What Makes a Good Project

A good project isn’t about fancy equipment. It’s about asking a clear question (or solving a real problem), testing it fairly, collecting enough data to trust what you found, and writing every step down. This page shows you how, with real projects that won big awards.

Science or engineering?Fair testsProject notebookData tablesAward winnersHow judges score102 project ideas

Two kinds of projects

Every project at the fair is one of two kinds. The kind you pick decides which steps you follow and which rubric the judges use. Here’s the short version: science starts with a question. Engineering starts with a problem.

Science experiment

“I wonder what happens if…”

  • You ask a question about how the world works.
  • You make a prediction (a hypothesis).
  • You run a fair test and measure what happens.
  • You end with an answer, even if it’s “my guess was wrong.”

Engineering project

“How can I build something that…”

  • You find a real problem or need.
  • You design and build a solution.
  • You test it, find what’s wrong, and make it better.
  • You end with a working design and data that proves it works.

Science Buddies puts it simply: “scientists study how nature works, engineers create new things.” The national science standards say the same thing: “While science begins with questions, engineering begins with defining a problem to solve.” Writing a computer program counts as engineering too. The fair has its own programming rubric for it.

Quick quiz: which kind is it?

Does road salt make lake ice melt faster than sand?

Science It’s a question about how nature works. You’d change the material and measure the melting.

Build a sled that carries more firewood without tipping over.

Engineering There’s a problem (tipping) and you’re building a fix. You’d test it, change it and test again.

Write an app that warns your family when the freezer door is left open.

Programming You’re building a solution out of code. Judges use the programming rubric.

Do spruce seedlings grow taller under red light or blue light?

Science One thing changes (light color) and you measure one result (height).

The scientific method, and why each step matters

The scientific method isn’t a list of boxes to fill in. Each step protects you from fooling yourself. Here’s what each one does for you.

  1. Ask a questionMake it something you can test and measure.Why“Is music good for plants?” is too fuzzy. “Do bean plants grow taller with 2 hours of music a day?” can be tested.
  2. Do background researchFind out what’s already known.Whyjudges want to see other ideas you considered, and why you chose yours.
  3. Make a hypothesisWrite your prediction before you test.Whyif you decide after, it’s easy to “see” what you hoped to see.
  4. Design a fair testChange one thing. Keep everything else the same.Whyif you change two things, you can’t tell which one caused the result.
  5. Collect and record dataMeasure carefully. Do at least 3 trials of each change. Write each one in your notebook right away.Whyone trial can be luck. Repeats show what really happens, but only if you wrote them all down.
  6. Analyze and concludeUse graphs and averages. Say what the data shows.Whya conclusion has to come from the numbers, not from your hunch.

Then share it, and let the results lead to your next question.

A wrong guess can still win. In 2011, 14-year-old Daniel Feeny took first place at the national Broadcom MASTERS for studying how wave force affects sea life in tide pools. His hypothesis wasn’t supported. “I discovered that nature doesn’t work that way, it’s not that simple,” he said. His careful test of an idea that turned out wrong was still real science. Source: Science News Explores

Fair tests: controlling your variables

A variable is anything that can change in your experiment. A fair test sorts them into three groups. Here they are, using a simple test: Which brand of paper towel soaks up the most water?

Independent variable

The one thing you change

The brand of paper towel.

Dependent variable

The thing you measure

How many milliliters of water each towel holds.

Controlled variables

Everything you keep the same

Towel size, water temperature, soak time, how long it drips, the same measuring cup.

Why does this matter? Say Brand A soaked for 30 seconds and Brand B soaked for 10. If Brand A wins, you can’t tell if it’s the better towel or just had more time. Your result is useless. Science Buddies puts it this way: “it is usually wise to have only one independent variable at a time.”

Write your controlled variables down, too. List them in your notebook before you start. If one changes by accident (the room gets warmer, or you run out of one brand of cup), write that down next to the trial. Then you’ll know which results to trust.

You also need a control group when you can. That’s a group that gets no treatment at all, so you have something to compare against. If you’re testing whether a fertilizer helps, the control is plants that get plain water.

The first fair test at sea: 12 sick sailors, 1747

Ship’s doctor James Lind wanted to find a cure for scurvy, a disease that killed many sailors. He picked 12 sick sailors who were “as similar as I could have them.” They slept in the same place and ate the same food. Then he split them into six pairs and gave each pair a different treatment.

  • Cider
  • Elixir of vitriol (acid)
  • Vinegar
  • Sea water
  • Oranges and lemonsAlmost well in a month
  • A purgative paste

The two men who got citrus were almost well within a month. Because Lind kept everything else the same, he could point to the one thing that made the difference. That’s the power of controlling variables. He also wrote down how each pair did, and in 1753 he published his results so other doctors could learn from them. Source: Houlberg, Clinical Medicine (2019)

How much data is enough?

Lind’s test had a weakness: only two sailors per treatment. What if one of them was just about to get better anyway? That’s why sample size, the number of things you test, matters so much. The ASEF rubric asks for at least 3 trials for each change you make. Science Buddies says “more is better.”

Here’s why. Look at this made-up paper towel data:

TrialBrand A (mL)Brand B (mL)
Trial 14251 ← odd one out
Trial 24436
Trial 34138
Trial 44335
Trial 54537
Average4339.4

If you only did Trial 1, you’d say Brand B wins. With five trials you can see that Trial 1 was odd. Maybe that towel was folded or got extra water. Brand A is better. More trials help you spot mistakes and find what’s really true.

But you can only catch an odd trial if you wrote down every trial, not just the average. If you only saved “Brand B: 39.4 mL,” the mistake would be hidden forever.

1 trial Could be luck. Could be a mistake. You can’t tell.
3 trials The fair’s minimum. You can find an average and spot a big mistake.
10+ trials Now you can trust the pattern. Graphs look clear.
SurveysYou need many more people than 3. Science Buddies says you’ll “almost always need many more than three participants!”

Big data, big result. Three teens from Kinsale, Ireland won the 2014 Google Science Fair Grand Prize for treating barley and oat seeds with helpful soil bacteria. They tested about 13,000 seeds over three years and compared them with an untreated control group. The treated seeds sprouted at least 50% faster. With that many seeds, luck can’t explain the result. Source: Modern Farmer

The engineering design process

Engineers follow a loop instead of a straight line. You’ll build, test and rebuild many times. Each new version is called an iteration, and judges want to see them.

  1. Define the problemWho needs help, and with what?Whya clear problem tells you what “success” looks like.
  2. ResearchWhat solutions exist? What’s wrong with them?Whythe rubric asks you to explain existing solutions and their problems.
  3. Design and predictSketch it. Choose materials. Calculate how well it should work.Whya prediction lets you check if your build met the goal.
  4. Build a prototypeMake a first version. It doesn’t have to be pretty.Whyyou learn the most from something real you can test.
  5. Test and measureRun fair tests, several times, and record every result in your log.Whydata proves your design works. “It seems better” doesn’t.
  6. Improve and repeatFix what failed, then test again.Whyevery failure teaches you something. Write down what you changed in each version, and why.

Steps 4–6 repeat until your design meets the goal.

Engineers still control variables. When you compare version 1 to version 2, test them the same way: same load, same distance, same temperature. Then you know the design change made the difference.

Keep a project notebook

Recording your data clearly is one of the most important parts of science and engineering. A project notebook (also called a lab notebook or log book) is where you write down everything you do, as you do it. Real scientists and engineers keep one for every project.

Here’s why it matters. In three weeks you won’t remember which trial was the one where the cup tipped over. Your notebook will. It’s how you find mistakes, how you prove your results to judges, and how someone else could repeat your work. A good rule: if it isn’t written down, it didn’t happen.

What goes in each entry

Every time you run a trial or build a new version, write an entry with these six parts:

  • Date and timeSo you can put your work in order and spot patterns, like “every Monday result is low.”
  • Trial or version numberGive every trial its own number. Never reuse one, even for a redo.
  • What you changedYour independent variable. For engineering, what’s new in this version.
  • What stayed the sameYour controlled variables. Note anything that changed by accident.
  • Results, with unitsThe exact numbers you measured. “12 min,” not “fast.”
  • Notes and surprisesAnything odd, any mistakes, and what you’ll try next.

Here’s what a real entry might look like. This student is testing what melts ice fastest: nothing, sand or road salt.

A made-up example. Notice the mistake is crossed out with one line, not erased.

Notebook rules

  • Write it right away. Don’t trust your memory, even for an hour.
  • Date every entry and number every page.
  • Use pen. Cross out mistakes with one line so they can still be read.
  • Never erase or tear out pages. A mistake is part of the record.
  • Always write units: mL, cm, °C, grams, seconds.
  • Record the weird results too. They often teach you the most.
  • Add sketches and photos. Tape in printouts, graphs and pictures of each build.
  • Paper or digital both work. A bound notebook is best. If you use a doc, date it and don’t delete things.

How careful records got the Wright brothers off the ground

In 1901, Wilbur and Orville Wright tested a new glider at Kitty Hawk. It was a letdown. It made only about one-third of the lift their math said it should. Their math used numbers that other inventors had trusted for more than 100 years. Something was wrong, but what?

Back home in Dayton, Ohio, they built a small wind tunnel in their bicycle shop. They tried as many as 200 model wing shapes. For nearly 50 of them, they ran careful tests and wrote down every number. Then they graphed the results.

Their records showed the problem. A key number everyone had used since the 1700s was too high. The old value was 0.005. The Wrights’ data said about 0.0033, and modern engineers agree they were right. Using their own data, they built the 1902 glider, the first aircraft a pilot could fully control. In 1903 they made the first powered airplane flight.

The Wrights didn’t win because they had better tools. They won because they measured carefully, wrote it all down and trusted their own data. Source: Smithsonian National Air and Space Museum

Record your data in tables

Draw your table in your notebook before you start. Then fill in one row for each trial as you go. Here are two made-up examples you can copy.

Science: a trial log

This is the ice test from the notebook page, with 10 g of material on each cube. Cube mass should stay the same, so writing it down catches bad trials.

What melts an ice cube fastest?
#MaterialCube (g)Time (min)Notes
1None3048
2Sand3041
3Salt3018
… trials 4–8 …
9Salt2612Small cube. Redo!
10Salt3018Redo of #9

When testing is done, make a summary table for your board. Explain any trial you left out. Never just hide it.

Minutes to melt a 30 g ice cube
MaterialTrial 1Trial 2Trial 3Avg.
None48504748.3
Sand41434041.3
Salt181718*17.7
* Redo. Trial 9 was left out (cube too small).

Engineering: a version log

Give each version its own row. This firewood sled is tested the same way every time: same hill, same rope, adding wood until it tips.

Firewood (kg) carried before tipping
Ver.What I changedTrials (kg)Avg.
1First build20, 22, 2121
2Runners 10 cm wider30, 29, 3130
3Deck 8 cm lower38, 40, 3738.3
4Added side rails37, 39, 3838

Wider and lower both helped. Rails didn’t, and that’s still useful: your log proves you tested the idea instead of guessing.

Table tips: label every column with its units, use one row per trial, and keep a notes column. A spreadsheet is great for averages and graphs, but write the raw numbers in your notebook first.

Units matter. Just ask NASA. In 1999, NASA lost the Mars Climate Orbiter, a spacecraft that had flown for nine months to reach Mars. One team’s software used English units. Another team’s software expected metric units. Nobody caught the mix-up, so the spacecraft flew too close to Mars and burned up in its atmosphere. One units mix-up wiped out years of work. Write your units every time. Source: NASA

Learn from award winners

These students won top prizes at state, national and international fairs. Look at what they did well, and steal their habits, not their ideas.

Science experiments

Science

Bacteria that help crops grow

Ciara Judge, Émer Hickey and Sophie Healy-Thow · 2014 Google Science Fair Grand Prize

Treated seeds sprouted at least 50% faster than an untreated control group, and crop yield went up as much as 70%.

Why it worked: a clear control group and thousands of seeds.

Modern Farmer · Scientific American

Science

Better electronics for the body

Grace Sun, 16 · 2024 Regeneron ISEF, top award ($75,000)

She tested four salts to improve tiny transistors that could work inside the body. The best one raised signal strength 97% compared with an untreated device.

Why it worked: she changed one thing at a time and compared each to a control.

Science News Explores

Science

Saving bug-eating insects

Poon Trakultangmun, Teepakorn Keawumdee and Pannathorn Siri · 2023 Regeneron ISEF, Young Scientist Award

They tested temperature, light and leaf vibration to help green lacewing eggs survive. Nearly six times as many eggs survived as in the usual container.

Why it worked: the usual method was the control, so the improvement was easy to measure.

Science News Explores

Science

Testing an old rocket rule

Eleanor Sigrest, 13 · 2016 Broadcom MASTERS, top prize

Designers assumed the best rocket nozzle angle was 15°. She 3D-printed nozzles at different angles and blew air through them. The best was about 20°.

Why it worked: she questioned something “everyone knows” and tested it.

Science News Explores

Science

Alaska: tree swallow chicks

Calvin Samson, grade 10, Bethel Regional High School · 2019 Alaska Science & Engineering Fair

“The Effects of Habitat on Tree Swallow Chick Mass” took 1st place in Animal Science and NOAA’s Taking the Pulse of the Planet Award.

Why it worked: a local question about Alaska wildlife that can be measured.

The Delta Discovery

Engineering and programming projects

Engineering

A motor with no magnets

Robert Sansone, 17 · 2022 Regeneron ISEF, first prize ($75,000)

Electric car motors need rare magnets. His magnet-free motor worked on the 15th try. One version melted on his desk. He tested it against a standard design and got 39% more torque.

Why it worked: he kept iterating and compared his design with an existing one.

Smithsonian Magazine

Engineering

Recycling 3D printer waste

Benjamin Davis, 16 · 2025 Regeneron ISEF, Young Scientist Award ($75,000)

He built a desktop machine that turns plastic scraps back into printer filament. He rebuilt it more than 50 times and tested over 50 setting combinations.

Why it worked: many iterations, each one tested and recorded.

3DPrint.com · Society for Science

Engineering

See-through car pillars

Alaina Gassler, 14 · 2019 Broadcom MASTERS, top prize ($25,000)

A camera outside the car shows the hidden view on the pillar inside. Her first image was blurry, so she added reflective fabric and 3D-printed parts to sharpen it.

Why it worked: a real problem (her mom’s blind spot) and a clear version 1 → version 2.

The Drive

Engineering

A fire alarm that knows people

Shanya Gill, 12 · 2023 Thermo Fisher JIC, top award ($25,000)

After a fire near her home, she built a low-cost heat camera that texts an alert if something hot is left with no one nearby. It spotted people 98% of the time.

Why it worked: she measured how well it worked, not just whether it worked.

Society for Science

Programming

Finding new planets, fast

Kaitlyn Wang, 17 · 2023 Regeneron ISEF, top award ($75,000)

Her program, ExoScout, searched space telescope data 120 times faster than older methods, with 97% accuracy. It found three new planets.

Why it worked: she compared her program with the existing method using real numbers.

Science News Explores

Notice what these winners have in common. They all compared against something: a control group, an older design or an existing method. They all collected lots of data. And they all kept track of it. You can’t say “it worked on the 15th try” or “I tested over 50 settings” unless you wrote down every try.

How judges score your project

ASEF judges for grades 6–12 use a 100-point rubric. There’s one for experiments, one for engineering and one for programming. All three are split into the same three parts.

  • 40 pts The process: method, design, data and analysis
  • 30 pts Presentation: board, talk and questions
  • 30 pts Innovation: new ideas, creativity and impact

The biggest part is your process. For an experiment, judges check that you labeled your variables and controls, did at least 3 trials for each change, and based your conclusion on the data. They also want to hear what could have gone wrong. For engineering, they look for calculations before you build, testing, iterations and a comparison with existing solutions.

Bring your notebook to the fair. Your board shows your best graphs. Your notebook is the proof behind them. Judges may ask to see it, and flipping to the page where a trial went wrong is a great way to answer a hard question.

Your board should explain the whole project even if you’re not there. In your talk, go beyond the board and explain why your work matters.

Climb the originality ladder

Judges give more points for a more original project. This is from the experiment rubric:

  1. ~3 ptsGoogled “science fair project ideas” and added your own twist.
  2. ~6 ptsFound existing research and tried to repeat the results.
  3. ~9 ptsImproved on existing research with a new approach or new data.

You don’t need a lab to reach the top step. Look at what scientists have already found about something in your town, then test it with your own local data. Read the full Grades 6–12 rubrics (PDF).

102 project ideas to get you started

Here are 102 ideas across every fair category, and more than half are about Alaska. Use one as a starting point, then make it yours. Change the place, the material or the question. That’s how you climb the originality ladder.

Science Engineering Programming Alaska Category number for registration
Animals and plants15 ideas · 8 Alaska

Categories: 0100 Animal Sciences · 1500 Plant Sciences

ProgrammingGrades 9–120100 Animal SciencesAlaska

Shifts in Alaska Salmon Run Timing

A salmon run is the yearly migration of adult salmon from the ocean up a river to spawn. Run timing can change as water temperatures change. A University of Alaska study at Auke Creek near Juneau found that pink salmon there have been migrating earlier over time as the water has warmed.

Question: Over the last 20 years, has the midpoint date of Alaska salmon runs (the day half of the season's fish have passed) moved earlier or later?

Why it stands out: It separates when salmon arrive from how many arrive. It builds a timing map that fishers and managers can use.

How to do it

Download daily fish counts from ADF&G Fish Counts for 3–5 rivers. Each river needs at least 10 complete seasons. For each river and year, find the run midpoint date: the day the running total reaches half of the season's count. Fit a trend line (linear regression) for each river, using each year (not each day) as one data point. Report how many days per decade the run moved, with a confidence interval (a range that shows how sure you are). Skip seasons with missing weeks or a change in counting method.

Learn more: Southeast Alaska pink salmon adapting to climate change, warming water (Anchorage Daily News)
Builds on: adfg.alaska.gov

ScienceGrades 6–80100 Animal SciencesAlaska

Snow Fleas: Springtail Activity and Snow Temperature

'Snow fleas' are not fleas. They are springtails, tiny six-legged animals that live on the forest floor and escape danger by flipping into the air with a forked, spring-loaded tail. Most people only notice them in spring, when they look like pepper sprinkled on melting snow, and scientists are still not sure exactly how they reach the snow surface.

