Digital mapping of spring wheat yield based on vegetation indices and estimation of its changes depending on the properties of anthropogenically transformed soils

Abstract

A comparative estimation of informativeness of vegetation indices NDVI, EVI, RVI, CTVI, SAVI, and MSAVI2 for digital mapping of yield of spring wheat grown in the southeast of Western Siberia is carried out. Using the obtained linear models, forecast maps of spring wheat yield were constructed, for which the data of spatial distribution of vegetation indices calculated by the Landsat-8 OLI satellite image (30 m resolution) had been served as an indicator and basic cartographic ground. Comparative analysis of the maps showed that the results of yield mapping based on vegetation indices NDVI, RVI, CTVI, SAVI, and MSAVI2 were identical. The results of mapping of spring wheat yield with the use of the vegetation index EVI were unsatisfactory, since in the area with sparse crops the values for yield were overstated by 2 times. The spring wheat productivity and vegetation indices were not statistically significantly different for the agro-dark-gray soil and agro-chernozems. The correlations were significant between the spring wheat yield and vegetation indices on the one hand, and pre-sowing water content and the content of exchangeable potassium on the other hand, while for humus the correlations were moderate. Insufficient pre-sowing water content of soils was a restrictive factor in the formation of spring wheat yield and did not allow agrochemical properties of soils to produce an effect in its increasing.