Validating First-Bloom Reports With Satellite Data

ISEF Category: Plant Sciences

Subcategory: Other  ·  Difficulty: Advanced  ·  Setup: Home Setup  ·  Time: Full Year

The Hook

A bloom date on a phone screen can turn into a statewide data point. That sounds simple, but weather, location, and human timing all bend the result. You can test how close crowd-sourced first-bloom reports come to satellite-based green-up signals. That makes your project useful for ecology, climate tracking, and citizen science.

What Is It?

Phenology is the study of seasonal life-cycle events, like first bloom, leaf-out, and fruiting. Think of it like nature’s calendar. Plants do not change on the same day everywhere, so researchers track when a species crosses a seasonal threshold in different places.

Your project compares two ways of marking spring. One comes from people reporting a first bloom through iNaturalist. The other comes from Sentinel-2 satellite data, where NDVI, or normalized difference vegetation index, estimates how green and active plants are. If the satellite and the human reports line up closely, that supports the crowd-sourced method. If they do not, you can ask why.

This is a validation project. Validation means checking whether one measurement method matches a trusted reference or a second independent signal. You are not just collecting pretty observations. You are testing how reliable citizen science can be across a whole state.

Why This Is a Good Topic

This is a strong science fair topic because you can measure real agreement between two data sources. You get a clear question, a clear number to compare, and a real ecological meaning behind the result. The project also connects to climate change, invasive species tracking, and conservation planning. You can realistically do the data collection, mapping, and statistics with free tools and public datasets.

Research Questions

• How does the timing of iNaturalist first-bloom reports compare with Sentinel-2 NDVI green-up dates across counties in one state?
• What is the effect of distance from weather stations on the gap between citizen-reported bloom dates and satellite-derived green-up dates?
• Does report density on iNaturalist change the agreement between first-bloom dates and NDVI thresholds?
• To what extent do different plant species show different offsets between bloom reports and satellite green-up signals?
• Which land-cover type, urban, suburban, or rural, shows the smallest difference between first-bloom reports and NDVI timing?
• How does elevation affect the difference between first-bloom reports and satellite-based green-up dates?

Basic Materials

• Laptop or desktop computer with internet access.
• Spreadsheet software such as Google Sheets or LibreOffice Calc.
• Free account for iNaturalist.
• Sentinel-2 imagery access through Copernicus Browser or a similar public viewer.
• State boundary map or county shapefile from a government source.
• Basic mapping tool such as QGIS.
• Weather and climate reference data from NOAA or the National Weather Service.
• Notebook for tracking search terms, filters, and data-cleaning decisions.

Advanced Materials

• Laptop or desktop computer with internet access.
• QGIS for geospatial mapping and overlay analysis.
• Python with pandas, geopandas, rasterio, and scipy for analysis.
• Google Earth Engine or another public remote-sensing platform for NDVI extraction.
• Sentinel-2 surface reflectance imagery.
• County or land-cover shapefiles from USGS or a state GIS portal.
• iNaturalist observation exports.
• NOAA climate normals or station data for comparison.
• Digital elevation model data from USGS.
• Statistical software such as R or Python notebooks for regression and correlation analysis.

Software & Tools

• QGIS: Maps observation points, county boundaries, land cover, and satellite layers for spatial comparison.
• Google Earth Engine: Extracts NDVI timelines and helps find green-up dates from Sentinel-2 imagery.
• Python: Cleans observation data, calculates date offsets, and runs correlation or regression tests.
• R: Fits statistical models and makes publication-style plots for agreement analysis.
• ImageJ: Helps if you need to inspect visual vegetation change from exported image panels.

Experiment Steps

1. Define one plant species, one bloom event, and one state so your comparison stays manageable.
2. Decide how you will turn satellite NDVI into a single green-up date that can match each report location.
3. Build a clean dataset by matching iNaturalist observations to satellite pixels, counties, or buffer zones.
4. Set controls that account for cloud cover, land cover, elevation, and report density.
5. Choose your agreement metric before you start, such as date difference, correlation, or classification accuracy.
6. Plan one validation check that tests whether your result holds across more than one year or one species.

Common Pitfalls

• Mixing first-bloom reports from different species, which makes the satellite comparison meaningless.
• Using raw NDVI without checking cloud contamination or image quality flags, which can shift the green-up date.
• Comparing point observations to broad satellite pixels without accounting for mixed land cover, which blurs the signal.
• Ignoring observation bias in iNaturalist, which can make popular places look more accurate than remote ones.
• Defining green-up and bloom with different rules for different sites, which breaks consistency across the state.

What Makes This Competitive

A stronger project goes beyond a simple date comparison. You can test multiple species, multiple counties, or multiple years, then ask which conditions improve agreement and which ones do not. Good control of spatial mismatch, weather noise, and observation bias will matter a lot. Strong statistics, clear maps, and a thoughtful validation method can turn this into a serious ecological analysis.

Project Variations

• Compare iNaturalist bloom reports with Sentinel-2 timing for one tree species across urban and rural counties.
• Replace county-level analysis with elevation bands to test whether mountain sites show bigger timing offsets.
• Use another phenology signal, such as leaf-out or peak green-up, and compare it with the same citizen-science reports.
• Test whether report accuracy changes before and after major rain events or warm spells.

Learn More

• USGS Land Cover data: Find land-cover layers and mapping resources through the USGS website and related data portals.
• NOAA Climate Data Online: Search for station records and climate normals to compare weather timing with bloom dates.
• NASA Earthdata: Browse remote-sensing tutorials and Sentinel-related resources through NASA Earthdata.
• Copernicus Sentinel-2 documentation: Read official mission and vegetation index guidance from the Copernicus browser and Sentinel documentation.
• iNaturalist Research-grade observations: Use the iNaturalist help pages and data export tools to learn how observation records are structured.
• PubMed: Search for review articles on phenology, citizen science validation, and remote sensing agreement studies.