Question: At what snow-surface temperatures are springtails (snow fleas) most common, and is the pattern different on sunny and shaded snow?

Why it stands out: Most people walk past the black specks on spring snow. You will map when and where they show up in your own neighborhood.

How to do it

Mark 10 fixed plots (25 × 25 cm squares) on late-winter snow. Put 5 in the sun and 5 in spruce shade. Count springtails in each plot at the same time every day for 4–6 weeks. Log air and snow-surface temperature with an inexpensive probe thermometer. Note the snow type each day (crust, slush, or fresh). Plot daily sun and shade averages against temperature, and find the threshold temperature where counts rise sharply.

Before you start: Adult supervision. Make a winter field safety plan. Choose sites near home, away from roads and thin ice. Keep away from moose.

Learn more: Springtails: a Sign of the Season (Alaska Science Forum, UAF Geophysical Institute)
Builds on: akentsoc.org

ScienceGrades 6–80100 Animal Sciences

Ant Trail Width and Head-On Collisions

Many ants travel along trails between their nest and food sources. Traffic scientists study ant trails because ants rarely form traffic jams: as trails get more crowded, ants spread out across the trail and adjust their speed. This project tests whether narrow sections of a trail, like a lane closure on a highway, cause more collisions.

Question: Do ants collide head-on more often on narrow sections of their own trails than on wide sections?

Why it stands out: It brings highway traffic science to an ant trail. Nothing in the ants' world gets changed.

How to do it

Film natural ant trails from a fixed spot. Don't use bait, and don't touch or block the trails. On 10 colonies, pick a narrow and a wide section of each trail. Film 10 minutes of each section on 3 different days. Count head-on collisions per 100 ants passing. Also count ants per minute (traffic density), so you can separate crowding from trail width. Use each colony as one data point (narrow-section rate minus wide-section rate).

Learn more: How Ants Go Marching and Never Cause Traffic Jams (Inside Science, American Institute of Physics)

ScienceGrades 6–121500 Plant SciencesAlaska

Does Smoke Water Speed Fireweed Seed Germination?

Fireweed is a tall plant with pink flowers that is common across Alaska and is one of the first plants to grow back after wildfires. Seeds of some plants sprout better after contact with chemicals from smoke, and liquid 'smoke water' can be bought to test this. Germination is when a seed starts to sprout.

Question: Does store-bought smoke water change how quickly fireweed seeds germinate?

Why it stands out: An Alaska plant that returns quickly after fires may respond to smoke very differently than a garden seed.

How to do it

Use one documented seed lot of fireweed and treat every seed the same way. Follow any pretreatment the seed supplier recommends. Compare three dilutions of purchased smoke water, used safely according to the label, with plain water as a control. Use four Petri dishes of 25 seeds per treatment. Run 'Grand Rapids' lettuce seeds the same way, since this variety is often used to test smoke-water response. This positive control checks that your smoke water works. Place dishes in random spots with equal light and temperature. Count germinated seeds daily for 21 days. Compare percent germination and average days to germination for each species.

Before you start: Adult supervision: purchased smoke-water handling, label directions, and sealed disposal of any moldy dish; no making fire or smoke

Learn more: Fireweed (Chamerion angustifolium) species review (USDA Forest Service Fire Effects Information System)

ProgrammingGrades 9–120100 Animal SciencesAlaska

Cold Pool Size and the Snow Crab Collapse

Snow crab in the eastern Bering Sea live on the sea floor in water just above freezing. The 'cold pool' is a band of ultra-chilled bottom water that forms after winters with sea ice. A marine heat wave kept sea ice from forming for two winters (2018–2019), billions of snow crab disappeared, and in 2022 Alaska canceled the snow crab harvest for the first time.

Question: Could the size of the Bering Sea cold pool (bottom water colder than 2 °C) have warned scientists before snow crab numbers crashed?

Why it stands out: You test a warning sign for a crash that wiped out about 90% of Bering Sea snow crab. You use the same data NOAA scientists use.

How to do it

Download NOAA's yearly survey counts of young and adult snow crab in the eastern Bering Sea. Also download cold-pool area, which is the area of bottom water colder than 2 °C. Using only years before 2018, compare crab numbers with cold-pool size 1, 2, or 3 years earlier (a time lag). Choose a threshold rule that flags years of low crab numbers. Compare it with a baseline forecast that just repeats last year's count. Lock in those choices before you test them on the years you have from 2018 through 2021, when the collapse was measured. Report prediction errors and false alarms, and list any missing survey years. Treat same-year matches as a separate comparison, not as an early warning.

Learn more: Alaska crab fishery collapse seen as warning about Bering Sea transformation (Alaska Beacon)
Builds on: alaskasnewssource.com

ProgrammingGrades 9–121500 Plant SciencesAlaska

Does First Frost Predict Birch Leaf Color Change?

In fall, leaves change from green to yellow or brown and then drop; the GLOBE program calls this green-down. Its timing can shift earlier or later from year to year depending on the weather. Students in GLOBE record leaf color on their own trees, creating records scientists can study.

Question: Does the date of the first fall frost help predict when birch leaves change color in Interior Alaska?

Why it stands out: Student leaf records test whether a familiar frost explanation beats the calendar.

How to do it

First make sure you can get raw GLOBE green-down (leaf color) records. Then choose at least three sites with 10 or more usable years and matching weather data. Define the color-change date the same way at each site. Compare a model using August temperature and first frost date with each site's average calendar date. Fit on earlier years, then test on whole later years. Also test one site left out of fitting. Report date errors, gaps in observations, and insect damage where recorded. A frost that comes after the color change cannot be an advance warning. Day length alone cannot explain year-to-year changes at one site, because it is the same every year on a given date.

Learn more: A Call for Daily Observations of Fall Color (NASA GLOBE Observer)
Builds on: akentsoc.org

ScienceGrades 6–121500 Plant SciencesAlaska

Does Splitting Long Days Affect Radish Growth?

In summer, much of Alaska gets 18 or more hours of daylight. A photoperiod is the number of hours of light a plant gets each day, and growers sometimes break up the dark period with night-interruption lighting because plants respond to the pattern of light and dark, not just the total. Radishes grow fast, so they make a quick test crop.

Question: Does splitting 18 hours of daily light into two blocks change radish growth when the total light stays the same?

Why it stands out: Both groups of plants get the same total light, but one group gets a break in the middle.

How to do it

Compare two repeating 24-hour schedules. One is 18 hours of light, then 6 hours dark. The other is 9 hours light, 3 dark, 9 light, 3 dark. Keep light intensity, temperature, soil, water, and radish variety the same. Use 10 separate pots per schedule in each of three four-week runs. Swap the schedules between grow compartments across runs, and rotate pots within each compartment. Compare leaf area and final root and shoot mass. Treat each run as a paired comparison. This models how growers might manage Alaska-like long summer light.

Before you start: Adult supervision: grow-light electrical safety and water containment

Learn more: Photoperiodic Lighting (Ohio Controlled Environment Agriculture Center, Ohio State University)

ScienceGrades 6–121500 Plant SciencesAlaska

Lettuce Growth per Kilowatt-Hour Under Different Photoperiods

A photoperiod is the number of hours of light a plant gets each day. Under grow lights, longer days usually mean more growth but also more electricity. Alaska winter days are short, so indoor growers rely on LED lights; one study of indoor lettuce found that growth per kilowatt-hour was highest at a middle photoperiod, not the longest one.

Question: Which daily light schedule (photoperiod) grows the most lettuce per kilowatt-hour of lamp electricity?

Why it stands out: The longest growing day has to earn its extra electricity.

How to do it

Grow one lettuce variety under 12, 16, 20, and 24 hours of LED light per day at the same light intensity. Use three separate grow compartments per schedule, with four pots each. Place compartments and pots in random spots. Keep temperature and watering steady, and meter the electricity for each compartment. After five weeks, measure fresh leaf mass, leaf damage (such as tipburn), and grams per kilowatt-hour. Treat each compartment, not each pot, as one replicate. Report only lamp energy, not total greenhouse cost. This models indoor winter growing in Alaska.

Before you start: Adult supervision: grow-light electrical safety and water kept away from plugs

Learn more: Optimal daily light integral for indoor hydroponic lettuce (Scientific Reports, via NIH PubMed Central)

EngineeringGrades 6–121500 Plant Sciences

A Homemade Clinostat With Proper Controls

Gravitropism is the way roots grow down and shoots grow up in response to gravity. A clinostat is a slowly turning device that keeps changing the direction of gravity for a plant, and scientists use it to imitate some effects of spaceflight. It does not remove gravity, so it is only a partial model of weightlessness.

Question: Does a homemade slow clinostat change root bending more than a control that spins but keeps gravity pointing the same way?

Why it stands out: A vertical-axis spinning control tests whether roots react to gravity changing direction or just to being spun.

How to do it

Build a guarded single-axis clinostat. Compare three groups at two slow speeds, keeping seedlings near the center of rotation. One group spins on a horizontal axis, so gravity's direction keeps changing. A control group spins on a vertical axis, so gravity's direction stays the same. A third group does not spin. With a teacher, surface-sterilize radish seeds in dilute bleach. Grow them on damp paper in closed clear containers, six per setting, and photograph roots after 48 hours. Measure root bending angle, improve the mount, and retest. Spinning changes gravity's direction for the plant but does not remove it. Throw away any container with visible mold without opening it.

Before you start: Adult supervision: dilute bleach, guarded motors, closed seed containers, and sealed disposal of anything contaminated; no growing microbes

Learn more: Magnets mimic gravity's guiding hand for plants growing in space (Science News Explores)
Builds on: societyforscience.org

ProgrammingGrades 9–121500 Plant Sciences

Detecting Plant Water Stress from Leaf Angles

When plants run short of water, their leaves can roll, droop, or change angle before obvious wilting. Leaf angle changes are a useful sign of how plants respond to drought. Scientists now use time-lapse photos and image-analysis software to measure leaf angles automatically and spot water stress early.

Question: Does a leaf-angle warning detect mild water stress sooner than a leaf-color warning?

Why it stands out: Your program has to spot a thirsty plant before it droops, without flagging healthy plants.

How to do it

Grow 24 separate bean pots. Randomly assign half to normal watering and half to mild drought. Weigh pots daily, and photograph them at the same time and lighting. Compare the leaf-angle and leaf-color warnings with a water-loss threshold set ahead of time. Stop drying at 20 percent loss of the water held after draining (pot capacity), or at the first visible wilt, whichever comes first. Then rewater. Lock the warning thresholds and test them on a second fresh batch of 24 pots. Record how early each warning comes and how many false alarms it gives for well-watered plants.

Before you start: Adult supervision: grow lights, water near electronics, and plant disposal

Learn more: All the angles: Automated image processing could aid crop evals (University of Nebraska–Lincoln)
Builds on: societyforscience.org

ScienceGrades 6–121500 Plant Sciences

Do Pea Roots Grow Toward Water Sounds?

In a 2017 study, scientist Monica Gagliano's team reported that pea roots grew toward the sound of running water. The plants seemed to avoid a recording of running water, and they ignored the sound when the soil was already moist. How plants might sense sound is unknown, so these results need more testing.

Question: Do pea roots grow toward recorded water sounds more often than toward a different sound of the same loudness?

Why it stands out: A surprising water-sound claim has to beat both silence and another sound.

How to do it

Grow peas in simple Y-shaped root choice chambers. Use 15 separate chambers each for a water sound, a non-water sound, and a silent speaker as a control. Pick the sound side at random. Match sound level, soil moisture, temperature, and measured chamber vibration, using speakers that are isolated from the chamber. Keep real flowing water outside the setup. After 10 days, score which arm the root entered first, using photos with coded labels so you don't know the treatment (blind scoring). Repeat with 15 new chambers per group. Any choice shows a response to this sound-and-vibration setup, not proof that roots can hear.

Before you start: Adult supervision: chamber construction, low-voltage speakers, and water kept away from electronics

Learn more: Hydrophonics (Northern Woodlands magazine)

ScienceGrades 6–120100 Animal Sciences

Tardigrade Freeze-Thaw Survival in Different Moss Habitats

Tardigrades, or water bears, are microscopic animals, most smaller than a pencil dot, that live in the thin film of water on moss. When their moss dries out or freezes, they protect their cells and go into a deep, nearly lifeless sleep, then wake up when conditions improve. They die if they stay dry or frozen too long, so habitats that freeze and dry often may favor tougher tardigrades.

Question: Do tardigrades (water bears) from roof moss survive repeated freezing and thawing better than ones from tree-bark or rock moss?

Why it stands out: Water bears are famous for surviving almost anything. You test whether harsher homes make tougher water bears.

How to do it

Collect 10 small moss samples from each habitat: roof edges you can reach from the ground, tree bark, and rocks. Do a practice search first with a microscope at 40–100× magnification. Soak each sample for a day, then remove and count the live tardigrades. Afterward, dry and weigh the moss that is left. Take animals from at least 5 samples per habitat and split them into groups of 20. Give the groups 0, 3, or 6 freeze-thaw cycles, keeping water amount, freeze time, and thaw time the same. Count the percentage that move again within 24 hours (survival rate). Compare survival by habitat, and count each original moss sample as one replicate.

Before you start: Adult supervision. Make a collection safety plan. Use only moss you can reach from the ground. Never climb onto roofs or use ladders.

Learn more: All About Tardigrades (Ask A Biologist, Arizona State University)
Builds on: akentsoc.org

ScienceGrades 6–121500 Plant Sciences

Which Conifer Needles Capture the Most Fog?

Fog is made of tiny water droplets floating in the air. When fog drifts through trees, droplets hit needles and leaves, join together, and drip to the ground; this is called fog drip. In some foggy forests, fog drip can be as heavy as light rain.

Question: Which conifer branch shape catches the most fog droplets for its size?

Why it stands out: The winning needles have to stay ahead after you account for branch size.

How to do it

Get five fresh branches, from separate trees, of each of three local conifers, with permission. Add a plastic-mesh control and an empty-holder control. Put each in the same fan-blown mist from a humidifier for one hour, in random order. Keep branches warmer than the dew point (the temperature at which water vapor condenses into liquid), so you measure droplets caught, not condensation. Weigh the water caught. Compare water per gram of branch and per front-facing area measured from a photo, and record needle spacing. Repeat on a second day with fresh branches. Call this a test of fog-droplet capture. Do not treat it as proof of water condensing from invisible vapor.

Before you start: Adult supervision: permitted branch clipping, clean humidifier use, contained mist, and electronics kept dry

Learn more: Fog drip (Glossary of Meteorology, American Meteorological Society)
Builds on: snexplores.org

ScienceGrades 6–80100 Animal Sciences

Worm Grunting: Which Vibrations Bring Earthworms Up?

Worm grunting is a way to collect fishing bait: people rub a piece of iron across a wooden stake driven into the ground, and the vibrations make earthworms crawl out. A Vanderbilt University scientist studied why it works. The leading idea is that the vibrations sound like a burrowing mole, a worm predator, so the worms flee to the surface.

Question: Which vibration brings the most earthworms to the surface: a steady tone at one frequency, or a tapping pattern like rain?

Why it stands out: It tests the science behind 'worm grunting,' a real way people collect bait. It pits two ideas against each other: worms fleeing a digging mole, or worms coming up for rain.

How to do it

Use identical bins of equally moist soil with 20 worms each. Bury the same vibration source in each bin. Choose 4 frequencies (pitches) that a vibration sensor in the soil can measure well. Each frequency must be below one quarter of the sensor's sampling rate (how many readings it takes per second). Compare them with a tapping pattern and a silent control, 5 trials each. Measure the actual frequency and vibration strength (amplitude) in the soil, and keep the amplitude equal. Count the worms that come up in 10 minutes. Run trials in random order, and give reused worms a day of rest. Keep store-bought worms contained; don't release them outdoors.

Learn more: Grunting for worms (Science News Explores)

ScienceGrades 6–121500 Plant SciencesAlaska

Wood Strength of Beetle-Killed Spruce Over Time

Spruce beetles are native bark beetles that have killed large areas of spruce in Southcentral Alaska. Dead trees are often salvaged for lumber or firewood, but wood can weaken over time as it dries, cracks, and decays. Testing the wood shows whether long-dead trees are still useful.

Question: Does the time since a spruce was killed by beetles relate to its wood's bending strength and dent resistance?

Why it stands out: Local salvaged wood gets measured tree by tree instead of judged by how it looks.

How to do it

Partner with a landowner or sawmill. Get one spruce species from five separate trees in each of three groups with a documented time since death. Add five recently cut healthy trees. Use only wood that was already cut down and has reliable records. Cut equal test sticks along the grain, and match their measured moisture content. Use a guarded test rig to measure the load that breaks each stick (bending strength) and the dent depth from a ball dropped from a fixed height. Average several sticks within each tree. Compare trees, rather than counting each stick as its own sample. Call the dent test impact resistance, not standard wood hardness or a building safety rating.

Before you start: Adult supervision: documented wood that was already cut down, saw cutting, guarded breaking loads, and land access; no cutting down trees or going near standing dead trees

Learn more: Spruce Beetle Outbreaks and Climate Change in Alaska (USDA Climate Hubs)
Builds on: knom.org

Earth, ice and volcanoes9 ideas · 7 Alaska

Categories: 0700 Earth & Planetary

EngineeringGrades 6–80700 Earth & PlanetaryAlaska

Preventing Soil Liquefaction Under Model Buildings

Soil liquefaction happens when strong earthquake shaking makes loose, water-soaked sand or silt lose strength and act like a liquid. Liquefied ground can no longer hold up weight, so buildings on it can sink or tilt. It needs three things together: loose sandy soil, groundwater filling the spaces between grains, and strong shaking.

Question: Shaking can make wet sand lose its strength and act like a liquid; this is called liquefaction. Which fix best keeps a model building from sinking: compacting the sand, gravel drains, or a geotextile (fabric) layer?

Why it stands out: You take a 2026 Nome ASEF winner's liquefaction experiment from the problem to possible fixes. Alaska's 1964 earthquake is the hook.

