One-time snapshot · 31 Oct 2026 · not updated. This was FarmBot's first look at the fields, made before the exact field edges were marked. It used a rough box around the village and square boxes around each plot. We keep it for reference and will not update it. New to satellite words? Read What the satellite sees. Current site · Pond fields (weekly) · Temple plot (weekly)
FarmBot

Methods

Methods & data sources

Where every number comes from: what the satellite saw, and what a computer estimated.

Sources

Data sources

SourceUsed forType
Sentinel-2 (Europe's free Copernicus photo satellites, about one photo every 5 days; L2A = photos already corrected for haze), via Microsoft Planetary Computer10 m colour and infrared photos, photo tiles 44QQF/44QQG. Formulas: greenness score (NDVI) = (B08−B04)/(B08+B04); water score (NDWI) = (B03−B08)/(B03+B08); second water score (MNDWI) = (B03−B11)/(B03+B11). Clear squares for greenness = squares the satellite's own sorter labels plants, bare soil or water; a fixed −1000 correction is applied to newer photos (processing baseline ≥ 4).Observed
ESA WorldCover 2021 v20010 m land-cover map (what covers each square). Only crop land (class 40) feeds the plot crop series, with persistent water and trees excluded.Classified (2021)
ISRIC SoilGrids (a global computer soil map) 2.0Global computer soil map, 250 m squares: texture, pH, organic carbon, nitrogen, nutrient-holding capacity (CEC) and how packed the soil is (bulk density), read from ISRIC's public files because its online service was down.Estimate
NASA POWER (a free NASA weather estimate) dailyRain, temperature, humidity, solar radiation and wetness around the roots, ~50 km squares.Estimate
Open-Meteo (a free weather service) historical weather APIComputer-reconstructed (reanalysis) rain, ET0 (water that sun and wind pull from a crop) and soil moisture.Estimate
OpenStreetMapRoads, river and village points used to check locations.Mapped
Census of India 2011 and LGD (Local Government Directory), via OdiapediaPopulation, households, area and land use for Nilabadi.Official (2011)

How

Method

  • Plots. Square boxes around two landmarks: 600 m around the fishing pond and 400 m north-north-east of the Agnigangamma Temple. They are not surveyed field boundaries.
  • Crop series. For each clear Sentinel-2 (European satellites that photograph the fields about every 5 days) date (1 Oct 2025 to 23 Oct 2026), FarmBot takes the median (middle value) greenness of crop-land squares in the box. A 20-day running median smooths it, and the peak, the drops below 0.7 and 0.5, the dry-season low and the point halfway to full green (50%) are read off that curve.
  • Pond water. Squares with water > 0 or second water > 0 within 150 m of the pond centre, converted to hectares (1 hectare ≈ 2.5 acres). For this metric only cloud, shadow and snow are removed, because the satellite's own sorter mislabels this pond as soil.
  • Locations. The pond was found by matching a map screenshot to Sentinel-2 water. The temple comes from a public place listing (map plus code X768+5R7) checked against imagery and OpenStreetMap.
  • Refresh. The site is generated from the data folders by a build script, and can be rebuilt weekly after new photos are downloaded.
    • OpenStreetMap roads and river over Sentinel-2
    How to read this: the lines are roads and the river from the free OpenStreetMap. They line up with the photo, which shows the location is right.OpenStreetMap roads and river over Sentinel-2 (21 Oct 2026) around the temple, used to check the location. © OpenStreetMap contributors.
  • Insights (experimental). Crop comparison: the median of each plot's crop-land greenness (NDVI) over clear dates (photos at least 60% cloud-free) in the shared 1–23 October window, 2025 vs 2026. Harvest heads-up: the weekly change in the 20-day smoothed greenness. Weak patches: each square's greenness minus the plot median, as a median over clear dates in each October. The cut is the bottom 20% in both seasons, after removing a 20 m edge strip next to land that is not crop land. Winter-crop budget: Nov 2025 – Feb 2026 Open-Meteo crop water need (ET0) minus rain, against the November median pond area (MNDWI) per metre of depth.
  • Pest rules. These are taken unchanged from the cited ICAR-NCIPM (India's national centre for pest management), ICAR-NRRI (National Rice Research Institute) and IRRI (International Rice Research Institute) documents. They are not tuned or scored for these plots yet.

Pest & disease sources

  1. ICAR-NCIPM (NICRA), Web enabled weather based prediction for insect pests of rice
  2. ICAR-NRRI (CRRI) Technology Bulletin 277, Rice blast disease and its integrated management
  3. ICAR-NRRI Agro-Advisory Service, first fortnight of September 2026
  4. ICAR-NRRI Agro-Advisory Service, October 2024 (district advisory incl. Koraput)
  5. Samir et al., J. Biopesticides 17(2):135–143 (2024), pest management in Koraput
  6. ICAR-IIRR, Demonstration of IPM practices in farmers' fields of Odisha (incl. Koraput)
  7. IRRI Rice Knowledge Bank, Sheath blight
  8. IRRI Rice Knowledge Bank, Bacterial blight
  9. IMD Agromet Advisory Services, district advisories

Read with care

  • Plot boxes are approximate. They are squares drawn around landmarks, not surveyed field boundaries.
  • Monsoon cloud gap. July–August had only one or two clear satellite photos, so transplanting dates can't be pinned down.
  • Temple plot location is still to be confirmed. The reference screenshot was rotated, so the box sits north-north-east of the temple; a true-west alternative was also checked.
  • Soil and weather figures are computer-estimated estimates, not measurements on these fields.