How to monitor cropland with satellite data
When agricultural teams ask how cultivation changed, start with two products that answer different questions. A classified crop map estimates which category a pixel belongs to. A vegetation index such as NDVI shows how strongly vegetation reflects in selected wavelengths on a given date. Together they help identify areas worth checking.
Use a crop map for class areas
Crop Maps Ukraine is a starting point for regional comparisons in Ukraine. Choose the exact two published product years, verify the legend for each, and read the raster values using the publisher’s service or download. Keep the source grid and class definitions with the export.
Clip each raster to the same administrative boundary version. Count pixels by class and convert to area using a method that accounts for pixel size. If you align categorical rasters to compare transitions, use nearest-neighbor resampling rather than interpolating class codes.
Use NDVI to inspect vegetation timing
NDVI is (near infrared − red) ÷ (near infrared + red). The Copernicus Sentinel-2 documentation lists the spectral bands. A common Sentinel-2 calculation uses B8 and B4. Mask clouds and shadows before creating a comparable time series.
Compare dates at similar stages of the growing season. A single low reading could come from harvest, bare soil, clouds or other conditions. Inspect multiple dates and record what was masked before interpreting a change.
Deliver a map and a table someone can audit
- Map both years with consistent colors where the publisher’s classes are comparable. Show masked and unknown pixels separately.
- Tabulate area by region, class and year. Include the total mapped area and valid coverage for each year.
- Export the clipped classified rasters and a source manifest with product versions, date windows, boundary version, pixel size and processing steps.
Frequently asked questions
Can NDVI alone identify a crop species?
NDVI measures a vegetation signal. Crop classification generally needs more evidence, such as multiple observation dates, additional spectral bands and training or validation data.
Can I call a decrease in mapped cultivation abandonment?
Report the change in the mapped class first. Confirm abandonment with independent field observations, records or additional imagery interpreted with local knowledge.
More guides on site selection and land
Start with a region and review the proposed crop products before analysis.
Start a cropland monitoring question