How to do it

Build a shake table on a guided slide with a fixed stop. Tape a phone to it running a free accelerometer app (which measures shaking), so every trial shakes the same. Fill identical tubs with water-saturated sand. Test 4 treatments: none, compacted (packed down), gravel drain columns, and a landscape fabric layer. Run 5 trials each. Measure how far a weighted block sinks and tilts. Reset the sand the same way every time: stir it, let it settle for a set time, and check the depth.

Learn more: Liquefaction of Soil During Earthquakes, Geology.com
Builds on: knom.org

ScienceGrades 9–120700 Earth & PlanetaryAlaska

Depth Hoar: What Grows Weak Snow Layers?

Depth hoar is a layer of large, loosely bonded snow crystals at the bottom of the snowpack. It forms when the ground stays relatively warm under a shallow, cold snowpack, creating a strong temperature gradient (a big temperature difference over a short distance). Because the crystals bond poorly, depth hoar is a weak layer involved in many dangerous avalanches.

Question: Depth hoar is the layer of large, sugary snow crystals at the bottom of the snowpack. Are these crystals larger in open fields or under spruce trees? How does their size relate to the temperature gradient (the temperature difference from the bottom to the top of the snow)?

Why it stands out: You study the weak snow layer behind many Alaska avalanches with a hand lens and a winter of patience.

How to do it

Pick 3 open sites and 3 forest sites near home. At each, put inexpensive temperature loggers at the ground and near the snow surface, recording every hour all winter. Each week, dig a small snow pit and measure the snow depth. Photograph crystals from the bottom layer on a gridded card through a hand lens. For each site, compare average crystal size with the average temperature gradient. (That is the temperature difference divided by snow depth.) Count each site, not each crystal, as one data point. Compare your depths with Community Snow Observations data.

Before you start: Adult supervision: make a winter field safety plan; use low, flat sites away from avalanche terrain, thin ice, and roads, and keep away from moose

Learn more: Depth Hoar / Basal Facets, Avalanche Encyclopedia, Avalanche.org
Builds on: communitysnowobs.org

ScienceGrades 6–120700 Earth & PlanetaryAlaska

Does Patchy Ground Cover Speed Permafrost Thaw?

Permafrost is ground that stays frozen for at least two years in a row. Above it is the active layer, which thaws each summer and refreezes each winter. In Alaska, plants and moss insulate permafrost from summer heat, so when cover is removed by fires, trails, or roads, the ground warms and thaws deeper.

Question: Picture frozen ground with the same total amount of insulation, but with bare gaps. Does it thaw deeper than ground under one continuous layer?

Why it stands out: You test whether a few bare gaps, like trails or burn scars on tundra, matter more than the average cover. The model fits in your hands.

How to do it

Freeze identical tubs of damp sand. Cover them with the same mass of insulation, either craft foam or dried moss. Lay it as one continuous layer, as clustered patches, or as stripes. Add a bare tub as a control (comparison), with 6 tubs for each setup. Let them thaw in the same room. Measure thaw depth with a thin probe at fixed times. Put temperature probes under the middle and edges of patches. This shows whether heat spreads sideways from the gaps.

Learn more: Permafrost Landscapes, Denali National Park, National Park Service

ScienceGrades 6–120700 Earth & PlanetaryAlaska

Does Salt Make Glacial Flour Settle Faster?

Glacial flour (also called rock flour) is very fine rock powder made when glaciers grind over bedrock. It is so fine that it stays suspended in water, which makes glacier-fed rivers and lakes look milky. This project tests whether salt, like the salt a river meets at the ocean, makes the particles clump together and settle faster.

Question: Glacial flour is the fine rock powder that makes glacier rivers look milky. Does salt make it settle faster, and how much salt does it take?

Why it stands out: You explain the sharp line where milky glacier water meets the ocean, using particle chemistry you can see in a tube.

How to do it

Sieve glacial silt, or buy garden rock dust. Stir equal amounts into tall clear tubes of water at 6 salt levels, from 0 to 35 grams per liter. (35 g/L is about as salty as the ocean.) Make 3 tubes for each level. Add tubes of salt water with no silt as blanks (for comparison). Shine a light through each tube. Record brightness with a light sensor or phone light meter every 10 minutes for 24 hours. Find the 'half-clear time' (when brightness is halfway back to the blank) for each salt level. Then find the salt level where it changes most.

Before you start: Adult supervision: if you collect silt near a glacial river, follow the ISEF Field Work Safety Plan

Learn more: Glacial Till and Glacial Flour, National Park Service

ProgrammingGrades 9–120700 Earth & PlanetaryAlaska

Does the Magnitude Cutoff Change Aftershock Decay?

Aftershocks are smaller earthquakes that follow a large earthquake, and their rate decreases over time. Omori's law is a formula that describes how fast that rate falls. Right after a big quake, some small aftershocks can be missed, so this project tests whether the smallest size you choose to count changes the result.

Question: Does the minimum magnitude you choose to count (the magnitude cutoff) change how fast Alaska aftershocks seem to decrease?

Why it stands out: You check whether a famous earthquake law changes depending on which quakes you choose to count.

How to do it

Pick 3 large, well-recorded Alaska earthquakes from the Alaska Earthquake Center catalog, such as the 2018 Anchorage M7.1. Count aftershocks in the same time bins at 3 magnitude cutoffs. Fit Omori's law (a formula for how aftershock rates drop over time) to each. Small quakes are often missed right after a big one. So before comparing, check when the catalog becomes complete (records nearly every quake) at each cutoff. Use bootstrapping (resampling your data many times) to put error bars on each decay rate.

Learn more: Aftershock Forecast Overview, U.S. Geological Survey
Builds on: earthquake.alaska.edu

ScienceGrades 6–120700 Earth & Planetary

Where Should Straw Mulch Go After Wildfire?

After a wildfire burns off plants and leaf litter, bare soil on hillsides washes away easily in rain; this is soil erosion. Crews often spread straw mulch to cover the ground, and research by the U.S. Forest Service found that the percentage of ground covered is the main control on erosion after fire. Testing where to place a limited amount of mulch can help protect soil and streams.

Question: On a bare hillside after a wildfire, does where you place straw mulch matter more than how much you use?

Why it stands out: You build on a Broadcom MASTERS project about erosion after wildfires. You turn 'where should the straw go?' into a tested rule.

How to do it

Fill identical trays with the same soil at the same slope. Cover 40% of each tray with straw in 3 layouts: strips across the slope (along the contour), strips running downhill, or scattered patches. Add bare trays and fully covered trays for comparison, with 6 trays of each. Pour the same simulated rain from a bottle with punched holes at a fixed height. Catch the runoff. Then dry and weigh the soil that washed off.

Learn more: Post-Fire Soil Erosion and How to Manage It, Joint Fire Science Program brief
Builds on: snexplores.org

EngineeringGrades 6–120700 Earth & PlanetaryAlaska

How Sensitive Is a Homemade Seismometer?

A seismometer is an instrument that detects ground shaking from earthquakes. Inside, a weight hangs on a spring and tends to stay still while the ground and the instrument's frame move, and electronics measure the difference. Low-cost seismometers built from small computers can detect earthquakes too, but they cannot sense the smallest or farthest ones.

Question: How large and how close does an Alaska earthquake need to be for a homemade seismometer (a shaking detector) to detect it?

Why it stands out: You build on a past Alaska fair seismometer project. You end up with a real curve showing the smallest earthquakes your town's sensor can catch.

How to do it

Build a seismometer from an Arduino or Raspberry Pi and a sensor. The sensor can be a MEMS accelerometer chip or a geophone (a magnet-and-coil sensor). First, record a quiet week to find your noise floor, the smallest shaking it can sense. Compare 2 sensor designs or mounts. Then record nonstop for 2 or more months on a concrete floor. Match your detections with the Alaska Earthquake Center catalog. Plot detection rate against magnitude and distance.

Learn more: Seismometers, seismographs, seismograms: how do they work? U.S. Geological Survey
Builds on: adn.com

ProgrammingGrades 9–120700 Earth & PlanetaryAlaska

Sampling Bias in Volunteer Snow Depth Data

Community Snow Observations is a citizen-science project in which backcountry skiers and other volunteers measure snow depth and share the data with scientists. Volunteers can only measure where they travel, so some places are measured much more than others; this is called sampling bias. Sampling bias can make an area's average snow depth come out wrong.

Question: If snow depths are measured mostly near trails and roads, how far off does the area's average snow depth become?

Why it stands out: You show which places a citizen-science snow map leaves out, and how much that matters.

How to do it

First, check Community Snow Observations data for 3 areas and dates with many measurements. One example is Thompson Pass, where the project began. Use all points in each set as your best available average. Have a computer draw 1,000 samples of the same size in three ways: random, clustered near trails or roads, and balanced across elevations (stratified). Compare how far each sample's average lands from the full average. Also compare it with the project's modeled snow map. Keep in mind that the full set is also volunteer data. So it isn't the true average for the landscape.

Learn more: About Us, Community Snow Observations

ScienceGrades 9–120700 Earth & Planetary

Does Silicone Putty Flow Like Glacier Ice?

Glaciers move partly by sliding on their beds and partly by slowly deforming inside, because ice under pressure behaves like a very thick, plastic material. The push that drives this flow, called driving stress, depends on the ice's thickness and the slope of its surface. This project uses silicone putty as a model to compare with the power-law rule scientists use for ice flow.

Question: Does a silicone-putty model glacier flow faster on steeper slopes the same way real glacier ice does?

Why it stands out: You get a hands-on test of the physics inside glacier models. You also show where a putty model works and where it breaks.

How to do it

Spread putty on the same rough ramp at 5 slopes and 2 temperatures, with 3 fresh trials each. Keep the putty at its test temperature during each run. Track markers on the surface with time-lapse photos. Measure thickness at the tracked section. Check that the putty touching the ramp does not slide. For each temperature, graph speed divided by thickness against driving stress. Driving stress is gravity's push, based on slope and thickness. Use log scales, which turn power laws into straight lines. Estimate the line's slope as an 'apparent flow exponent.' Only do this where flow is steady and there is no sliding. Compare it with ice's value of about 3. Explain that spreading putty may not meet these conditions.

Learn more: Deformation and sliding, AntarcticGlaciers.org

Environment and climate7 ideas · 6 Alaska

Categories: 1000 Environmental

ProgrammingGrades 9–121000 EnvironmentalAlaska

Can Winter Weather Predict Spruce Beetle Damage?

Spruce beetles are small bark beetles, native to Alaska, that tunnel under the bark of spruce trees and can kill them. A large outbreak has killed spruce across Southcentral Alaska in recent years, and warmer conditions help the beetles grow faster and attack more trees. Each summer, the U.S. Forest Service maps new tree damage from small airplanes in an aerial detection survey.

Question: Does winter temperature help predict next year's spruce beetle damage on aerial survey maps better than last year's damage pattern alone?

Why it stands out: A weather-based warning has to beat the simple guess that new damage appears next to old damage.

How to do it

Use Alaska aerial detection survey maps and weather records from 2015–2025. Keep only forest areas that were surveyed in both years of each comparison. Start with a baseline model that uses only the earlier year's damage. Compare it with models that also add winter temperature, snow depth, and distance from earlier damage. Train your models on earlier years. Test them on later years and on map areas left out of training. Report the prediction errors and the number of usable region-years (one region in one year). Call your result "mapped damage," not total damage, because how much area was flown and delays in spotting dying trees can change what shows up on the maps.

Learn more: Spruce Beetle Outbreaks and Climate Change in Alaska (USDA Climate Hubs)
Builds on: fs.usda.gov

ScienceGrades 6–121000 EnvironmentalAlaska

Do Road Culverts Change Stream Insects?

A culvert is a pipe that carries a stream under a road. Poorly designed or poorly maintained culverts can block fish from moving up and down a stream, and Alaska has a program to find and fix them. Stream insects such as mayflies, stoneflies, and caddisflies (benthic macroinvertebrates) are often used to judge stream health, so they can show whether a culvert changes the stream.

Question: Do stream insect communities change more across a culvert than across nearby stretches of stream with no culvert?

Why it stands out: A reference stretch tests whether the culvert effect is just normal change along a stream.

How to do it

Pick three culverts that you can reach safely. Sample three shallow riffles above and below each one. Riffles are fast, shallow, rocky spots. Use the same net area and sampling time every time. On the same streams, sample matching upstream and downstream reference stretches with no culvert. Identify mayfly, stonefly, and caddisfly families. Record habitat, water temperature, and turbidity (how cloudy the water is). Compare the above-below change at each culvert with the change on its reference stretch. Count each crossing as one separate site. Don't claim the culvert caused any difference, since this shows a correlation, not a cause.

Before you start: Adult supervision: land access permission, road and water hazards, gloves, and a field safety plan; no deep wading or handling fish

Learn more: Fish Passage Improvement Program, Alaska Department of Fish and Game
Builds on: akentsoc.org

ScienceGrades 9–121000 EnvironmentalAlaska

Humidity Correction for Low-Cost Smoke Sensors

Low-cost air sensors, such as PurpleAir, measure fine particle pollution from wildfire smoke and cost far less than official air monitors. In Alaska, they can give smoke information to rural communities that have no official monitor. Their readings need to be checked against official monitors and corrected, and this project tests whether adding humidity to that correction makes them more accurate.

Question: Does correcting for humidity make low-cost particle sensors more accurate during Alaska wildfire smoke?

Why it stands out: The correction has to work on smoke events it has never seen.

How to do it

Use public hourly records from two or more low-cost sensors placed next to an official (regulatory) air monitor. Check that their times, locations, and units match. Pick at least six smoky weeks that have humidity records. Build a simple correction equation using the first four weeks. Then compare the errors of corrected and uncorrected readings on the remaining full weeks. Check the very smoky hours separately too. If you can't find paired records, get approval to place a sensor beside a monitor before you start. Never create smoke.

Learn more: UAF project seeks to provide air quality data for rural, remote Alaska areas, Alaska Public Media

EngineeringGrades 6–121000 EnvironmentalAlaska

Rainwater Catchment and First-Flush Diverters

Rainwater catchment means collecting rain from a roof for household use, which some Alaska homes rely on. The first rain off a roof washes away pollen and other buildup. A first-flush diverter sends that first, dirtiest water away so cleaner water reaches the storage tank, but every gallon diverted is water the household loses.

Question: Which first-flush diverter size makes stored roof rainwater the least cloudy while wasting the least water?

Why it stands out: Your design has to balance cleaner-looking water against the water a household loses.

How to do it

A first-flush diverter sends away the first, dirtiest water from a roof. Install three diverter sizes and a no-diverter control under equal roof areas. Do all sampling at ground level. Over 10 rainstorms, rotate the devices among the downspouts. Measure the water diverted, turbidity (cloudiness), conductivity (a measure of dissolved salts), and pH. Compare the stored water within each storm, and record the days since the last rain. Then improve the best diverter and test it in three later storms. These tests cannot show whether the water is safe to drink.

Before you start: Adult supervision: ground-level collection, gloves, care with unknown water, and a field safety plan; no climbing and no drinking

Learn more: Rainwater Catchment Systems, Your Northern Home (Cold Climate Housing Research Center)
Builds on: adn.com

ProgrammingGrades 9–121000 EnvironmentalAlaska

Are Alaska Harbors Warm Enough for Green Crabs?

The European green crab is an invasive species: a crab from another part of the world that can spread quickly and harm local shellfish and shoreline habitats. It was first found in Alaska in July 2022 at the Annette Islands Reserve in southern Southeast Alaska, and it is a banned species in the state. Its larvae drift on ocean currents, so scientists watch which nearby waters could support it.

Question: Which Alaska harbors have recorded water temperatures inside the published ranges where European green crabs can survive and reproduce?

Why it stands out: Your map separates measured water temperature from the much harder question of whether crabs will actually invade.

How to do it

Find water-temperature records for at least 10 Alaska coastal sites that cover the same months. Look up published temperature limits for green crab survival and for breeding. Count how many days each site falls inside each range. Count again after adding 1 °C and then 2 °C, and label these as 'what if' warming scenarios. Use the 2022 green crab discovery at the Annette Islands Reserve as background. Show how the site rankings change with missing data or with different published limits. Remember that warm enough water alone does not mean crabs will arrive and build a population.

Learn more: European Green Crab species profile (Alaska Department of Fish and Game)
Builds on: thecordovatimes.com

ProgrammingGrades 9–121000 EnvironmentalAlaska

Invasive Plant Hotspots or Just More Observers?

An invasive plant is a non-native plant that spreads on its own and crowds out native species. In Alaska, sightings of non-native plants are stored in the AKEPIC database, run by the University of Alaska Anchorage. Because people mostly report plants where they go looking, maps of reports can show where observers are, not only where plants are; scientists call this sampling bias.

Question: Do the hotspots on maps of Alaska invasive-plant reports change when you account for how hard people were looking (sampling effort)?

Why it stands out: You ask whether a hotspot comes from the invader or from the people looking for it.

How to do it

Choose one invasive plant with enough public records, such as those in AKEPIC. Divide the Alaska records into equal-size grid cells and three time periods. Compare raw report counts with two effort-adjusted counts from the same reporting system. One is reports per observer-day (one person searching on one day). The other is reports of your plant divided by all plant records in that cell. Redraw each map 100 times using random resamples of the records (bootstrapping) to see how stable the hotspots are. Explain that both effort measures are imperfect. Also explain that having no reports does not prove a plant is absent. Use existing records only, and collect no invasive plants.

Learn more: Alaska Exotic Plants Information Clearinghouse (AKEPIC), Alaska Center for Conservation Science, UAA
Builds on: accs.uaa.alaska.edu

EngineeringGrades 6–121000 Environmental

Filtering Laundry Microfibers Without Slowing Drainage

Microfibers are tiny threads that break off clothing in the wash. Fibers from synthetic fabrics like polyester and nylon are a kind of microplastic pollution. Washing machines usually have no filter to catch them, so many go down the drain to wastewater plants, and some reach rivers and the ocean; add-on filters can catch many of them.

Question: Which washable mesh filter traps the most model laundry fibers without slowing water drainage too much?

