How to Do Real Plant Sciences Research at Home: A High School Student’s Guide to Free Tools, Affordable Kits, and Public Databases

Plant science used to live behind the doors of university greenhouses and seed-company R&D labs. Today, a windowsill, a smartphone, and a free Google account can run the same kinds of experiments.

This guide is your starting point. It covers three things: the affordable kit you can assemble for the price of a textbook, the free software that professional botanists actually use, and the public databases that hold decades of satellite, genomic, and trait data waiting for a new question.

Why this is possible now

Three shifts in the last decade quietly opened plant research to high school students.

First, public satellite data went free and friendly. NASA's Landsat archive, ESA's Sentinel-2 imagery, and MODIS products are all queryable through Google Earth Engine on a free account. You can watch a forest green up, a drought spread, or a crop region stress out, all from a Chromebook.

Second, plant genomes and trait databases became openly downloadable. The 1001 Genomes project for Arabidopsis, Phytozome, Ensembl Plants, the TRY trait database, GBIF occurrence records, and the AlphaFold protein structure database now sit one click away. Real comparative genomics is a Python notebook away.

Third, sensors got cheap and image analysis got smart. A $30 USB microscope, a $15 pH/EC pen, a $5 capacitive soil-moisture sensor, and a phone camera now collect data that would have required a $20,000 plant phenotyping rig in 2010. Open tools like PlantCV, ImageJ, and ilastik turn those images into numbers.

Put it together: a kitchen counter with a Kratky hydroponic jar, a phone on a tripod, and a laptop running PlantCV can produce the same kind of phenotyping dataset that fills published papers.

The plant sciences home kit

Group your kit by what each item lets you do. None of these depend on a school lab.

Plants and seeds (your model organisms)

• Arabidopsis thaliana seeds, the lab workhorse, about $10 per packet.
• Wisconsin Fast Plants (Brassica rapa) for full life cycles in 35 days.
• Lemna minor (duckweed), free from any pond, for rapid bioassays.
• Radish, lettuce, basil, mung bean, oats for fast germination experiments.
• Tomato seedlings, peppers, and legumes for longer agronomy projects.
• Moss and succulents for low-maintenance ecology and physiology work.

Growing setups

• Kratky passive hydroponic jars (mason jar plus net cup, under $5 each).
• Cheap grow LEDs, including narrow-band 450/525/660 nm sets for action-spectrum work.
• Seedling trays, peat pellets, and clear plastic domes from any garden center.
• A windowsill with consistent light or a small shelf with a timer.

Sensors and electronics

• Arduino Uno or an ESP32 board, around $10 to $20.
• DHT22 temperature and humidity sensor.
• Capacitive soil-moisture sensors (avoid the corroding resistive kind).
• BH1750 lux sensor for quantifying light.
• pH and EC pen, about $15, essential for any hydroponic or soil study.
• A smartphone camera plus a small tripod.

Imaging and microscopy

• USB microscope, around $30, for stomata, pollen, and root hairs.
• Foldscope, the paper origami microscope, for fieldwork.
• Flatbed scanner for high-resolution root and leaf scans.
• Optional: a clip-on thermal camera attachment for leaf-temperature work.

Wet-lab consumables you already own

• Vinegar, baking soda, hydrogen peroxide, table salt, sugar, milk.
• Eggshells, coffee grounds, banana peels, wood ash for amendment studies.
• Clear nail polish for stomatal impressions.
• Zip bags, mason jars, and a kitchen scale that reads to 0.1 g.

A complete starter kit sits in the $80 to $200 range, depending on how many sensors you add.

Signature technique: smartphone time-lapse plus PlantCV

If one workflow opens the most projects in plant science, it is automated image-based phenotyping. You shoot a daily time-lapse of your plants, then let PlantCV measure traits that used to require hours of manual ruler work. Here is the 5-step version.

1. Set up a fixed imaging station. Tape your phone to a tripod aimed at a flat board with your plants. Add a color-checker card and a ruler in the frame. Same angle, same distance, every day.
2. Automate capture. Use your phone's built-in interval timer or a free time-lapse app. Daily shots at the same hour control for circadian effects.
3. Install PlantCV in Google Colab. No local install needed. PlantCV is a free Python library built on OpenCV, specifically for plant images.
4. Build a segmentation pipeline. PlantCV walks you through masking the plant from the background, then extracts leaf area, perimeter, color channels, and shape descriptors automatically.
5. Export a clean dataset. You end up with a CSV of traits per plant per day. Plot growth curves in Python or R, fit logistic or Gompertz models, and run statistical comparisons across your treatment groups.

The same pipeline handles germination assays, drought trials, color-change studies, and disease progression.

The dry-lab side: free software you can install today

Plant research now has a full open-source software stack. Install these on your laptop or run them in Colab.

Image analysis and phenotyping

• PlantCV: Python library for high-throughput plant image analysis.
• ImageJ / Fiji: the standard for measuring leaves, roots, stomata, and lesion areas.
• RhizoVision Explorer: free root-architecture quantification from flatbed scans.
• RootPainter: deep-learning root segmentation you train with a few clicks.
• ilastik: interactive machine-learning segmentation for any plant image.

Geospatial and remote sensing

• Google Earth Engine: cloud platform with petabytes of free satellite imagery.
• QGIS: full-featured open-source GIS for mapping and spatial analysis.

Genomics and bioinformatics

• BioPython: read, write, and analyze sequences in Python.
• MEGA: free phylogenetics with a friendly GUI.
• psRNATarget, CRISPOR: web tools for miRNA targets and CRISPR guide design.
• AlphaFold (via Colab notebooks): protein structure prediction on free GPUs.

