Scholarly record

METHOD OF FORECASTING THE AGRO-ECOLOGICAL STATE OF SOILS ON THE EXAMPLE OF THE SOUTH OF UKRAINE

D.S. Breus, Olga Yevtushenko, С.В. Скок, Olena Rutta

Abstract

Land resources are the main means of production and a factor of socio-economic development and ecological sustainability of environment[1]. The condition of soils is one of the main indicators of the ecological state of the territories, because of the direct impacts from internal factors, which are caused by the use of soils in agricultural production[2] and external influences caused by anthropogenic activities[3]. Unsatisfactory condition of the soil cover in Ukraine, which is confirmed in scientific works of the leading Ukrainian scientists, determines necessity of soil fertility forecasting, which in its term will allow to identify risk zones with the most unfavorable conditions for agricultural activity and to determine the optimal reclamation measures to improve the qualitative state of degraded soil[4]. Forecasting the agro-ecological state of soils makes it possible to establish the spatial-temporal patterns of its changes under the influence of anthropogenic and natural factors. The complexity of the forecasting processes is determined by the multifactorial and temporal conditionality of the destruction of the natural properties of soil fertility[5]. The use of traditional statistical methods to forecast the corresponding complex stochastic and dynamic processes significantly reduces the reliability of the obtained results. The article uses the method of artificial neural networks, which provides the possibility of nonlinear interpretation of large arrays of input data, interactive adaptation of created models to new information, with high accuracy interpret retrospective arrays of data and highly accurate forecasting the nonlinear processes[6]. Using the Statistics Neural Networks (SNN) module, neuromodels of the three-layer perceptron's architecture were created to forecast soil fertility in South of Ukraine in the soil layer 0...20 cm for the main agrochemical parameters[7].

Publication details

Back to publication list