Why it stands out: The best filter has to catch small fibers without making the sink back up.

How to do it

Prepare contained batches of store-bought short cotton fibers as a model for laundry lint. Measure fiber length and thickness under a school microscope, and state the size range. Compare three mesh pore (opening) sizes with an empty holder as the control. Use six fresh batches of equal mass for each design. Count the fibers caught by filtering the outflow water and examining it under the microscope. Record drainage time before and after five clean-and-reuse cycles. Then improve the holder design and repeat. Keep all water and fibers for sealed disposal. Limit your conclusions to the fiber sizes you tested.

Before you start: Adult supervision: handling wet fibers, cutting tools, and catching all test water; no dry dust and nothing poured down the drain

Learn more: Microplastic Fibers (Ocean Conservancy)

Chemistry8 ideas · 7 Alaska

Categories: 0500 Chemistry

ScienceGrades 9–120500 ChemistryAlaska

Temperature Error in Indicator-Dye pH Tests

pH measures how acidic water is. The pH scale is logarithmic, so a change of one pH unit means a tenfold change in acidity, and even small changes in ocean pH matter. Scientists tracking ocean acidification need precise measurements, so it is important to know whether a cheap test, which uses an indicator dye that changes color with pH, stays accurate in cold water.

Question: A cheap indicator-dye color test can measure pH (how acidic water is). Does cold seawater make it read wrong by enough to hide real changes from ocean acidification?

Why it stands out: You check the measuring tool before anyone trusts a cheap trend in coastal water chemistry. Most projects skip this step.

How to do it

Make artificial seawater with a buffer (a mix that holds pH steady) at pH values near normal ocean pH, for example 7.7, 7.9, and 8.1. Check each one with a calibrated pH meter. Then read each with the same indicator dye in a phone light box at 5, 15, and 25 °C. Make 5 separate samples for each combination. Find the smallest pH difference the color method can reliably tell apart (its resolution). Also find how much temperature shifts the reading (temperature bias). Real pH changes a little with temperature too, so compare against the meter at each temperature.

Before you start: Adult supervision: you will use indicator dyes and buffer chemicals; wear goggles

Learn more: A primer on pH (NOAA Pacific Marine Environmental Laboratory)
Builds on: aoan.aoos.org

EngineeringGrades 6–80500 Chemistry

Correcting Lighting Errors in Phone Color Tests

Many low-cost chemical tests use paper strips that change color, and people often read them with a phone camera. Light sources such as warm bulbs, LEDs, and daylight have different color temperatures, so a camera can record the same strip as different colors. Photographers fix this with white balance, often by including a neutral gray or white reference in the photo.

Question: Different room lights can throw off a paper color test read with a phone camera. Can a printed color reference strip placed next to the test correct those errors?

Why it stands out: You fix the main reason cheap color tests fail in real kitchens and classrooms: the lighting.

How to do it

Soak paper strips in 5 known concentrations of food dye. Set them up with and without a printed gray-and-color reference strip beside them. Photograph them under 3 light sources: a warm bulb, a cool LED, and a window. Estimate dye concentration from the color, using 5 strips per condition. Use one set of strips to build your correction rule (calibration). Then test the rule blind on fresh strips someone else made, without telling you their concentrations.

Learn more: Understanding White Balance (Cambridge in Colour)

ScienceGrades 6–120500 ChemistryAlaska

Coffee-Ring Patterns to Identify Water Sources

When a drop of liquid with tiny particles in it dries, the particles collect at the edge and leave a ring; this is called the coffee-ring effect. It happens because the drop's edge gets stuck in place and water evaporates fastest there, so liquid flows outward and carries particles with it. Different dissolved minerals may leave different ring and crystal patterns, which a computer might learn to tell apart.

Question: Can the coffee-ring pattern left by a dried water drop tell well water, city tap water, rain catchment, and glacial river water apart?

Why it stands out: It builds on a Broadcom MASTERS coffee-ring winner. It asks whether a phone could be a cheap first screen for rural Alaska water (not a safety test).

How to do it

First make reference solutions: table salt (sodium chloride) and Epsom salt (magnesium sulfate) at 4 concentrations each. Dry 10 drops of each on clean glass to build a reference set. Then collect 4–5 real water types on 3 different days. Photograph each dried drop with a clip-on phone microscope. Measure ring width and crystal coverage in free ImageJ software. Train a simple classifier (a program that sorts samples into groups) on two collection days. Test it blind on the third day, with distilled water as a blank.

Before you start: Adult supervision for sampling near rivers. Follow the ISEF Field Work Safety Plan.

Learn more: Why do coffee stains dry to leave a ring? (The Naked Scientists)
Builds on: snexplores.org

ScienceGrades 9–120500 ChemistryAlaska

Crushed Shells as an Ocean Acidification Buffer

The ocean absorbs carbon dioxide (CO2) from the air, which lowers seawater's pH and makes it more acidic; this is called ocean acidification. It also leaves less carbonate in the water, which animals like oysters and clams need to build their calcium carbonate shells. This project tests whether waste shells from seafood processing can buffer seawater, meaning help keep its pH from dropping.

Question: Can crushed crab, clam, and oyster shells, which seafood processors throw away, keep seawater from becoming more acidic when CO2 is added?

Why it stands out: It turns a waste product of Alaska fisheries into a test of a possible local fix.

How to do it

Clean, crush, and sieve each shell type to one grain size (for example 1–2 mm). Add 0, 5, 10, or 20 g per liter to jars of artificial seawater, 3 jars each. Add the same amount of freshly opened club soda (carbonated water) to every jar. Log pH every 10 minutes for 2 hours with a calibrated pH meter. Also measure buffering capacity, which is how much acid the water can neutralize. Titrate by counting the drops of dilute vinegar needed to lower pH by 0.5. Compare your starting values with Alaska Ocean Acidification Network monitoring data.

Learn more: What is Ocean Acidification? (NOAA National Ocean Service)
Builds on: aoan.aoos.org

ScienceGrades 6–120500 ChemistryAlaska

Grit and Salt Mixtures for Breaking Ice Bonds

Road crews spread salt because it lowers the freezing point of water and keeps ice from bonding to pavement. Grit, such as sand or small gravel, does not melt ice but gives tires and boots traction. Salt moves with meltwater into nearby streams and groundwater, so finding ways to use less of it helps protect water.

Question: Does mixing gravel grit with salt reduce the amount of salt needed to break ice free from pavement?

Why it stands out: You look for a salt-saving recipe that matters for Alaska roads, cars, and nearby streams.

How to do it

Freeze same-size blocks of water in identical cups onto chilled concrete pavers. Test 4 salt doses, each with or without the same amount of sieved grit. Add two controls (comparison groups): grit only and no treatment. Use 5 blocks per group, all at one freezer temperature. After 30 minutes, pull each block sideways with a spring scale or luggage scale. Record the force needed to break the ice bond. Collect all salty water so it can be disposed of safely.

Before you start: Adult supervision: freezer work with heavy pavers

Learn more: How salt works and overview of deicing chemicals (Minnesota Pollution Control Agency Stormwater Manual)

ScienceGrades 6–80500 ChemistryAlaska

Comparing De-Icers at Alaska Winter Temperatures

De-icers work by lowering the freezing point of water, so ice melts into brine (salty water) even below 0 °C, and by keeping ice from bonding to pavement. Each chemical has a lowest practical working temperature: dry rock salt (sodium chloride) is too slow below about 15 °F (−9 °C), while calcium chloride keeps working down to about −20 °F (−29 °C). Knowing these limits helps people choose a de-icer that works in an Alaska winter.

Question: How much ice do rock salt, calcium chloride, magnesium chloride, and potassium acetate melt at -5, -15, and -25 °C, and which melts the most per dollar?

Why it stands out: It tests store-bag claims against real Alaska winter temperatures. It tells your family which bag to buy.

How to do it

Freeze identical ice discs in the same dishes. Test equal de-icer doses at -5, -15, and -25 °C. Include two controls: untreated ice and sand only. Use separate discs for the 15-, 30-, and 60-minute end points. Do 3 trials for each de-icer, temperature, and end point. Log the temperature beside the dishes. At each end point, quickly drain the brine through a chilled sieve and weigh the ice that is left. Subtract it from the starting ice weight. Note that brine clinging to the ice may cause some error. Divide the estimated grams melted by the dollar cost of the dose you actually used.

Before you start: Adult supervision. Some de-icers irritate skin and eyes, so wear gloves and goggles.

Learn more: How salt works and overview of deicing chemicals (Minnesota Pollution Control Agency Stormwater Manual)

ScienceGrades 9–120500 ChemistryAlaska

Glow Stick Reaction Rate and Temperature

A glow stick makes light by chemiluminescence. Bending it breaks a glass tube of hydrogen peroxide so it mixes with another chemical, and the energy from their reaction makes a dye glow. Like many reactions, it runs faster when warm and slower when cold, which is why a glow stick kept in the freezer lasts longer.

Question: How does temperature change the rate of a glow stick's light-producing reaction, and can that predict how long one stays bright on a -20 °C night?

Why it stands out: It pulls a college-level chemistry result out of a dollar-store toy. Then it checks the prediction on a real Alaska winter night.

How to do it

Activate identical glow sticks in water baths at 5 temperatures from 0 to 40 °C, 3 sticks each. Measure brightness in a dark box at a fixed distance. Log brightness every minute for 2 hours. If brightness decays exponentially (fades by the same fraction each minute), calculate a decay rate for each stick. Add 273.15 to each Celsius temperature to get Kelvin. Graph the natural log of decay rate against 1/Kelvin (an Arrhenius plot). From the slope, estimate an apparent activation energy, which tells how strongly temperature affects the fade rate. Predict brightness at -20 °C. Then test 3 more sticks in a freezer or outdoors, with a thermometer beside them.

Before you start: Adult supervision for hot water baths. Never cut glow sticks open.

Learn more: The chemistry of glow stick colours (Compound Interest)

ScienceGrades 9–120500 ChemistryAlaska

Scratch Width and Rust Creep Under Paint

Paint protects steel by forming a barrier between the metal and the environment. When paint is scratched, the bare steel rusts, and because rust takes up much more space than steel, it lifts the paint next to the scratch. This lets corrosion (rusting) keep spreading under the paint, a problem called undercutting or rust creep.

Question: Does a wider scratch in paint let corrosion (rust) spread farther underneath it?

Why it stands out: A controlled scratch is a tougher, more real-life test than comparing perfect coats. It echoes a Kotzebue-inspired Alaska fair project on corrosion (rusting).

How to do it

Give identical steel washers the same number of coats of one water-based paint. Use a guide to make scratches of 3 measured widths. Also include unscratched painted washers and bare washers, with 6 washers per group. Give all of them the same mild saltwater wet-and-dry cycles for 3 weeks. Before the test, practice a tape test (gently lifting loose paint with tape) on spare washers. At the end, rinse and dry every washer and use the same tape test to reveal rust under loose paint. Photograph each washer on a grid. Use free ImageJ software to measure rust creep (rust that spread past the original scratch). Say in your report that rust under paint that stays stuck down may be missed.

Before you start: Adult supervision: sharp metal and paint fumes

Learn more: Paint: protective coatings for steel (American Galvanizers Association)
Builds on: adn.com

Biochemistry, cells and microbes10 ideas · 9 Alaska

Categories: 0300 Biochemistry · 0400 Cellular & Molecular · 1300 Microbiology

ScienceGrades 9–120300 BiochemistryAlaska

Alginate Encapsulation to Protect Berry Pigments From Light

Crowberries and wild blueberries get their dark color from anthocyanins, natural pigments whose color and stability depend on pH, light, and temperature. Food scientists try to protect fragile molecules by encapsulation, which means wrapping them in a protective coating. Alginate is a gel made from seaweed that forms beads when dripped into a calcium solution.

Question: Does a thin alginate (seaweed gel) coating slow how fast light fades the anthocyanin pigment in crowberry or wild blueberry juice?

Why it stands out: It uses an Alaska berry to test a trick food scientists use to protect fragile molecules.

How to do it

Mash and strain picked or frozen crowberries (or wild blueberries) into one batch of juice. Adjust it to one pH (how acidic it is). Make juice-filled alginate beads with 0, 1, or 2 extra coating layers. Match the amount of pigment in each group. Put 5 batches per group under the same LED light, with matching jars kept in the dark as controls. Each day, dissolve a few beads in sodium citrate solution and measure the color of the liquid. That way the cloudy beads can't throw off the camera reading.

Before you start: Adult supervision. Use a field safety plan for berry picking in bear country, or use frozen berries. Use food-grade sodium alginate, calcium chloride, and sodium citrate.

Learn more: Anthocyanidins and anthocyanins: colored pigments as food, pharmaceutical ingredients, and the potential health benefits (Food & Nutrition Research, via NIH PubMed Central)

ScienceGrades 6–120400 Cellular & MolecularAlaska

Cold Acclimation and Freeze Damage in Kale

Many plants can survive freezing only after a period of cool weather first, a process called cold hardening or cold acclimation. Shorter days and cooler temperatures signal plants to get ready, and during hardening plants remove extra water from their cells so ice is less likely to form inside them. Cold-tolerant plants also make chemical changes in their cells that help them survive Alaska winters.

Question: Do kale leaves that are cold-hardened for two weeks near freezing have less cell damage after freezing than leaves grown warm?

Why it stands out: It measures cold hardening, the trick that lets plants survive northern winters, with a cheap conductivity meter. You can test garden kale after Alaska's first fall frosts.

How to do it

Grow 10 kale plants indoors. Move 5 of them to about 4 °C for 2 weeks to cold-harden. Cut same-size leaf discs from every plant. Freeze them for 0, 1, 3, or 5 freeze-thaw cycles. Soak the discs in equal amounts of distilled water. Measure electrolyte leakage (salts leaking out of damaged cells) with an inexpensive conductivity meter. Then boil the discs and measure again to get 100% leakage. Report percent damage, and count each plant (not each disc) as one replicate (one repeat of the test).

Before you start: Adult supervision for boiling leaf discs and handling hot containers. Keep water away from electrical equipment.

Learn more: Factors Affecting Cold Hardiness Development (UAF Cooperative Extension Service)

ScienceGrades 9–120400 Cellular & MolecularAlaska

Day Length and Mitosis in Onion Root Tips

Mitosis is the process in which one cell divides into two, and it happens most in growing tissue such as the tips of roots. The mitotic index tells how actively a group of cells is dividing: count the cells in mitosis under a microscope and divide by the total number of cells. Comparing mitotic indexes shows whether a change, like constant light, speeds up or slows down cell division.

Question: Does 24-hour light, like an Alaska June, change the mitotic index (share of dividing cells) in onion root tips compared with a 12-hour day?

Why it stands out: It asks a classic microscope question in a way only an Alaska student would think of. It also checks whether midnight sun scrambles a plant's daily clock.

How to do it

Grow onion bulbs over jars of water wrapped in foil, so the roots stay dark. Give the green shoots 12, 18, or 24 hours of light at the same brightness, with 6 bulbs per group. After 5 days, cut root tips at the same clock time. Soften them in dilute hydrochloric acid and stain them. Under a microscope, count how many of 500 cells are in mitosis; this is the mitotic index. Bonus: sample in the morning, at noon, and at night. See whether constant light erases the daily (circadian) rhythm of cell division. Longer days also mean more total light, so say so, or add a dimmer 24-hour group.

Before you start: Adult supervision for the stain and dilute hydrochloric acid. Wear goggles and gloves.

Learn more: Mitotic Index (MHCC Biology 112, Open Oregon Educational Resources)

ScienceGrades 9–120400 Cellular & MolecularAlaska

Day Length and Stomatal Density in Radish Leaves

Stomata are tiny pores on the surface of a leaf that let gases move between the inside of the leaf and the air. Scientists count them by painting clear nail polish on a leaf, peeling it off, and looking at the print under a microscope. This project tests whether long Alaska summer days change how many stomata a leaf makes.

Question: Do radish leaves grown under 20-hour days, like an Alaska summer, make more stomata (tiny leaf pores) than leaves grown under 12-hour days, even with the same total light per day?

Why it stands out: It separates day length from total light. Most light-and-plant projects mix these two up.

How to do it

Grow 12 radish plants per group under 12-, 16-, and 20-hour days (photoperiods). Dim the long-day lights so each group gets the same total light per day (daily light integral). Check this with the same light meter. Add a fourth group with 20 hours at full brightness to test extra daily light. Paint clear nail polish on leaves to make impressions, and lift them off with clear tape. Count stomata per square millimeter (stomatal density) under a microscope, and average 3 spots per plant. Compare plants, not microscope spots. The matched-light groups differ in both day length and brightness, so say that neither one alone can be called the cause.

Before you start: Adult supervision. Use nail polish in a well-aired area away from flames and heaters.

Learn more: Stomata Printing: Microscope Investigation (California Academy of Sciences)

ScienceGrades 9–120400 Cellular & MolecularAlaska

Detecting Invasive European Green Crab With eDNA

The European green crab is an invasive species: it arrived from Europe and harms local sea life by eating young salmon and king crab, destroying eelgrass, and outcompeting native species. All animals shed cells into the water, and this environmental DNA (eDNA) can be filtered from a seawater sample and tested to show a species is nearby. NOAA and partners, including the Metlakatla Indian Community, use eDNA and traps together to watch for green crab in Southeast Alaska.

Question: Can environmental DNA (eDNA) in seawater detect invasive European green crab at sites where traps catch them, and stay negative where traps don't?

Why it stands out: It follows a Juneau student's ISEF-winning eDNA (environmental DNA) path into one of Alaska's newest invasive species problems.

How to do it

Green crab was first found in Alaska in 2022 through Metlakatla Indian Community trapping. With an agency or university mentor, filter 1-liter seawater samples, 3 per site. Sample sites where traps have caught green crab, sites where they haven't, and unknown sites. Carry bottles of clean water as field blanks to catch contamination. In the lab, run a published green crab qPCR assay (a test that detects and counts DNA copies). Include positive controls (samples known to have crab DNA) and inhibition checks (tests that nothing in the sample is blocking the reaction). Compare detections with trap records. Clean gear between sites, and never move live animals or plants.

