Hayfield Mapping in the Floodplain Landscapes of Southern Russia Based on Multitemporal Sentinel-2 Data

Abstract

This paper proposes a new method for mapping hayfields in floodplain landscapes based on the use of multitemporal spectral–zonal data of earth remote sensing (ERS) of high spatial resolution (Sentinel-2) using an expert threshold of spectral brightness coefficient (SBC) in the red channel (the maximum composite of values for the vegetation period) for freshly cut vegetation adjusted for the values of the maximum composite for the growing season of the normalized difference vegetation index (NDVI). The regularities of changes in the values of SBC in the sloping and nonsloped territories in the RGB and NIR channels, as well as the values of the NDVI and NDWI indices, are revealed. Annual sloping areas within the Volga-Akhtuba floodplain (VAF) in Volgograd oblast are mapped. Here, an average of 12 000 ha (8%) of the territory is mowed annually, while most of the area is mowed in August–September (more than 65% of the area). Most sloping areas are 1 to 10 ha. At the same time, over the past 6 years, there has been a tendency toward an increase in both the total annual mowed areas and the areas of hayfields. It is revealed that the main annually mowed areas are concentrated around infrastructure facilities: closer to consumers and transport routes.