Crop monitoring is software that reads satellite images of farmland to show what crop is planted in each field, how well it is growing and how much of it has been damaged. It is also called satellite crop monitoring. Its buyers are banks that lend to farmers, insurers that pay crop claims, and governments that forecast harvests.
How does it work? Earth observation satellites pass over every field in India every few days. Some carry optical cameras, which see the field as a photo does, and also see light that the human eye cannot see. Healthy leaves reflect a lot of near-infrared light, so the software can measure how green and dense a crop is in each image. Other satellites carry radar, which sees through clouds. Radar matters in India, because the main growing season, the kharif season, falls in the monsoon, when clouds hide most fields.
The software first marks each field's boundary, often using land records in a geographic information system (GIS). Then AI models, trained on thousands of fields whose crops are known, recognise each crop from how it changes over the season. Rice, cotton and soybean each grow and turn green on their own timetable. AI that reads images of the ground in this way is called geospatial AI.
Consider an insurance company that covers soybean farmers in Latur district, in Maharashtra, under the government-backed crop insurance scheme, the Pradhan Mantri Fasal Bima Yojana (PMFBY). In late September, three days of heavy rain flood the low-lying fields. Farmers in 40 villages report losses. The insurer must check each claim before it pays, and sending inspectors to every field takes weeks. Its crop monitoring software compares radar images from before and after the rain. It shows that about 18% of the soybean area in those villages stood under water for more than two days, and it shows exactly which fields. The insurer sends its inspectors to those fields first. It also checks that each claim matches a field that was actually sown with soybean. Its insurance core system then records each claim and its assessment.
A bank uses the same kind of images differently. Before it lends for the season, it looks at five years of images of a farmer's fields to see what was grown and how reliably. Many farmers have no credit score, so this field history can stand in for one in the bank's underwriting. Alternative credit scoring combines it with other data. Governments use the same images to forecast harvests.
How is it different from satellite imagery analytics in general? Imagery analytics reads satellite images for any purpose, such as counting ships in a port or tracking construction. Crop monitoring is built only for farmland, with models trained on India's crops and seasons.