Before you start: Mentor and lab required: use an agency or university qPCR lab. Before starting, get site permissions and SRC review (the fair's Scientific Review Committee) of sampling, extraction chemicals, and test methods. Follow lab rules for protective gear. Shoreline sampling needs adult supervision and a field safety plan.

Learn more: Using eDNA to Monitor Alaskan Waters for Invasive European Green Crabs (NOAA Fisheries)
Builds on: fisheries.noaa.gov

ScienceGrades 9–120300 Biochemistry

Bead Size and Immobilized Enzyme Reaction Rate

Enzymes are proteins that speed up chemical reactions; amylase, for example, breaks starch into sugar. Chemists can trap an enzyme inside gel beads, called immobilization, so the enzyme can be filtered back out of the mixture. For the enzyme to work, the starch has to diffuse (spread) into the bead, and this project tests whether that slows the reaction.

Question: Do larger alginate gel beads slow how fast a trapped (immobilized) enzyme breaks down starch, because starch diffuses into them too slowly?

Why it stands out: It shows whether the slowdown comes from the enzyme or from the traffic jam getting into the bead. Engineers face the same problem in enzyme reactors.

How to do it

Immobilize the same total amount of amylase (an enzyme that turns starch into sugar) in calcium alginate beads (a gel made from seaweed). Make beads in 3 sizes, and compare them with the same amount of free enzyme. Run 5 separate batches per size in starch solution at a fixed temperature and stirring speed. Use enzyme-free beads as a blank. Every few minutes, remove a small sample of liquid and add dilute iodine, which turns blue-black when starch is present. Measure how fast the blue-black color fades with a phone color app. Bonus: repeat at 5 °C to see if cold makes the bead-size effect bigger.

Before you start: Adult supervision. Dilute iodine stains and irritates skin, so wear goggles and gloves.

Learn more: Immobilization of Lactase Enzyme on Alginate Beads – A Quick Test (ChemEd X)

ProgrammingGrades 9–121300 MicrobiologyAlaska

Does Water Temperature Predict Shellfish Toxin Levels?

Paralytic shellfish poisoning (PSP) is a dangerous illness caused by saxitoxin, a toxin made by certain algae during harmful algal blooms. Clams and mussels filter these algae from seawater and store the toxin, and cooking or freezing does not remove it. Alaska has among the highest PSP rates in the nation, and shellfish that people harvest themselves are not tested.

Question: Does recent water temperature help predict shellfish toxin (saxitoxin) readings at an Alaska monitoring site?

Why it stands out: The warning has to hold up in a new season and with the gaps between real samples.

How to do it

On the Alaska Harmful Algal Bloom (AHAB) Network data portal, pick one site with at least three seasons of water temperature data. It also needs one shellfish toxin measured the same way the whole time. Build two models. One uses recent average water temperature. The baseline uses only the time of year. Fit both on earlier seasons and test them on the last season. Count false alarms, and count missed warnings (times the toxin crossed the regulatory limit with no warning). Then remove some sample dates to see how much later a warning would come. This project cannot tell anyone whether shellfish are safe to eat.

Learn more: Alaska has among the nation's highest rates of paralytic shellfish poisoning (Anchorage Daily News)
Builds on: ahab.aoos.org

ScienceGrades 6–120300 BiochemistryAlaska

Cold-Induced Sweetening in Stored Potatoes

When potatoes are stored too cold, some of their starch turns into sugar, a change called cold-induced sweetening. The extra sugar makes the potatoes taste sweet and makes fries brown or darken before they are cooked through inside. Potato experts warn that potatoes stored below about 40 °F (4 °C) often make dark fries.

Question: Does storing potatoes at root-cellar temperatures turn their starch into sugar (cold-induced sweetening), and does that make them brown faster when cooked?

Why it stands out: It shows cold sweetening, a plant's survival trick. You see it in the same potatoes Alaska families keep in cellars all winter.

How to do it

Store one potato variety at about 2 °C, 8 °C, and 20 °C, with 3 separate bags at each temperature. Use glucose test strips made for testing food. Calibrate them against known glucose solutions, not table sugar. Also do a spike test: add a known amount of glucose to potato juice and check that the reading rises as expected. Every week for 6 weeks, test juice from 5 potatoes per bag. Then oven-bake equal-size slices for the same time. Score browning from photos with free ImageJ software. Average the readings within each bag, and compare the glucose and browning trends. Bonus: recondition some cold potatoes by warming them for a week, then test again.

Before you start: Adult supervision for hot oil or the oven when frying.

Learn more: Never Store Potatoes On Ice (Ask Dr. Potato, Idaho Potato Commission)

ScienceGrades 6–80300 BiochemistryAlaska

Light and Freezing Stability of Anthocyanin Plant Pigments

Anthocyanins are water-soluble pigments that give many flowers, berries, and vegetables like red cabbage their red, purple, and blue colors. Their color depends on pH (how acidic something is): red in acid and bluer as pH rises. Light, temperature, and pH all affect how long the color lasts, which matters for anyone hoping to use them as natural dyes.

Question: Do anthocyanin pigments from fireweed flowers, crowberries, or red cabbage keep their color longest under light and repeated freezing and thawing?

Why it stands out: It tests whether the flower that colors Alaska roadsides every July could be a long-lasting natural dye.

How to do it

Pick a cup or two of July fireweed petals and freeze them until you start. Make water-and-vinegar extracts from fireweed, crowberries, and red cabbage. Start them all at the same color strength and pH (how acidic they are). Split each extract into 4 groups: fridge, room temperature in the dark, room temperature under LED light, and 3 fixed freeze-thaw cycles. Use 3 vials per group, plus matching water-and-vinegar blanks. Check that the lit and dark vials stay at the same temperature. Give the freeze-thaw group a fixed schedule over the same 2 weeks. Each day, photograph every vial in a light box with a white reference card. Measure the percent of color remaining with a free color app.

Before you start: Adult supervision. Make a plant-collection safety plan. Pick away from busy roads, and avoid wildlife.

Learn more: Anthocyanidins and anthocyanins: colored pigments as food, pharmaceutical ingredients, and the potential health benefits (Food & Nutrition Research, via NIH PubMed Central)

ScienceGrades 6–121300 MicrobiologyAlaska

Yeast Fermentation of Birch Sap and Syrup

Yeast are tiny fungi that ferment sugar, turning it into carbon dioxide gas and alcohol. In Alaska, people tap birch trees in spring and boil the sap down into birch syrup. Birch sap sugar is mostly glucose and fructose, while maple sap is mostly sucrose (table sugar), so yeast may not use them at the same rate.

Question: Does baker's yeast produce gas at different rates in birch products and sugar water that start with the same refractometer (Brix) reading?

Why it stands out: An Alaska forest product goes up against table sugar and maple syrup, using the same yeast and a carefully matched starting sugar level.

How to do it

Buy pasteurized birch sap, birch syrup, maple syrup, and table sugar. Dilute each to the same Brix reading on a refractometer (a tool that estimates dissolved sugar from how the liquid bends light). Add the same amount of baker's yeast to four bottles of each liquid. Keep one bottle of each with no yeast as a control. Hold all bottles at one temperature for 48 hours. Run tubing from each bottle into an open water-displacement setup (the gas pushes water out of an upside-down container), so no pressure can build. Compare how fast and how much gas forms. Equal Brix readings may not mean equal sugar the yeast can use. Fill out ISEF Form 3 (Risk Assessment), and never drink the mixtures.

Before you start: Adult supervision: pasteurized ingredients, baker's yeast only, pressure-free gas collection, Form 3, and no drinking or eating the mixtures

Learn more: Birch Sap Sugaring Off (Alaska Science Forum, UAF Geophysical Institute)

Health, medicine and people14 ideas · 9 Alaska

Categories: 1200 Medicine & Health · 0200 Behavioral & Social

ProgrammingGrades 6–120200 Behavioral & SocialAlaska

AI Image Generators and Alaska Stereotypes

AI image generators make pictures from text prompts after learning from huge collections of images. Researchers have found that these tools often exaggerate stereotypes, for example showing people in some jobs as almost always one race or gender. Seeing biased images can strengthen people's own stereotypes, so it matters how AI pictures real places like Alaska.

Question: When asked for everyday Alaska scenes, do AI image generators show stereotypes such as igloos, dog sleds, and fur parkas?

Why it stands out: It adds an Alaska twist to a national-winning study of AI bias. Nobody has to be surveyed.

How to do it

Write 10 everyday prompts, such as 'a teenager in Bethel after school' or 'a family dinner in Anchorage in July.' Run each one 5 times in 3–4 image generators, and record the version and date. Before you start, write a coding checklist (scoring rubric) for every image: igloo, dog sled, fur parka, snow in a summer scene, modern houses, trucks and cars. Have a friend score half the images blind, without knowing which generator made them. Then measure how often you two agree (inter-rater agreement). Compare each generator's scores with census data and real photos of those towns.

Learn more: AI image generators tend to exaggerate stereotypes (Science News Explores)
Builds on: societyforscience.org

ProgrammingGrades 9–120200 Behavioral & SocialAlaska

Tracing Alaska's '95% of Food Is Imported' Claim

People often say that Alaska imports 95% of its food. Alaska Public Media found the number traces back to a rough estimate from 1977 and has not been checked or updated since, even though it is still widely repeated. Counting and coding how a claim appears across many documents is a research method called content analysis.

Question: How often do news stories and official reports repeat the claim that Alaska imports 95% of its food, and how often do they say where the number came from?

Why it stands out: Alaska Public Media traced the famous number to a rough 1970s estimate that 'lacks units.' You map how a shaky statistic spread.

How to do it

Before you search, write down your search terms and sources. Also write a citation scale: none, vague, names a source, or links to a real study. Collect every use of the claim you can find since 2000 in news archives, legislative documents, and websites. Aim for 100 or more. Code each one for year, type of outlet, and citation score. Have a second person code 20% of them to check that you agree (inter-rater reliability). Chart how the claim spread over time, and compare citation rates across outlet types.

Before you start: none (an optional phase 2 survey would need IRB approval, a review board that checks studies with people)

Learn more: That figure you've heard on how much food Alaska imports? It's not real (Alaska Public Media)

ProgrammingGrades 9–120200 Behavioral & SocialAlaska

Modeling Food Sharing in Subsistence Communities

In rural Alaska, many families harvest fish, game, and plants for food, a way of life called subsistence. Extended families often work together to harvest and process food and then share it widely within and between communities. An agent-based model is a computer simulation in which each 'agent' (here, a household) follows simple rules, so you can see how the whole group does over time.

Question: In an agent-based computer model of households that hunt and fish, how variable do yearly harvests have to be before sharing food leads to better survival than keeping it?

Why it stands out: It takes a 2025 ISEF-winning game-theory idea and asks a survival question. That question matters anywhere a bad fishing year is real.

How to do it

Build a simple agent-based model in Python or NetLogo. In it, households catch a random amount of food each year. Each one follows one strategy: always share, never share, or reciprocal sharing (share only with households that shared before). Change harvest variability, meaning how much harvests swing from year to year (10 levels, 100 runs each). Measure how many households survive 50 years and which strategy spreads. Use the year-to-year variability in real ADF&G salmon counts as one test level. Run a sensitivity check: see whether your answer changes when you tweak the rules. Say clearly that a model shows what is possible, not how real communities behave.

Learn more: The Harvest and Use of Wild Resources by Communities Within or Near Northern Alaska Parklands (National Park Service)
Builds on: societyforscience.org

ProgrammingGrades 9–120200 Behavioral & SocialAlaska

Seasonal Affective Disorder in Google Search Trends

Many people feel down when days get shorter in fall and winter and feel better when longer daylight returns in spring. When these mood changes are serious, doctors call it seasonal affective disorder (SAD), a type of depression that usually starts in late fall or early winter. Light therapy and vitamin D are treatments for this winter pattern of SAD.

Question: As days get shorter, do Alaskans search for 'tired,' 'light therapy lamp,' and 'vitamin D' more than people in Florida do, a possible sign of winter seasonal affective disorder?

Why it stands out: It studies winter blues across a whole state. You don't have to survey a single person.

How to do it

Download 5 or more years of monthly Google Trends data for 6 sleep and mood terms in Alaska, Washington, and Florida. Trends scores each state on its own 0 to 100 scale. So compare the shape of each state's yearly cycle (winter peak divided by summer low), not raw scores. Match each month to its average day length. Add a control term with no link to daylight ('pizza') and a holiday term ('gift') to spot December effects. Flag months where Alaska's search volume is too low for Trends to report.

Learn more: Seasonal Affective Disorder (National Institute of Mental Health)

ScienceGrades 9–120200 Behavioral & SocialAlaska

Showing Uncertainty on Glacier Flood Risk Maps

In Juneau, a lake called Suicide Basin forms beside Mendenhall Glacier, held back by the glacier's ice. When the water suddenly breaks out, it causes a glacial outburst flood down the Mendenhall River, which has happened almost every year since 2011. Scientists say the size of each flood is hard to predict, so flood maps have to deal with uncertainty.

Question: Do people read a flood-risk map more accurately when it shows a shaded zone of uncertainty instead of one sharp flood line?

Why it stands out: It tests whether being honest about uncertainty makes a map more useful. That is a real question for a town that has flooded from a glacier lake nearly every year since 2011.

How to do it

Make a fictional flood map of a river town, based on Juneau's glacial outburst floods. Make two versions: one with a sharp flood boundary, and one with the same boundary plus a shaded uncertainty band. Get IRB approval first (a review board that checks studies with people). Then randomly assign 60 or more volunteers to one version. Ask 6 questions you wrote ahead of time, such as 'Should this household move its car?' Score answers with an answer key made in advance. Keep colors and wording the same. Write questions so the band doesn't simply give away the answers.

Before you start: Needs IRB approval.

Learn more: New study highlights findings and predictions for Juneau glacial outburst flood (U.S. Geological Survey)

ScienceGrades 9–120200 Behavioral & SocialAlaska

Snow-Covered Sidewalks and Pedestrians in the Street

When plows clear roads, they often leave a ridge of snow along the edge of the sidewalk, and many sidewalks are not shoveled. Then pedestrians may walk in the road, where they face both icy pavement and passing cars. This is a winter safety problem in snowy towns, including in Alaska.

Question: After Alaska snowstorms, how many pedestrians walk in the street instead of on the sidewalk, and does it depend on how well the sidewalk is cleared?

Why it stands out: It counts how many people end up walking in the road after storms. Every Alaska town faces this winter safety question.

How to do it

Pick 6 blocks or intersections. Count pedestrians on the sidewalk versus the road at the same times of day. Count on dry days and on the 1–3 days after each of 5 storms. Rate sidewalk clearing on a 0–3 scale (0 = not cleared, 3 = fully cleared), and record temperature and new snow depth. Compare each site with itself before and after storms. Don't talk to people, film faces, or record who they are. Snow and plowing happen together, so report correlations (links), not proof of cause.

Before you start: Adult supervision. Make a winter roadside field safety plan. Watch from a safe public spot away from traffic, snowplows, and ice.

Learn more: Ice, snow and slippery sidewalks: For many seniors, winter is 'hip season' (archived pedestrian safety site, UNT Libraries government documents)

ProgrammingGrades 6–120200 Behavioral & Social

Gender Pronouns in AI Chatbot Job Stories

AI chatbots run on large language models, which learn to write by studying huge amounts of text written by people. A UNESCO study found that these models often show gender bias, for example describing women in home and family roles far more often than men. Counting pronouns in AI-written stories is one simple way to measure this kind of bias.

Question: When an AI chatbot writes a short life story for a fictional geologist, nurse, or bush pilot, does the job change whether it uses he, she, or they?

Why it stands out: It takes a national-winning project on AI image bias and applies it to text. Anyone can repeat your test.

How to do it

Pick 8 jobs, such as nurse, engineer, geologist, bush pilot, and kindergarten teacher. Add plain 'scientist' as the comparison. Write one gender-neutral prompt. Ask the same AI model for 30 short fictional biographies per job, and repeat on 3 different days. Count he, she, and they using a coding rule you write before you start. List unclear cases instead of guessing. Record the model name and version so anyone can rerun your test.

Learn more: Generative AI: UNESCO study reveals alarming evidence of regressive gender stereotypes (UNESCO)
Builds on: societyforscience.org

EngineeringGrades 9–121200 Medicine & Health

Mapping Pressure Under Seat Cushions with DIY Sensors

Sitting or lying in one position for a long time can squeeze skin against bone and cut off blood flow, causing pressure sores (also called pressure ulcers or bedsores). People who use wheelchairs are at higher risk, so seat cushions are designed to spread out weight and lower the highest pressure points. A pressure-mapping mat shows where those high-pressure spots are.

Question: Can a low-cost pressure-mapping mat show which spacing of cut foam channels lowers the peak pressure under a model seat?

Why it stands out: The pressure map shows whether a cushion spreads force out or just moves the high-pressure spot.

How to do it

Build a grid of pressure sensors from pressure-sensitive conductive film, such as Velostat, and a school microcontroller (a small programmable circuit board). Calibrate each sensor cell with known weights. Load and unload each cell to check for hysteresis (readings that lag behind), and press nearby cells to check for crosstalk (false signals). Compare three foam-channel spacings of equal thickness with uncut foam, using five separately built cushions per design and the same weighted seat model. Measure peak pressure and contact area before and after 100 compression cycles. Then improve one pattern and retest it on five fresh cushions. Do not test on people or make medical claims.

Before you start: Adult supervision: secured loads, cutting tools, and low-voltage wiring

Learn more: Pressure Sores (MedlinePlus, U.S. National Library of Medicine)
Builds on: societyforscience.org

EngineeringGrades 6–121200 Medicine & HealthAlaska

Insulated Pouches to Keep Vials From Freezing

Some medicines, such as insulin, must not freeze, and insulin that has been frozen should not be used. In an Alaska winter, a person waiting outside at a bus stop can face temperatures far below freezing. Insulation slows heat loss, so a well-designed pouch can delay how fast a vial cools to 0 °C.

Question: Which reusable insulated pouch keeps a vial of water from reaching freezing the longest during a simulated Alaska bus-stop wait?

Why it stands out: The closure has to protect the vial just as well as the insulation does.

