Neural network forecast of humus content in slightly eroded ordinary chernozems of the Rostov region

The article presents a long-term forecast of the humus content in the slightly eroded ordinary chernozems of Rostov region based on automated neural networks. The research was conducted over a long period of time in 1986-2022 to study crop rotations, tillage methods and fertilizer application levels on the slope of the Bolshaya Log ravine with a steepness of up to 3.5-4o in the Aksai district of the Rostov region. The forecast revealed a decrease in humus content in the next hundred years without fertilizers. The application of medium doses of fertilizers makes it possible to maintain the humus content at the initial level in crop rotations with varying proportions of perennial grasses, while increased doses confirm the expanded reproduction of soil fertility. Keywords: AUTOMATED NEURAL NETWORKS, HUMUS, FORECAST, SOIL, CROP ROTATIONS, FERTILIZERS