Application of Artificial Intelligence for Predicting Erosion of Biochar Amended Soils

Abstract

Recently, incentives have been provided in developed countries by the government for commercial production of biochar for soil treatment, and other construction uses with an aim to reduce a significant amount of carbon emissions by 2030. Biochar is an important material for the development of circular economy. This study aims to develop a simple Artificial Neural Network (ANN) based model to predict erosion of biochar amended soils (BAS) under varying conditions (slope length, slope gradient, rainfall rate, degree of compaction (DoC), and percentage of biochar amendments). Accordingly, a model has been developed to estimate the total erosion rate and total water flow rate as a function of the above conditions. The model was developed based on available data from flume experiments. Based on ANN modelling results, it was observed that slope length was the most important factor in determining total erosion rate, followed by slope gradient, DoC, and percentage of biochar amendment. The percentage of biochar amendment was a leading factor in the total water flow rate determination as compared to other factors. It was also found that the reduction in erosion is relatively minimal during an increase in slope length up to 1.55 m, reducing sharply beyond that. At a slope length of 2 m, erosion is found to be reduced by 33% (i.e., 2.6 to 1.75), whereas the total flow rate decreases linearly from 1250 mL/m2/min to 790 mL/m2/min. The ANN model developed shows that soil biochar composite (SBC) with 5% biochar amendment gave the best results in reducing soil erosion. This study can be a helpful tool in providing preliminary guidelines for using biochar in erosion control.