How to do it

Compare three pouch designs that cost the same, a store-bought insulated lunch pouch, and a vial with no insulation (control). Use three separate pouches per design. Start sealed water vials at 20 °C. Run six freezer tests at a logged temperature. Measure how long each vial takes to cool to 0 °C and its temperature after 30 minutes. Keep the vial position and starting amount of water the same. Then improve the pouch closure and test again. The 30-minute wait is a scenario you chose, not a measured bus wait. Use no real medicine and make no claims about storing drugs.

Before you start: Adult supervision: freezer handling, sealed plastic vials, and temperature loggers

Learn more: Insulin and syringes: storage and safety (MedlinePlus, U.S. National Library of Medicine)

EngineeringGrades 6–121200 Medicine & HealthAlaska

Water-Saving Handwashing Designs Tested on Model Hands

In 2020, about 1,700 homes in roughly 31 Alaska Native communities had no running water or sewer, so families haul water and store it in drums. Handwashing helps prevent disease, but every liter counts when water must be carried. Low-water designs, such as a tippy-tap (a hanging water jug tipped by a foot pedal), aim to rinse hands well with less water.

Question: Can a pulsed-flow nozzle remove the same amount of washable gel from a model hand while using less water?

Why it stands out: A rural Alaska water-budget problem becomes a repeatable test with no human testers.

How to do it

Coat identical plastic model hands with equal amounts of washable colored gel (a stand-in for dirt), and hold them in a fixed position. Compare three designs: steady flow, pulsed flow (water in short bursts), and a tippy-tap. Test each with six fresh gel coatings. Measure the water used, and photograph the color left behind under fixed lighting. Adjust the most promising design. Then test six new coatings with a different gel pattern, using the original design as your control. Report gel removed per liter. Make no claims about killing germs or preventing illness.

Before you start: Adult supervision: tool use, keeping water contained, and protected pump wiring

Learn more: In rural Alaska, access to water is preventive medicine (Anchorage Daily News, sponsored by ANTHC)

ScienceGrades 6–80200 Behavioral & Social

Pedestrian Countdown Signals and Late Crossings

Pedestrian countdown signals show how many seconds are left to cross the street. A Federal Highway Administration study of more than 300 intersections in Charlotte and Philadelphia found that pedestrian crashes fell about 9% after countdown signals were installed. Engineers still want to know how the timers change people's crossing choices.

Question: At crosswalks with countdown timers, do more people start crossing late, after the flashing 'Don't Walk' hand begins, than at crosswalks without timers?

Why it stands out: It gives a city traffic engineer local data on late crossing starts at intersections with and without countdown timers.

How to do it

From a public sidewalk, watch 4 similar intersections with the same number of lanes and speed limit. Two should have countdown timers and two should not. Do 10 one-hour sessions per site at matched times. Record the signal phase (Walk, flashing Don't Walk, or steady Don't Walk) when each person stepped off the curb. Don't talk to anyone, film faces, or write down who people are. Compare the share of late starts, using each session as one data point.

Before you start: Adult supervision. Make a roadside field safety plan and watch from well back from traffic. Public observation only: no talking to people and no records that identify anyone. Grades 6–8 need teacher sign-off.

Learn more: Safety Evaluation of Pedestrian Countdown Signals (Federal Highway Administration)

EngineeringGrades 9–121200 Medicine & HealthAlaska

Comparing Filters for DIY Box-Fan Air Cleaners

Wildfire smoke contains very small particles that are harmful to breathe, and Alaska summers often bring smoky days. A do-it-yourself (DIY) air cleaner made from a box fan and an air filter is a low-cost way to reduce smoke indoors. Filters are rated by MERV (Minimum Efficiency Reporting Value), and the EPA recommends MERV 13 filters to catch the very small particles in smoke.

Question: Which filter rating on a box-fan air cleaner removes test particles fastest for each dollar and each watt of electricity?

Why it stands out: The cheapest filter has to show how much clean air each dollar buys.

How to do it

Work with a lab mentor. Compare three filter MERV ratings on the same guarded box fan inside a sealed test chamber. Make particles with a contained salt-water mist, and let the mist dry before you measure. Match the humidity and starting particle level each time. Record how fast particle levels drop (the decay rate) and how much electricity is used, with six trials per filter. Add two controls: fan with no filter, and no fan. Check your two particle sensors side by side first. Correct for particles that settle on their own, then estimate the clean air delivery rate (CADR, how much particle-free air the cleaner provides). Compare filter and energy costs. Call this a particle-removal test, not a direct test of wildfire smoke or health effects.

Before you start: Mentor/lab: sealed particle chamber, contained mist, drying checks, safe fan loading, and supervised disposal

Learn more: Research on DIY Air Cleaners to Reduce Wildfire Smoke Indoors (U.S. EPA)
Builds on: alaskapublic.org

EngineeringGrades 6–121200 Medicine & Health

Designing a Dispenser Gate That Releases One Bead

Many machines must release exactly one small object at a time, such as automatic pill dispensers and parts feeders in factories. The part that controls the release is often called a gate or escapement. Common failures are jams, double drops (two objects at once), and missed drops.

Question: Which dispenser gate design releases exactly one bead at a time across different bead sizes?

Why it stands out: A dispenser that works 99 times out of 100 still has a failure worth finding.

How to do it

Build three copies each of three gate designs, plus a simple sliding gate as a baseline to compare against. Run 100 drops per copy with each of three wooden-bead sizes, in random order. Keep the hopper fill level within a set range. Count jams, double drops, and missed drops, then improve the best gate. Test three new copies on a bead shape you have not used yet, 100 drops each. Use no pills and no human testers.

Before you start: Adult supervision: cutting tools, moving gates, and small-part containment

ScienceGrades 9–121200 Medicine & Health

Do Cold Fingers Affect Pulse Oximeter Accuracy?

A pulse oximeter is a small device, usually clipped on a fingertip, that shines light through the skin to estimate blood oxygen level (SpO2) and pulse rate. The FDA says the readings are estimates and can be affected by poor circulation, skin temperature, skin pigmentation, nail polish, and other factors. The FDA advises making sure your hand is warm before taking a reading.

Question: Do low-cost pulse oximeters disagree more with a higher-quality reference oximeter when fingers are slightly cool?

Why it stands out: A missing reading may matter as much as a wrong one.

How to do it

First get IRB approval (from an Institutional Review Board, which protects people in studies), and have a health professional review your plan. Then enroll 20 volunteers as they come indoors on a cold day. Measure finger skin temperature. Take three readings each from two low-cost oximeters and one reference oximeter, changing the order of devices each time. Repeat after each person has warmed up indoors for 15 minutes. Compare how far the low-cost devices are from the reference (bias), how long each takes to give a reading, and how often each fails to give one. Give no one any health feedback. Remember that the reference device is not perfect either.

Before you start: Needs IRB approval; mentor/lab: a health professional reviews the plan, no making anyone colder on purpose, and no health feedback to any person

Learn more: Pulse Oximeter Basics (U.S. Food and Drug Administration)

Engineering and robotics11 ideas · 5 Alaska

Categories: 0800 Engineering

EngineeringGrades 9–120800 EngineeringAlaska

Does Freezing Weaken Zinc's Rust Protection?

Steel rusts (corrodes) faster in wet, salty conditions. Connecting zinc to steel protects it through cathodic protection: the zinc corrodes first and is used up in place of the steel, so zinc is called a sacrificial anode. This project tests whether Alaska's freeze-thaw cycles change how well that protection works.

Question: Zinc connected to steel can protect the steel from rusting. Does repeated freezing and thawing weaken that protection in salty, wet sand?

Why it stands out: This follow-up to a Kotzebue-inspired project asks whether zinc's rust protection survives winter freezing and thawing.

How to do it

Fill 24 cups with clean sand, the same amount of salt, and the same amount of water. Put one identical steel strip in each. Wire a zinc strip to the steel in half the cups, and leave the rest unprotected. For 20 days, put six cups of each kind through a daily freeze-thaw cycle. Keep six of each at a steady cold temperature just above freezing, such as in a fridge. Log temperature with a probe in each group. Measure the electric current between zinc and steel during thaw. At the end, clean the steel with one teacher-approved method and weigh how much metal was lost.

Before you start: Adult supervision: metal edges, freezer handling, and teacher-approved rust removal

Learn more: What is Cathodic Protection?, American Galvanizers Association
Builds on: adn.com

EngineeringGrades 9–120800 EngineeringAlaska

Cold Weather and Low-Cost Motion Sensors

A passive infrared (PIR) motion sensor detects changes in the infrared heat given off by warm objects. A microwave motion sensor sends out radio waves and senses the echoes from moving objects. Because PIR sensors detect heat differences, cold air could change how well they spot a warm target, which matters for ideas like moose-warning systems.

Question: How does cold weather change how far away low-cost motion sensors can detect a warm model target?

Why it stands out: You take a moose-warning idea and test what low-cost sensors can really detect.

How to do it

Compare two inexpensive motion sensors. One is a passive infrared (PIR) sensor, which senses changes in heat. The other is a low-cost microwave motion sensor. Pull a warm-water target on a rope across a closed yard at marked distances and three angles. Do five passes per setting in each of three safe air temperature ranges. Log the target and air temperature, the speed, and how often each sensor detects the target. Also count false alarms when the yard is empty. Repeat everything with a room-temperature target. Measure distances with a tape measure. Use no people or animals as targets, and don't claim the sensors can detect moose.

Before you start: Adult supervision: limits on time in the cold, and a field safety plan that keeps you away from roads, water, and wildlife

Learn more: PIR Motion Sensor, Adafruit Learning System
Builds on: societyforscience.org

EngineeringGrades 9–120800 EngineeringAlaska

Insulating Foundations on Thawing Permafrost

Permafrost is ground that stays frozen year-round, and in many parts of Alaska it contains a lot of ice. Heat from a building can thaw the ground below, and as the ice melts the soil settles unevenly, cracking or tilting the building. Builders protect permafrost with insulation, gravel pads, and raised or adjustable foundations that keep the ground frozen.

Question: Can a ring of rigid foam insulation around a model footing (the base that supports a building) slow its settling as icy soil thaws?

Why it stands out: Simple, replaceable insulation takes on the thawing-permafrost problem behind Alaska foundation damage.

How to do it

Freeze equal columns of clean sand and water, with a measured thin ice layer under each footing. Compare footings with rigid foam insulation rings of two widths with a footing that has no insulation. Use six separate columns per design and equal loads. Log temperature and settlement (sinking) during the same thaw period. Improve the insulation design. Then test six new columns against six uninsulated controls. Treat your results as a model, not as building advice.

Before you start: Adult supervision: secured loads, freezer handling, and spill containment

Learn more: What do I need to be aware of when building on permafrost?, Your Northern Home (Cold Climate Housing Research Center)

EngineeringGrades 6–120800 EngineeringAlaska

Keeping Sensor Vents from Icing Shut

Outdoor weather sensors sit inside housings with vents that let air reach the sensor. In cold, humid air, frost can build up on vents and block airflow, so a humidity sensor responds slowly or gives wrong readings. The shape of the vent may change how quickly it ices up, which matters for instruments that run unattended through an Alaska winter.

Question: Which vent shape keeps a weather sensor housing working after frost forms?

Why it stands out: A tiny vent opening becomes a weak point you can measure in an Alaska winter instrument.

How to do it

Build three copies each of three vent types: straight, downward-facing, and baffled (with inside walls that block a straight path). Give them all the same opening area. Add a fully open-vent reference. Give each housing five equal moisture-and-freeze cycles in random order. Measure frost coverage, airflow speed at a fixed fan setting, and how quickly the humidity sensor responds to a change. Compare each housing with its own results from before freezing. Improve the best vent and test three new copies.

Before you start: Adult supervision: freezer handling, contained moisture, and low-voltage electronics

EngineeringGrades 6–120800 Engineering

Best Plow Blade Angle for a Battery Robot

A snowplow blade can face straight ahead or be angled to push snow off to one side. Battery-powered robots and small electric plows have limited energy, so the blade angle that moves the most snow per unit of energy (watt-hour) matters. Measuring energy use directly lets you compare blade designs fairly.

Question: Which blade angle lets a small robot clear the most snow for the battery energy it uses?

Why it stands out: The plow has to earn its battery use by moving snow you can measure.

How to do it

Fit one small robot with the same straight blade set at three angles. Before each run, reset a short course with the same snow mass, depth, and packing. Do six runs per angle. Measure energy use with a logger that records current and voltage. Compare grams of snow cleared per watt-hour of energy. Use the blade set straight across (square-on) as your baseline, and start every run with the battery at the same charge. Then adjust the best angle and repeat six runs with fresh snow.

Before you start: Adult supervision: moving blades, time in the cold, and protected low-voltage wiring

ScienceGrades 6–120800 Engineering

Folded Cardboard Roof Strength When Damp

Folding a flat sheet into a pattern, as in origami, can make it much stiffer without adding material, because the shape (geometry) carries the load. Engineers study folded paper structures for uses like quick-to-build shelters and bridges. This project tests whether the fold that is strongest when dry is still the strongest after the cardboard gets damp.

Question: Which cardboard fold pattern stays strongest after it gets damp?

Why it stands out: The fold that is strongest when dry may lose its lead once it gets wet.

How to do it

Compare three fold patterns and one flat panel. Give them all the same roof span and the same card-stock mass. Build five new panels of each pattern for each of two conditions. One condition is dry. The other is misted evenly with a measured amount of water. Add equal small sandbags across each panel until it buckles (bends and collapses). Then calculate the failure load per gram (how much weight it held for each gram of card). Keep the supports fixed. Describe your test as a model of a damp roof under load.

Before you start: Adult supervision: use stable supports, keep the loads contained, and keep hands clear of collapsing panels

Learn more: Paper Tubes Make Stiff Origami Structures, Georgia Tech
Builds on: societyforscience.org

EngineeringGrades 9–120800 Engineering

Slip Detection for Gentle Robot Grippers

A robot gripper must squeeze hard enough to hold an object but not so hard that it crushes soft items. When an object starts to slip, it makes tiny vibrations that a sensor can pick up. Slip detection lets a robot tighten its grip only when needed, the way human fingertips do.

Question: Can an inexpensive vibration sensor help a robot gripper hold soft objects with less squeezing force?

Why it stands out: The robot must notice a slipping object before it squeezes harder.

How to do it

Fit a low-voltage gripper with a piezoelectric vibration sensor (it makes a small voltage when it vibrates) and a calibrated force sensor. Tune two controllers on separate foam practice objects. One squeezes harder when it senses a slip; the other uses a fixed force. Then lock both settings. Test each controller 10 times on each of three new foam stiffnesses, in random order. Record drops, dents, and peak force. Only compare force among grips that hold the object for the same 10 seconds.

Before you start: Adult supervision: pinch points, low-voltage wiring, and tools

ScienceGrades 6–80800 Engineering

Which Shoe Sole Grips Ice Best?

Ice is slippery because water molecules at its surface move easily, and how slippery it is depends on temperature. Researchers found that ice friction is lowest around -7 °C (19 °F) and rises at colder temperatures. Grip is measured with the coefficient of friction, the force needed to slide an object divided by its weight.

Question: Which shoe sole material grips ice best as the ice temperature changes?

Why it stands out: You measure winter grip without anyone stepping onto the test ice.

How to do it

Test four sole materials, including smooth rubber, with three separate samples of each. Attach each sample to the same weighted sled. Pull the sled with a force gauge on level ice at three measured temperatures. Do five pulls per sample at each temperature, in random order. Divide the pulling force needed to start the sled moving by the sled's weight. This number is the coefficient of static friction, your measure of grip. Use a fresh lane of ice each time. Record how rough the ice surface is and how worn the sole is. Keep all tests on a contained tray on a bench.

Before you start: Adult supervision: cold trays, sharp traction parts, and pulling loads that are held securely

Learn more: The slipperiness of ice explained, ScienceDaily (Max Planck Institute for Polymer Research)

EngineeringGrades 9–120800 Engineering

Which Tendon Material Lasts in a Robot Hand?

Many robot hands move their fingers with tendons, strong cords pulled by motors in the forearm, much like the tendons in a human hand. Tendons can stretch slowly over many cycles (creep) and wear where they rub on pulleys. Then the fingers lose grip strength and position accuracy, so engineers test tendon materials for durability.

Question: A tendon-driven robot hand uses cords, like tendons, to pull its fingers. Which low-cost tendon material keeps the hand gripping well after 500 grip cycles?

Why it stands out: The winning hand must keep its grip after hundreds of repeats.

How to do it

Use a school 3D printer to build three identical hands for each tendon material. Use a calibrated force sensor to measure grip force. Compare fishing line, kite string, and aramid thread (a very strong synthetic fiber, such as Kevlar), all at the same starting tension. Every 100 cycles, measure grip force, tendon stretch, and finger position error. Use the same motor motion and object each time. Then improve the pulley design. Compare it with the original using three new tendon sets.

Before you start: Adult supervision: printer use, wiring, and moving fingers

Learn more: Tendon-Driven Humanoid Hands, Humanoid.guide
Builds on: societyforscience.org

EngineeringGrades 6–120800 EngineeringAlaska

Reducing Water Slosh When Hauling Water

In some rural Alaska communities, homes have no running water, and people haul water from a central source and store it in containers. When a container moves over bumps, the water sloshes and can spill out. A baffle is a divider placed inside a tank to break up the water's motion and reduce sloshing.

Question: Which removable divider (baffle) best reduces spills from a model water container on rough paths?

Why it stands out: A tabletop bump shows how much hauled water a simple insert could save.

How to do it

Build three copies each of three baffle spacings, plus a container with no baffle. Give them all the same water mass and a vented lid. Pull each one on a guided cart over a fixed bump at a marked speed for 10 trials. Catch the water that escapes through the same small lid opening and weigh it. Film how far the container tilts at its peak, and check the cart's acceleration with a sensor. Then improve the best baffle and test it on a second bump. Describe this as a model of hauling water for a household.

