Mapping forest plantations of the Volgograd region according to remote sensing data using the BSFI and NDWI indices

The possibilities of adjusting the results of mapping wood plantations based on the bi-seasonal forest index (BSFI) according to Sentinel-2 data and calculated seasonal composite images of the NDWI index maximum values are presented. It was found out that with using a seasonal composite image of the mentioned values, calculated with the NIR and SWIR2 ranges, the manufacturer`s accuracy in raster cross-validation increases from 82,5 to 92,8 %. At the same time, the initial areas of wood zones decreased by 7 %, and those of false identifications shrank by 67 %. Comparison of mapping results with statistical data showed small differences in areas with a large number of natural forest plantations (in 18 of 33 areas they do not exceed 15 % of the design ones). Moreover, in lands with a large number of protective green spaces, discrepancies can reach up to 210 % of the design values. It is shown that significant distinctions in design and actual mapping materials can be associated with poor preservation, fires, illegal logging, as well as problems of accounting protective forest plantings. A fairly accurate method for adjusting BSFI was proposed (the manufacturer’s accuracy increased from 82,5 to 92,8 %) based on the maximum composite NDWI image calculated using the NIR and SWIR2 bands of the Sentilel-2A and Sentilel-2B satellites in a test area within the Volgograd region