Losses Assessment for Winter Crops Based on Satellite Data and Fuzzy Logic

This paper considers the method of the winter crop classification map producing in terms of climatic and weather abnormal conditions in 2020. Given that the traditional method of construction involves the use of a training sample, which is collected in ground surveys along the roads. This sample could not be collected under the strict quarantine regime, that is why the classification map was created based on the sample obtained as a result of the photointerpretation. Both, optical Sentine1-2 and SAR Sentinel-l satellite data were used. This is due to the fact, that the period of the winter crop classification map producing fell exactly on the period of time (April and May 2020), when the area of study Odesa region (as well as the whole territory of Ukraine) had a high percentage of cloud cover. At the same time, radar imaging techniques allow us to bypass obstacles such as clouds, but also have lower sampling quality. Therefore, it was decided to combine the obtained classification maps based on radar and optical data by fuzzy logic, considering the degree of belonging of each pixel by the value of the normalized difference vegetation index (NDVI).As a result, the obtained classification maps based on photointerpretation sample have an accuracy close to 95%. The fuzzy logic method allows to increase this value by selecting only the best pixels from classification maps based on radar and optical satellite data.