Before you start: Adult supervision: a secured cart, contained water, and electronics kept dry

Learn more: In rural Alaska, access to water is preventive medicine, Alaska Native Tribal Health Consortium (sponsored, Anchorage Daily News)

EngineeringGrades 6–120800 Engineering

Designing a Storm Drain Guard That Resists Clogging

Storm drains carry rainwater off streets, and drain guards (grates) keep large debris out. Fallen leaves can clog the openings, so water backs up and floods streets and yards. The pattern of the openings may change how quickly a guard clogs.

Question: Which storm drain guard opening pattern lets the most water through before leaf litter blocks it?

Why it stands out: The winner has to keep draining when a new shape of leaf litter shows up.

How to do it

Build three guards with different opening patterns. Also build a plain grid guard to compare against. Give all of them the same total open area. Keep the water flow rate fixed. Add identical flexible plastic leaf shapes in measured batches. Do six trials per guard, resetting between trials. Measure water depth, overflow, and the leaf mass needed to reach a set water depth. Then improve the best guard and run six new trials with a second leaf shape.

Before you start: Adult supervision: cutting plastic, keeping water contained, and keeping pumps dry

Learn more: 'Only rain down the drain': keep storm drains clear of leaves, WHYY

Energy and transportation7 ideas · 4 Alaska

Categories: 0900 Energy & Transportation

ProgrammingGrades 9–120900 Energy & TransportationAlaska

Battery Storage in an Alaska Village Microgrid

Most rural Alaska villages are not connected to a large power grid. Each runs its own small grid, called a microgrid, usually powered by diesel generators, and the fuel is expensive to ship in by barge or plane. Adding solar panels and batteries can reduce diesel use, but each extra battery may save less than the one before.

Question: For a model of one Alaska village microgrid, at what point does adding more battery storage save too little diesel fuel to pay for itself?

Why it stands out: One graph shows the point where buying more battery storage stops paying for itself.

How to do it

Choose one community listed in the Power Cost Equalization (PCE) reports (state records of village electricity use and costs). Estimate its hour-by-hour electricity demand (load profile), and label it clearly as an estimate. Match it to the village's reported yearly electricity use. Use hourly solar data to compare three solar sizes and a range of battery sizes against running on diesel alone. Calculate diesel use, unmet demand, and yearly cost. State your assumptions for battery efficiency losses and each generator's minimum load (the lowest power it can safely run at). Run 30 scenarios with different weather and demand. Show where the savings level off. Also show how the answer changes with fuel price and with your hourly-demand estimates.

Learn more: What is a microgrid?, Alaska Center for Energy and Power, University of Alaska Fairbanks

EngineeringGrades 6–120900 Energy & TransportationAlaska

Vertical Bifacial Solar Panels on Snowy Ground

Bifacial solar panels produce electricity from both their front and back sides, so they can use light reflected off the ground. Vertical panels facing east and west catch more morning and afternoon sun than panels tilted toward the south. Researchers at the University of Alaska Fairbanks have tested vertical bifacial panels in Fairbanks, where snow and low sun angles make solar power harder to predict.

Question: Does a vertical bifacial (two-sided) solar panel gain more from snowy ground than the same panel tilted toward the south?

Why it stands out: The snow next to a panel may matter as much as the sun above it.

How to do it

Use three matched pairs of small bifacial solar panels. In each pair, mount one panel vertically facing east and west and tilt the other toward the south. Give each panel the same suitable electrical load. Log its power, its temperature, and any shading. Collect at least 20 full days over six weeks, with both snowy and bare ground if you can. Swap the panels between mounts each week. Compare daily energy per rated watt within each pair. Report ground cover separately from cloud cover and temperature. Don't claim snow caused differences between dates, since other conditions change from day to day too.

Before you start: Adult supervision: secure mounts at ground level, weather-rated wiring, and a winter field safety plan

Learn more: Study on solar PV model effectiveness prepares future in Alaska, UAF Alaska Center for Energy and Power

ScienceGrades 6–80900 Energy & TransportationAlaska

How Long Should a Block Heater Run?

An engine block heater is an electric heater that plugs into an outlet and keeps a car's engine, oil, and coolant warm on cold nights. A warmer engine starts more easily, wears less, and produces less pollution right after starting. In Anchorage, the city recommends plugging in for about two hours before driving when it is 20 °F or colder, often using a timer.

Question: On cold Alaska mornings, how much more does each extra hour of block heater use warm the engine?

Why it stands out: You find out whether the last hours of plugging in add much warmth at all.

How to do it

Work with an adult. Attach a temperature probe to the same safe spot on the outside of the engine. Use an outdoor-rated plug-in energy meter and a timer. Over 24 mornings, randomly choose heating times of two, four, or eight hours. Park the car for the same length of time each night. Compare how much warmer the probe is than the outdoor air, and how much electricity (kilowatt-hours) was used. Take the starting temperature and the overnight weather into account. Never run the engine while you measure. Report only surface warming, not fuel savings or whether the engine is ready to drive.

Before you start: Adult supervision: parked vehicle, household (mains) electricity, heater used only as the maker approves, and care in the cold

Learn more: Should you get an engine heater?, Alaska Public Media
Builds on: adn.com

ScienceGrades 6–120900 Energy & Transportation

Keeping Batteries Warm in Cold Weather

Batteries deliver less energy in the cold because their internal resistance rises and their usable capacity drops. In an Alaska winter, this can make devices shut off early. Insulating a battery, or adding a small heat source like a chemical hand warmer, may keep it warm enough to deliver more energy.

Question: Can an insulated pouch with a hand warmer get more usable energy out of a battery in cold weather?

Why it stands out: You test a fix for cold batteries and measure how much extra energy it really gives.

How to do it

Use six matched rechargeable nickel-metal hydride (NiMH) battery packs in a holder rated for the test temperature. Use a suitable fixed load and a logger that records voltage and current. Test four conditions in random order: room temperature, cold and uninsulated, cold and insulated, and cold and insulated with a hand warmer. Recharge only at room temperature. Stop each test at the maker's cutoff voltage (the lowest safe voltage). Log the pack temperature to stay within its rated limits. Compare the energy in watt-hours each pack delivers, and count the cost of the hand warmer separately.

Before you start: Adult supervision: follow battery ratings and temperature limits, protect wiring, and supervise hand-warmer use

Learn more: BU-502: Discharging at High and Low Temperatures, Battery University

ScienceGrades 6–120900 Energy & TransportationAlaska

Does Wet Moss Insulation Recover After Drying?

Insulation slows heat loss mostly by trapping air, and its resistance to heat flow is measured as R-value. Water conducts heat much better than air, so when water replaces the trapped air, wet insulation lets more heat escape. Natural materials such as moss have been used as insulation, and they can easily get damp.

Question: Does adding the same amount of water increase heat loss more in dried moss than in foam insulation?

Why it stands out: You pick up where a 2026 Nome winner stopped. Does natural insulation recover once it dries, or does the damage stay?

How to do it

Build three identical wall panels of dried store-bought moss. Build three more of rigid foam insulation, with the same thickness and area. Put each panel on a warm box and measure the power needed to keep a fixed temperature difference. Test the panels dry, after two measured amounts of water, and again after they dry out. Keep airflow and test time the same. Run each condition three times in random order. Report how much insulating value each material loses when wet and how much it regains after drying. Don't make claims about full sod buildings.

Before you start: Adult supervision: safe low-voltage heating, dry electrical connections, and careful material handling

Learn more: What Happens When Metal Building Insulation Becomes Wet?, National Insulation Association (Insulation Outlook)
Builds on: knom.org

EngineeringGrades 6–120900 Energy & Transportation

A Reusable Interior Window Insert Seal

Windows lose heat in two main ways: through the glass (conduction) and through air leaks around the frame. An interior window insert, sometimes called an interior storm window, is a removable panel that presses into the window frame. Its edge seal blocks air leaks, so the seal has to keep working after the insert is taken out and put back many times.

Question: Which edge seal on a removable interior window insert keeps heat loss low after 20 removals?

Why it stands out: A reusable winter window insert has to survive being taken out and put back.

How to do it

Build five copies each of three seal designs, plus a standard insert sealed with tape. Fit them on identical model windows. Use the same warm box, starting temperature difference, and insert thickness each time. Measure how fast the box cools before and after 20 cycles of removing and reinstalling the insert. Compare with a window that has no insert. Record any condensation and visible gaps, keeping humidity the same. Then improve the best seal and test five new copies. Keep the amount of heat stored in the box and inserts the same.

Before you start: Adult supervision: safe warm-water heat sources, cutting tools, and moisture near sensors

Learn more: Interior storm windows as a cost-effective window weatherization option, Center for Energy and Environment

EngineeringGrades 9–120900 Energy & Transportation

Sliding Ballast for a Wave Energy Converter

A wave energy converter is a device that turns the up-and-down or rocking motion of ocean waves into electricity. Many designs work best when waves arrive at one particular period (the time between waves), but real waves vary. Moving weight, called ballast, inside the device might help it respond well to a wider range of wave periods.

Question: Can sliding ballast (movable weight) help a model wave energy converter produce power across more wave periods?

Why it stands out: The generator has to keep working when the waves stop arriving on a regular schedule.

How to do it

Use a contained wave tank with a paddle that makes waves at three measured periods. The period is the time between waves. Compare three fixed ballast positions with a sliding ballast of the same total mass. Run six wave sequences per setting, and let the water settle between runs. Measure wave height, how far the ballast moves, and the electrical energy delivered to one resistor. Use ballast fixed in the center as your baseline. Measure the sliding friction. Then improve the track and test six new sequences of irregular waves.

Before you start: Adult supervision: contained water, moving parts, and low-voltage wiring

Builds on: snexplores.org

Physics and space8 ideas · 3 Alaska

Categories: 1400 Physics & Astronomy

EngineeringGrades 9–121400 Physics & AstronomyAlaska

Building a Magnetometer to Detect Geomagnetic Storms

A magnetometer measures the strength and direction of Earth's magnetic field. During geomagnetic storms, caused by activity on the Sun, the field changes more than usual, and the changes are especially strong at high latitudes like Alaska. NOAA's planetary K-index (Kp) uses magnetometer readings from observatories around the world to score the disturbance from 0 to 9 every three hours.

Question: Can a homemade magnetometer tell real local geomagnetic disturbances apart from sensor drift?

Why it stands out: Your instrument has to prove it sees space weather, not just its own temperature changes.

How to do it

Build a magnetometer using an RM3100 magnetic sensor chip, and add a temperature logger. Before field use, test its noise and how much its readings drift with temperature. Mount it securely away from cars, wiring, and loose metal. Record every 10 seconds for six to eight weeks. Compare hourly changes with a public Alaska reference magnetic observatory, such as the USGS College observatory near Fairbanks, and with NOAA's Kp index. Use quiet days to set a detection threshold, then test it on later days. If the reference data or your sensor is not good enough, say so. Do not treat Kp as the true local reading, because it is a worldwide average.

Before you start: Adult supervision: weatherproof low-voltage wiring, site access, and a winter field safety plan

Learn more: Planetary K-index (NOAA Space Weather Prediction Center)

ScienceGrades 6–121400 Physics & AstronomyAlaska

How Snow Cover Slows Ice Growth on Ponds

Lake and river ice grows thicker as cold air pulls heat out through the ice. Snow on top of the ice traps air and acts like a blanket, slowing that heat loss. A study of Alaska lakes and rivers found that snow is the main factor controlling how thick the ice gets by late winter.

Question: How much does a measured layer of snow slow ice growth in outdoor model ponds?

Why it stands out: The snow blanket on top of the ice becomes your variable in a safe model of Alaska winter lakes.

How to do it

Set up identical shallow tubs of water on dry land as model ponds. Use six tubs each with no snow, thin snow, and thick snow. Insulate the sides equally. Measure the starting water mass and ice thickness, and put temperature probes at fixed depths. For four cold weeks, measure snow depth and ice growth using fixed markers or by looking safely from the edge. Compare the groups with Stefan's equation, a simple ice-growth model based on freezing degree-days (a running total of how cold it has been). Treat each tub as a separate trial, and report effects from the tub walls. Never walk onto lake ice, and never use the results to judge whether ice is safe to travel on.

Before you start: Adult supervision: contained water, measurements from the tub edge, cold exposure, and a field safety plan

Learn more: Observation-derived ice growth curves for Alaskan lakes and rivers (The Cryosphere, 2020)
Builds on: knom.org

ScienceGrades 6–81400 Physics & AstronomyAlaska

Testing Wall Thickness in Model Snow Shelters

Snow is a good insulator because it traps lots of air, and air conducts heat poorly, much like a down jacket. That is why people in cold places can use snow to build warmer shelters. How well a snow wall slows heat loss depends on its thickness and on the snow's density (how tightly packed it is).

Question: Do thicker walls slow heat loss in small model snow shelters that have the same inside size?

Why it stands out: An Alaska winter shelter question gets a fair test with 15 separately built shelters.

How to do it

Build five tiny shelters at each of three wall thicknesses around the same removable form. Give them equal vents, and measure the snow density. Put an identical sealed bottle of 40 °C water inside each one. Log the bottle, inside-air, and outside temperatures for four hours. Compare cooling rates with a bottle that has no shelter, and place shelters in random spots. Keep shelters too small for anyone to go inside. Explain that the results measure heat loss in a model, not survival safety.

Before you start: Adult supervision: snow handling, warm-water filling, cold-exposure limits, and a field safety plan; no going inside shelters

Learn more: How snow insulates the ground (Alaska Science Forum, UAF Geophysical Institute)

ScienceGrades 6–81400 Physics & Astronomy

Does Fresh Snow Absorb More Sound Than Packed?

Fresh, fluffy snow is full of tiny air pockets that trap sound waves, so it absorbs sound much like the foam panels in a recording studio. Hard, smooth surfaces such as pavement or ice reflect sound instead. As snow packs down, hardens, or melts, it loses much of its quieting effect.

Question: Does loose snow reduce sound levels more than packed snow on the same test ground?

Why it stands out: The hush after snowfall becomes a test that separates snow texture from changing weather.

How to do it

Make four equal plots on one flat site: bare ground, loose snow, packed snow, and an icy crust. Use the same speaker and a microphone at a fixed height, at two marked distances. Play test tones at a safe volume in random order, with five recordings per surface on four calm days. Turn off the microphone's automatic gain control (automatic volume). Calibrate its levels against a borrowed sound level meter, and record wind and temperature. Compare changes in decibels with the bare plot. Call the result total sound reduction, not pure absorption.

Before you start: Adult supervision: icy plots, safe sound levels, and a winter field safety plan

Learn more: Why does everything get so quiet after a snowfall? (Montreal Science Centre)

ScienceGrades 9–121400 Physics & Astronomy

How Speed and Depth Change a Boat's Wake

A boat moving across water leaves a V-shaped wake called a Kelvin wake, first explained mathematically by Lord Kelvin. In deep water, theory predicts that each arm of the V sits about 19.47 degrees from the boat's path, no matter how fast the boat goes. At higher speeds the most visible waves can form a narrower V, and the simple theory assumes deep water.

Question: How do towing speed and water depth change the wake angle you can see behind a model boat?

Why it stands out: A familiar V-shaped wake shows how a textbook angle and the angle in a photo can measure different things.

How to do it

Tow one small model hull through a long tank of water at five measured speeds and two water depths. Do six runs per setting, waiting for the water to settle each time. Film from above with distance markers. Measure the actual speed and the angle of the brightest wave crests before waves reflect off the tank walls. Compare with a still-water reference and with the theoretical deep-water Kelvin angle of 19.47 degrees. Keep that theoretical edge separate from the brightest angle you see. Record tank width, depth, hull length, and wavelength. These show where the tank is too small for a fair test of the theory.

Before you start: Adult supervision: contained water, guarded towing parts, and secured overhead camera

Learn more: Kelvin wake pattern (Wikipedia)
Builds on: societyforscience.org

ScienceGrades 6–121400 Physics & Astronomy

How a Small Added Mass Changes String Harmonics

A vibrating string fixed at both ends forms standing waves, and each pattern has its own frequency, called a harmonic. Each harmonic has nodes, points that barely move, and antinodes, points that move the most. Physics models predict that a small added weight shifts a harmonic most when it sits near that harmonic's antinodes and least when it sits at a node.

Question: How does moving one small added mass along a string change its first four vibration frequencies (harmonics)?

Why it stands out: A mass placed at a node can leave one harmonic almost unchanged while shifting another.

How to do it

Stretch a string at low tension between two fixed points. Check its tension with a force gauge before each trial. Attach the same small bead and clip at five marked spots, keeping the added mass the same. Compare with a bare string and with a clip-only control. Record the sound with a microphone for six fresh tension setups at each spot. Use audio software that shows a frequency spectrum to find the first four frequency peaks. Compare the changes with a simple string model, including spots at nodes, where a vibration pattern barely moves.

Before you start: Adult supervision: anchored low-tension string and small secured masses

Learn more: The first 7 harmonics of standing waves on a vibrating string (UNLV Physics)

ScienceGrades 9–121400 Physics & Astronomy

Image Stacking Versus Long Exposures for Resolving Stars

Astrophotographers often take many short photos and combine them, called image stacking, instead of taking one long exposure. Random noise changes from frame to frame while the real signal stays the same, so stacking can make the signal stand out. Short exposures also blur less when the camera shakes, but every frame adds some of the camera's own noise.

Question: Can many short, aligned photos resolve two dim point lights better than one long exposure with the same total time?

Why it stands out: Two tabletop stars show when more photos help and when extra camera noise wins.

How to do it

Make two dim artificial 'stars' from LEDs behind pinholes, set at several small separations. Shake the camera at three repeatable levels, plus a still-camera reference. Take 10 image sets per setting. Each set has one long exposure and short exposures with the same total time, stacked both with and without alignment. Lock camera settings, and take dark frames (photos with the lens covered) to measure the camera's own noise. Decide what counts as 'two separate lights' (resolved) before you score images with coded file names. Report when stacking helps by cutting blur, and when it hurts through added camera read noise or poor alignment.

