How Satellite Data Tells Indian Farmers Exactly When to Sow

India's kharif season has always been a negotiation with uncertainty. A farmer in Vidarbha or the Deccan plateau watches the sky, checks the soil between his fingers, asks neighbours what they heard on the radio. The southwest monsoon arrives differently every year, early in one district, late in the next, stalled for two weeks over the Western Ghats while the plains wait. Sow too early into dry soil and the seed rots. Wait too long after the first rains and the growing window closes before the crop matures. The margin for error is measured in weeks, sometimes days.

Traditional knowledge handles this well on average. It fails on the outliers, and in Indian agriculture the outliers, a drought year, a delayed onset, an early withdrawal, are frequent enough to be the rule rather than the exception. The India Meteorological Department has issued district-level forecasts for decades, but a district can span hundreds of kilometres of varying terrain. The forecast that covers Aurangabad city does not capture what is happening in the black cotton soil belt forty kilometres east of it.





ISRO's earth observation programme now operates a constellation that includes the Resourcesat series, Cartosat satellites, and the RISAT radar satellites, which can see through cloud cover, useful precisely because the monsoon arrives under heavy cloud. The European Space Agency's Sentinel-1 radar data is also freely available and widely used by Indian agricultural research institutions.


These satellites measure several things a farmer needs to know. Normalised Difference Vegetation Index, or NDVI, tracks how much green biomass is actively growing across a field, it distinguishes a field that looks green from one that is actually photosynthesising. Soil moisture indices derived from microwave radar tell agronomists how deep the water has penetrated after a rain event, which matters more than surface wetness for germination. Land surface temperature anomalies flag heat stress before it shows in the plant. And repeat-pass imaging, the same field captured every five to twelve days, builds a time series that reveals whether soil conditions are improving, stable, or deteriorating.

ISRO's National Remote Sensing Centre in Hyderabad processes this data and feeds it into advisory systems that state agriculture departments and the Indian Council of Agricultural Research can push down to district agriculture officers, and increasingly, directly to farmers through apps and SMS services.

The gap between satellite data and a farmer's decision used to be wide. The data existed, the researchers used it, but the translation into an actionable planting date for a specific village in a specific week was slow and bureaucratic. That gap has narrowed through a combination of better data pipelines and mobile penetration.

The FASAL project, Forecasting Agricultural output using Space, Agro-meteorology and Land based observations, has been running since the mid-2000s and is jointly managed by the Ministry of Agriculture and ISRO. It uses satellite-derived crop area estimates and soil moisture readings alongside ground weather stations to produce pre-harvest yield forecasts and, more recently, sowing advisories. Farmers in some districts now receive SMS alerts that combine satellite soil moisture data with IMD rainfall forecasts to suggest an optimal sowing window, not a season-level average, but a week-level recommendation for their block.

Kisan Suvidha, the central government's farmer services app, aggregates some of this data. State-level systems in Andhra Pradesh, Telangana, and Maharashtra have gone further, building district-specific dashboards that agriculture extension officers use during field visits. The officer arrives at a village with data on that village's fields, not a regional average.



Satellite data improves the odds; it does not eliminate risk. The revisit cycle of most optical satellites means a cloud-covered field during the monsoon can go unobserved for the exact days that matter most. Radar satellites solve the cloud problem but require more processing expertise to interpret. Ground-truth calibration, sending someone to actually check whether the satellite's soil moisture reading matches what the soil contains, is expensive and patchy across India's farming districts.

There is also the question of what a farmer can do with the information once they have it. A planting advisory is useful only if the seed is available, the soil preparation is done, and the labour is ready. Satellite data can tell a farmer in Bundelkhand that the optimal sowing window opens in four days. Whether they can act on that in four days depends on credit, supply chains, and infrastructure that no satellite can fix.

What the data does is shift the uncertainty from the unknowable to the manageable. A farmer who previously had no way to know whether the soil fifty centimetres down had enough moisture to support germination now has an estimate. It may be imperfect. It is better than nothing, and in a system where a single failed kharif can push a household into debt, better than nothing is worth a great deal.