Barley Yield Estimation Performed by ANN Integrated With the Soil Quality Index Modified by Biogas Waste Application

Abstract

Today, the evaluation of soil quality and crop yield has become a critical issue in meeting the increasing population's food needs. The current study aims to analyse and predict the effect of biogas waste (BW) application on soil quality and barley yield. The yield of barley grown in the soil with 0 (B0), 10 (B1), 20 (B2), 30 (B3) and 40 (B4) t ha-1 BW applied and the physical, chemical and biological properties of the soil were examined. In determining the soil quality index (SQI), the analytic hierarchy process and linear combination technique were used, 27 soil indicators in the total data set (TDS) and 10 soil indicators were evaluated separately due to the minimum data set (MDS) created with a principal component analysis (PCA). The relationship between SQI values obtained based on application and barley yield was estimated by applying general regression equations and Levenberg-Marquardt training algorithm in artificial neural networks (ANN). The quality of soil, which was the II class, at the 0 t ha-1 (control) BW for both data sets with biogas waste application was defined as the III and IV soil quality classes. While the increases in barley crop yield were similar to the soil quality index values obtained with the MDS (SQI(MDS)), the optimum yield was obtained at the 30 t ha(-1) BW; with this application, an increase of 35.62% barley crop yield was achieved compared to the 0 t ha(-1) BW. For both data sets, the coefficient of determination (R-2) by general regression in the yield estimates from the SQI had a prediction accuracy of 0.87-0.88. At the same time, the values in ANN were determined as 0.91-0.92. Among the estimation methods, the highest R-2, low root mean square error (RMSE) - 125.5 kg and Akaike information criterion (AIC) - 359.58 were determined by ANN. The study concluded that biogas waste application increases soil quality and barley yield. The MDS can be adopted successfully in soil quality determination, and the barley crop yield can be predicted with high accuracy from the soil quality with ANN.