Modeling, statistics, and machine learning

• Python with scikit-learn and PyTorch for ML models on plant data.
• R with phytools, ape, and tidyverse for phylogenetics and stats.
• AquaCrop-OS: open-source crop water-productivity model from the FAO.
• MaxEnt: species distribution modeling on climate and occurrence data.

Running the same tools that fill published methods sections changes how research feels. You are not simulating science. You are doing it.

Public databases that count as real data

Re-analysis of public data is itself a legitimate research path. Many ISEF-level projects never touch a pipette.

Satellite and climate

• NASA MODIS, Landsat, Sentinel-2 (through Google Earth Engine): vegetation indices, surface temperature, soil moisture.
• WorldClim and CHELSA: high-resolution global climate layers.
• NOAA and MesoWest: weather station data for hyperlocal modeling.

Agriculture

• USDA NASS: U.S. county-level yields, planted acreage, and prices.
• FAOSTAT: global agricultural statistics.

Biodiversity and occurrence

• GBIF: hundreds of millions of plant occurrence records worldwide.
• iNaturalist: citizen-science observations with photos and locations.
• PlantNet: image-based plant identification with a public API.

Genomes, transcriptomes, and traits

• NCBI GenBank and GEO: sequences and gene-expression datasets.
• Ensembl Plants and Phytozome: curated genomes for crops and model species.
• 1001 Genomes: natural variation in Arabidopsis thaliana.
• AraGWAS: genome-wide association results for Arabidopsis traits.
• TRY Plant Trait Database: millions of trait records across species.
• KEW Seed Information Database: seed mass, dormancy, and longevity data.
• AlphaFold DB: predicted 3D structures for almost every known plant protein.

A serious project can be built entirely on these, with the only "lab" being a Colab notebook.

How to combine wet and dry: the strongest project shape

The strongest plant projects mix hands-on measurement with computational analysis. Two patterns work especially well.

Pattern A: home experiment plus image-based modeling. You run a controlled growth experiment with two or three treatments on your windowsill. You capture a daily time-lapse, extract traits with PlantCV, then fit a growth model or train a small classifier in Python. The hands-on side gives you ownership of the data; the computational side gives you statistical power.

Pattern B: public-data model plus a local ground-truth check. You build a model from satellite, GBIF, or trait data, then validate one prediction with a small field or windowsill study. For example, predict urban-edge phenology from Sentinel-2, then walk a transect with iNaturalist to confirm.

Judges respond strongly to this shape because it shows both real-world data collection and modern data analysis, the two halves of a working plant scientist.

Choosing a phenomenon that has not been done

Originality is a process, not a guess. Run this 3-step check before you commit to a question.

1. Google Scholar. Search your candidate question in three different phrasings. Read the abstracts of the top 10 hits and the most recent 5 hits.
2. Society for Science abstracts archive. Search prior ISEF and Regeneron abstracts for your keywords. This catches projects that never became journal papers.
3. PubMed and Google Scholar review articles. Filter for review papers in the last 5 years. Reviews tell you where the field thinks the open questions are.

If you find adjacent work, that is good news, not bad. It means the topic is alive, and your job is to identify the specific twist (a new species, a new variable, a new method, a new geography) that has not been tried.

A realistic timeline

• One to two weeks. A focused replication or a single measurement campaign. Good for first-time researchers, regional science fair entries, or pilot data.
• One to two months. A full hybrid project with a real experiment plus a computational analysis. Strong fit for regional and state fairs.
• A full year. Multi-generation, multi-season, or large-dataset projects suitable for ISEF-track competition.

If this is your first project, start with the 1-to-2-week version and let the questions you bump into shape the longer study.

A starter checklist

Before you pick a specific phenomenon, get these in place.

1. A clean, consistent workspace with stable lighting and a power outlet.
2. A free Google account for Colab and Earth Engine, with Earth Engine sign-up submitted (approval is fast for students).
3. A local Python environment (Anaconda or Miniconda) with NumPy, pandas, scikit-learn, BioPython, and Jupyter installed.
4. PlantCV installed in Colab, ImageJ installed locally, and QGIS installed if your project is spatial.
5. Your sensor and imaging kit assembled, calibrated, and tested on a throwaway plant.
6. A paper or digital lab notebook with dated entries, ready for daily logging.
7. A single written sentence that states your research question in plain language.

Once these are checked off, you are ready to pick a phenomenon and start collecting data.

Where to go next

Plant Sciences at ISEF splits into seven subcategories. Each one has its own MehtA+ project guide that builds directly on the kit on this page. Pick the subcategory that pulls at you most.

• Agriculture and Agronomy (AGR): crop yield, soil amendments, irrigation, fertility, and food production.
• Ecology (ECO): plant communities, biodiversity, invasive species, urban habitats, and ecosystem processes.
• Genetics and Breeding (GEN): Mendelian inheritance, comparative genomics, GWAS, and selection experiments.
• Growth and Development (DEV): germination, tropisms, hormones, dormancy, and seedling physiology.
• Pathology (PAT): plant diseases, host-pathogen interactions, disease forecasting, and biocontrol.
• Plant Physiology (PHY): photosynthesis, stomata, transpiration, stress response, and metabolism.
• Systematics and Evolution (SYS): phylogenetics, morphometrics, biogeography, and trait evolution.
• Other (OTH): cross-cutting projects in phenotyping rigs, ML pipelines, citizen science, and meta-studies.

A windowsill, a phone, and a laptop are enough to run a real plant science project. The only thing left is to choose the question.

Project ideas in this category (87)