A New Method for Estimating Soil Fertility Using Extreme Gradient Boosting and a Backpropagation Neural Network

• DOI: 10.3390/rs14143311
• Publication Date: 2022-07-11T04:06:21Z
• Authors: Yiping Peng; Zhenhua Liu; Chenjie Lin; Yueming Hu; Li Zhao; Runyan Zou; Ya Wen; Xiaoyun Mao
• Keywords: Backpropagation

Soil fertility affects crop yield and quality. A quick, accurate evaluation of soil is crucial for agricultural production. Few satellite image-based studies have quantified during the growth period. Therefore, this study proposes a new approach to quantitative fertility. Firstly, optimal spectral variables were selected using integration an extreme gradient boosting (XGBoost) algorithm with variance inflation factor (VIF). Then, based on where red-edge indices introduced first time, estimation models developed backpropagation neural network (BPNN) assess The model was finally adopted map Sentinel-2 imagery. This performed in Conghua District Guangzhou, Guangdong Province, China. results our research are as follows: (1) five (inverted chlorophyll index (IRECI), vegetation (CVI), normalized green-red difference (NGRDI), position (REP), triangular greenness (TGI)) variables. (2) BPNN established provided reliable estimates fertility, determination coefficient (R2) 0.66 root mean square error (RMSE) 0.17. nonlinear relation found between (3) provides potential mapping images, R2 0.62 RMSE 0.09 measured estimated results. suggests that proposed method suitable paddy fields.