Before you start: Adult supervision: dim LEDs, secured camera rig, and low-voltage wiring

Learn more: ELI5: How Image Stacking Reduces Noise and Recovers Signal (PetaPixel)

EngineeringGrades 6–121400 Physics & Astronomy

Ranking Streetlight Blue Light with a DVD Spectrometer

A spectrometer splits light into its colors (wavelengths) so you can see which colors a lamp gives off. The closely spaced tracks on a DVD act as a diffraction grating, so a cardboard box and a DVD can make a simple classroom spectroscope. Different streetlights, such as sodium lamps and LEDs, give off different mixes of colors, including different amounts of blue light.

Question: Can a homemade DVD spectrometer rank local streetlights by how much blue light they give off?

Why it stands out: A cheap instrument has to measure its own bias before it can compare lights.

How to do it

Build a spectrometer from a piece of DVD, which acts as a diffraction grating, and give it a fixed camera mount. Calibrate wavelengths using the known mercury lines of a sealed fluorescent lamp. Lock exposure and white balance, and subtract a dark frame (a photo taken with no light). From the sidewalk, take three photos each of 20 or more streetlights of different types, such as high-pressure sodium, warm-white LED, and cool-white LED. First test the setup on lamps whose makers publish their spectra. Then rank the streetlights by a blue-to-total light ratio. Report that this ratio is specific to your instrument, because a camera sensor responds to some colors more strongly than others.

Before you start: Adult supervision: night work near roads with a field safety plan, sharp cut disc edges, and never opening fluorescent lamps

Learn more: A DVD Spectroscope: A Simple, High-Resolution Classroom Spectroscope (Journal of Chemical Education, hosted by NASA Langley)

Computer science and math13 ideas · 9 Alaska

Categories: 0600 Computer Science · 1100 Math

ProgrammingGrades 6–120600 Computer ScienceAlaska

Simplifying Wildfire Map Shapes for Slow Internet

Wildfire maps show each fire's edge, called a perimeter, as a shape made of many corner points (vertices). Line simplification is a computer method that removes some of those points so the file gets smaller and loads faster. Removing too many points makes the shape less accurate, which matters in rural Alaska towns with slow internet that need fire maps during fire season.

Question: How much can you simplify the fire perimeter shapes on an Alaska wildfire map before they become too inaccurate to be useful on a slow rural internet connection?

Why it stands out: You put a number on a real rural Alaska problem: how much map detail a slow connection can handle during a fire.

How to do it

Download 30 past fire perimeter shapes from AICC (the Alaska Interagency Coordination Center). Simplify each at 4 settings with a free tool such as mapshaper, which removes vertices (corner points). Measure file size and load time on a connection slowed down on purpose (throttled), with 5 runs each. Also measure the largest gap in meters between the simplified edge and the original. Decide before you start how much error is too much, for example 100 m. Choose a setting. Then check it on 10 new fire shapes you didn't use.

Learn more: Simplification guide, mapshaper documentation
Builds on: fire.ak.blm.gov

ProgrammingGrades 6–120600 Computer ScienceAlaska

Testing Aurora Alerts: Kp Index and Cloud Cover

The aurora, or northern lights, glows when charged particles from space strike gases high in Earth's atmosphere. NOAA's Space Weather Prediction Center uses the Kp index, a 0–9 scale of how disturbed Earth's magnetic field is, to decide when to issue geomagnetic storm alerts, and aurora watchers are among the people who rely on them. A high Kp still does not guarantee a sighting at your house, because clouds can block the sky.

Question: Aurora alerts use the Kp index (a 0–9 measure of geomagnetic activity). How often does a Kp alert correctly predict that you'll see the aurora from your house, and does adding cloud cover help?

Why it stands out: You build and honestly test an alert that Alaskans really check on winter nights.

How to do it

Write a script that saves NOAA's Kp forecast and a cloud-cover forecast every evening before dark. That way you only use what was known when the alert went out. Point a camera north from a fixed spot. Take sky photos every few minutes for 60 or more nights. Label each night 'aurora seen' or not. Build alert rules on the first 40 nights (training set): Kp only versus Kp plus clouds, at several Kp cutoffs. Then test them on the last 20 nights, which you held back (test set). Score hits, misses, and false alarms.

Learn more: Planetary K-index (NOAA Space Weather Prediction Center)

ProgrammingGrades 6–120600 Computer Science

Does an Image Classifier Learn Backgrounds Instead?

An image classifier is a computer program that learns to sort pictures into groups, such as 'mug' or 'shoe.' Sometimes it learns a shortcut instead of the object: if every mug photo was taken on the same table, it may learn the table. This problem, called shortcut learning, is a major reason AI that scores well in testing can fail in the real world.

Question: Does an image classifier (a program that sorts pictures into groups) really learn to recognize the objects? Or does it learn the backgrounds the objects were photographed on?

Why it stands out: With your own photo set, you show a big reason why AI that looks great in tests fails in the real world.

How to do it

Photograph 6 household objects, such as a mug, spoon, and shoe, on 6 different backgrounds. Take 20 photos of each object on each background. Train the same simple classifier two ways; Google's Teachable Machine works. In one, split the photos into training and test sets at random. In the other, leave whole backgrounds out of training. Repeat 5 times. Test both on a 7th background the model has never seen.

Learn more: Shortcuts: How Neural Networks Love to Cheat, The Gradient

ProgrammingGrades 9–121100 MathAlaska

Can Simple Models Beat Official Salmon Forecasts?

Before each fishing season, the Alaska Department of Fish and Game (ADF&G) publishes a preseason forecast of how many salmon will return. For Bristol Bay sockeye, the forecast is built from age classes (how many years fish spent in fresh water and in the ocean) and past run sizes. Forecasting is hard: since 2014, every Bristol Bay sockeye return has been larger than the state's preseason forecast.

Question: Can a simple model predict salmon returns as well as ADF&G's official preseason forecasts for one well-recorded Alaska salmon run?

Why it stands out: Your model gets only the information the official forecasters had before each season.

How to do it

Choose a run with at least 15 years of both preseason forecasts and final return counts (total run size). Check that both count the same fish, such as the same stock and the same parts of the run. Predict each later year using only data you would have had before that season (a hindcast test). Compare three forecasts: the official one, an average of past years, and an age-based sibling model (which predicts older fish from the number of younger fish from the same spawning year that returned the year before) if matching age data exist. Score each one on mean absolute error, bias (whether it guesses too high or too low), and percent error. Decide ahead of time how to handle very small returns. Report results year by year, including very large and very small runs.

Learn more: State biologists forecast the 2026 Bristol Bay sockeye run (Alaska Beacon)
Builds on: adfg.alaska.gov

ProgrammingGrades 9–121100 MathAlaska

Comparing Earthquake Size Patterns Across Alaska Regions

Small earthquakes happen much more often than large ones: each step down of one magnitude unit usually brings about 10 times as many quakes. On a graph of the logarithm of the count versus magnitude, this makes a nearly straight line, called the Gutenberg-Richter relationship, and its slope is called the b-value. Very small quakes are often missed, so scientists first find the magnitude of completeness, the smallest size a catalog records reliably.

Question: Does the b-value (the slope of the earthquake size pattern) differ between Alaska regions once you account for missed small quakes and the fitting method?

Why it stands out: Your fitting method has to pass a test with a known answer before it can explain Alaska earthquakes.

How to do it

Use the same time period of the Alaska Earthquake Center catalog for the Aleutians, Southcentral, and Interior Alaska. Each region needs at least 500 usable quakes. Find each region's magnitude of completeness. Then pick one safe cutoff that works for all three. Compare a straight-line fit to a histogram (a bar graph of counts) with the maximum-likelihood method, which uses each quake's magnitude directly. Test both methods on synthetic (computer-made) catalogs where you already know the b-value. Use 1,000 bootstrap resamples to show uncertainty. Repeat by blocks of time, and again with aftershock clusters removed (declustering). Catalog changes over time and related quakes can shift the answer.

Learn more: How Often Do Earthquakes Occur? (SERC, Carleton College)
Builds on: earthquake.alaska.edu

ProgrammingGrades 9–120600 Computer ScienceAlaska

Compressing Data from a Remote Weather Station

Data compression means storing or sending information in fewer bytes. Lossless compression lets you rebuild the original data exactly, while lossy methods throw some detail away. Remote Alaska weather stations often send data by satellite, which costs money and battery power for every byte, so smaller messages matter.

Question: Which data compression method sends the most weather information from a remote Alaska weather station for the fewest bytes?

Why it stands out: Remote stations pay for every byte sent by satellite, in money and battery. You find where the savings are.

How to do it

Get a month of 1-minute data from a home weather station or a public one. Compare four methods: gzip (a common general-purpose compressor), delta encoding (saving only the change from the last reading), delta encoding plus variable-length integers (small numbers take fewer bytes), and sending a reading only when the value changes. Run them on a microcontroller (a small computer chip) such as an ESP32. Measure file size, time to compress, and how accurately you can rebuild the original data. Check whether short events like a wind gust go missing. Test at 3 sampling rates.

Learn more: Time-Series Compression Algorithms, Explained, Tiger Data (Timescale) blog

ProgrammingGrades 9–121100 MathAlaska

Predicting Juneau Glacial Outburst Flood Peaks Early

Suicide Basin is a lake dammed by the Mendenhall Glacier near Juneau. Almost every year since 2011, it has suddenly drained under the glacier in a glacial outburst flood, raising Mendenhall Lake and the Mendenhall River. A July 2016 flood sent more than 16,000 cubic feet of water per second down the river and damaged homes, so an early estimate of a flood's peak helps people prepare.

Question: Can the first 12 hours of a Mendenhall Lake rise predict how high the lake will crest during a glacial outburst flood?

Why it stands out: Your model must predict a real Juneau flood using only what was known as the water rose.

How to do it

Use public USGS Mendenhall Lake water-level records for at least five well-recorded releases from Suicide Basin. Keep basin and lake measurements separate. Use one fixed rule to decide when each lake rise starts. Fit simple growth curves using only the first 12 hours. Test each model on a flood that was left out of fitting. Compare its errors in crest height and timing with an average of past floods. The crest is the flood's highest water level. Report how often the crest comes before 12 hours. Where you have enough floods, test using only earlier years. This is a research test, not a safety forecast; official flood warnings come from the National Weather Service.

Learn more: New study highlights findings and predictions for Juneau glacial outburst flood (USGS)

ProgrammingGrades 6–120600 Computer Science

Detecting a Dripping Faucet by Sound

A leaky faucet wastes water one drop at a time, and small household leaks add up over a year. The EPA says a faucet dripping once per second can waste more than 3,000 gallons a year. Acoustic leak detection means using a microphone and a computer program to recognize the sound of a leak, even when other household noise is present.

Question: Can a $40 Raspberry Pi (a tiny computer) with a microphone detect a dripping faucet in a noisy house?

Why it stands out: You make a cheaper version of a Stockholm Junior Water Prize finalist's leak-sound idea. Then you prove it works in rooms it has never heard.

How to do it

Record 3–4 measured drip rates plus a no-drip control (comparison) with TV, fan, or podcast noise in the background. Do at least 3 separate recording sessions per condition across 3 rooms. Compare three methods: a fixed loudness threshold, a threshold that adjusts to the background noise, and a small trained audio classifier (a program that sorts sounds). Train on some sessions. Test on whole sessions and rooms you kept out of training. Report correct detections on drip clips and false alarms on no-drip clips. At each drip rate, catch and measure the water for a timed interval. Only then convert the rate to liters wasted per day.

Learn more: Fix a Leak Week, U.S. EPA WaterSense
Builds on: news.cision.com

ProgrammingGrades 6–120600 Computer Science

Measuring Water Turbidity With a Smartphone Camera

Turbidity is how cloudy water is. It is caused by tiny particles such as clay, silt, and algae, and it is measured by shining light into the water and recording how much the particles scatter it. High turbidity blocks sunlight in the water and can harm habitat for fish and other aquatic life.

Question: Turbidity is how cloudy water is. Can a phone photo of a marked tube measure it as accurately as a real turbidity meter?

Why it stands out: You bring a phone tool, in the style of the Stockholm Junior Water Prize, to a measurement almost every water project needs.

How to do it

Make 20 turbidity standards (samples of known cloudiness) by adding clay to water in steps. Measure each with a borrowed turbidity meter; ask a school, utility, or watershed group. Photograph each in a clear tube over a printed black-and-white target, under 3 lighting setups. Write code that measures how much the target fades. Fit a calibration curve (a line that turns fading into a turbidity reading). Then test it on 20 new samples you mix separately and on real creek water. Compare your results with the meter.

Before you start: Adult supervision: if you collect creek water, use the ISEF Field Work Safety Plan and take samples from a safe bank without wading

Learn more: Turbidity and Water (USGS Water Science School)
Builds on: newea.org

ProgrammingGrades 9–121100 MathAlaska

Forecasting Geomagnetic Storms with the Kp Index

Activity on the Sun can disturb Earth's magnetic field, and strong disturbances are called geomagnetic storms. NOAA rates the disturbance with the planetary K-index (Kp), a 0–9 score for each three-hour period, averaged from magnetometer readings at observatories around the world. A Kp of 5 or more counts as a geomagnetic storm.

Question: Does the current Kp value help predict whether the next three-hour period will have high Kp?

Why it stands out: The forecast has to give honest odds, including times when a storm dies down instead of continuing.

How to do it

Use eight full years of NOAA Kp data. Decide what counts as high Kp before you start. Build two models on the first five years. One is a climatology model that uses a fixed chance (how often high Kp happened overall). The other is a persistence model that uses the previous three-hour Kp. Test both on the three later years. Compare their probability errors (Brier scores) and their reliability (how often the predicted chances match what happened). Estimate uncertainty by resampling whole months. Try shorter training windows to check for changes over the 11-year solar cycle. Don't treat high Kp as a promise of visible aurora in Alaska.

Learn more: Planetary K-index (NOAA Space Weather Prediction Center)

ProgrammingGrades 9–120600 Computer ScienceAlaska

Machine Learning to Detect Volcanic Plumes on Webcams

A volcanic plume is a cloud of gas and ash that rises from an erupting or restless volcano. Airborne volcanic ash is a serious hazard to aircraft even hundreds of miles from an eruption, because it can reduce visibility, damage flight controls, and cause jet engines to fail. The Alaska Volcano Observatory, which helped write the linked fact sheet, watches Alaska's volcanoes with webcams, satellites, and other sensors.

Question: Can a computer program detect visible volcanic plumes (clouds of gas and ash) on Alaska Volcano Observatory webcams, and does letting it answer 'can't see' reduce false alarms?

Why it stands out: You teach a computer when to say 'I don't know.' That is a key skill for any warning tool.

How to do it

Download at least 1,000 archived webcam images from the Alaska Volcano Observatory. Label each one: visible plume, clear, cloudy, or uncertain. Have a second person check 200 of your labels. Train a small image classifier (a program that sorts pictures) using transfer learning, which starts from a model already trained on other photos. For the final test, keep whole dates and volcanoes out of training. Compare the same detector with and without a view-quality filter that skips poor views. Report false alarms, missed visible plumes, and the share of test images skipped. If the filter rejects a frame you labeled as a plume, count it as a missed plume. Keep uncertain labels separate. Describe your results as detecting visible plumes, not as proof that a plume contains ash.

Learn more: Airborne Volcanic Ash—A Global Threat to Aviation (U.S. Geological Survey Fact Sheet 2010-3116)
Builds on: avo.alaska.edu

ProgrammingGrades 9–120600 Computer ScienceAlaska

Predicting Rapid Wildfire Growth in Alaska

Alaska wildfires often burn more than a million acres in a season. Fires start and spread in dry, warm, and windy weather with a spark, and lightning is a major cause: in late June 2025, over 80,000 lightning strikes set off 162 new fires in two weeks. Predicting which fires will grow fast helps fire managers decide where to send crews.

Question: Using past Alaska fire records and weather data, can a simple model predict which wildfires will double in size by the next day?

Why it stands out: You apply a national-winning wildfire AI idea to Alaska's own fire seasons, and you test it honestly.

How to do it

Check records from AICC (the Alaska Interagency Coordination Center) for fires with daily size reports across at least 3 seasons. Aim for far more than 100 fire-days, with enough examples of fires that doubled and fires that did not. For each day, save the fire size and weather known by a fixed evening cutoff. Then label whether the size doubled by the next evening. Train two models: logistic regression (a simple probability model) and a decision tree (a branching yes-or-no model). Compare them with two baselines (simple guesses): 'never doubles' and 'repeat yesterday's growth result.' Test on a whole season you kept out of training. Report correct doubling alerts (hits), missed doublings, and false alarms separately, plus the number of doubling cases. If there are too few, report that the data can't support a reliable test.

Learn more: Alaska's million-acre wildfire season, typical in size, had some unusual features (Alaska Beacon)
Builds on: fire.ak.blm.gov

ProgrammingGrades 6–81100 Math

One Shared Lunch Line or Separate Lines?

Queueing theory is the math of waiting lines. With separate lines, one server can sit idle while another line backs up; a single shared line that feeds all servers avoids this. In practice the difference can be small and depends on details, so computer simulations are used to test real situations.

Question: Would one shared lunch line feeding all serving stations cut long waits compared with separate lines at the same stations?

Why it stands out: You check whether a faster average hides students stuck in very long waits.

How to do it

First get IRB approval (from an Institutional Review Board, which checks studies involving people). Then record anonymous arrival and service times at 10 school lunches. Use no names or photos, and don't change the line. Build a computer simulation of the line from six lunches. Check how well it predicts wait times at the other four. Compare shared and separate lines over 1,000 simulated lunches with the same staff and arrival pattern. Measure both average waits and the longest waits. Test your assumptions about students switching lines and servers working at different speeds. Don't recruit anyone for surveys.

Before you start: Needs IRB approval

Learn more: One Queue or Two (Allen Downey, Probably Overthinking It)

Check the rules before you start. Projects with people (even surveys), animals with backbones, microbes or hazardous chemicals need approval before you begin. Look for the “Before you start” note on each idea, read the fair rules, and ask your teacher. Questions? Email [email protected].

“Learn more” links explain the topic. “Builds on” links point to the past winner or public data an idea grew from.

Your good-project checklist

Tick each one you can honestly say yes to.

Ready for more? See Middle School Information, Senior High Information and The Road to ISEF.

Sources and credits

All project examples belong to the students named. We describe them here to teach, and we link to the articles where we found them.