Remote Sensing for Risk Mitigation in Agricultural Financings: Multitemporal Change Detections in Agricultural Areas using the Delta NIR and Delta NDVI Models

Abstract

Agribusiness is one of the main pillars of the world economy, where the agriculture sector is fundamental to the economy of countries and contributes strongly to poverty reduction, positive balance of trade and inflation control. The countries have government policies that subsidize agricultural credits with low financing rates in order to encourage the sector. With the expansion of agricultural areas to meet the growth in the world population, its necessary a technological revolution in the field, as is increasingly necessary to mitigate risks in the financeable area, with quick and transparent inspection. Thus, a tool in Python language was develop containing a methodology for monitoring the cycle of cultivation in agricultural areas with emission of alerts messages in cases of deviations in the behavior of the planted area. Delta NDVI and Delta NIR models were developed to monitor agricultural cycle and were able to perform remotely the multitemporal detection of changes. The study was in an area in Brazil, using 9 images from the Nano Satellite Planet, from 2017 to 2019. The results were assertive, because the classifications of methods detected changes in the patterns of agricultural, emitting signals in cases of deviation from behavior with alerts for loss of vegetation from initial cycle, full and final of maturation. As the variation between models was not expressive, Delta NIR was an attractive alternative for change detections because uses only one band, the processing is less costly, with low response time and good performance, with execution in less than half time. Thus, the models were assertive and can facilitate inspection by of the countries with subsidies for agriculture, whether by inspectors from the Government or by Financial Institutions, in addition to reducing costs in the operational process concentrating visits only for areas of large hectares.