Development of a method for assessing the state of land fertility based on modeling the productivity of agricultural crops in natural zones of the Altai region

The paper presents the results of research on the development of a method for calculating the total soil score (TSS) based on modeling the yield of agricultural crops to assess the fertility level of agricultural lands compared to reference sites, taking into account their selection for intrazonal and interzonal differentiation using the geometric mean of the relative values of key soil, hydrothermal, climatic and vegetation indicators, adapted to the conditions of the Altai Territory. The research is based on the combination and improvement of methods developed by the State Research Institute of Land Resources and methods of crop yield modeling according to L.M. Burlakova using machine learning models. The calculations used long-term statistical data from Rosstat (2007-2024), agrochemical indicators, ERA5-Land climate reanalyses (January-May 2007-2024), NDVI/MODIS satellite indices (2013-2014) and OpenLandMap/SoilGrids digital soil maps. A zonal approach with separate calibration of the Random Forest model for seven natural zones was applied. The importance of features was analyzed, and the metric characteristics of various models were assessed. It was shown that the key determinants of crop yields are hydrothermal and hydrological indicators, availability of mineral nutrition elements, and energy balance factors. The developed methodology will improve the accuracy of crop yield forecasts for natural zones, as well as assess the suitability of land plots by calculating the total soil score for making decisions on the priority introduction of land into agricultural circulation, including unused arable land, in accordance with the State Program for the Effective Involvement of Agricultural Land into Circulation and Development of the Land Reclamation Complex of the Russian Federation (2021). The results obtained can serve as a scientific basis for agricultural planning, agromonitoring, and adaptive management of soil resources.