Produção Nacional
Spatial Prediction of Soil Depth Via the Combination of Multiple Remote Sensing Techniques
2022; RELX Group (Netherlands); Linguagem: Inglês
10.2139/ssrn.4145255
ISSN1556-5068
AutoresGabriel Pimenta Barbosa de Sousa, José Alexandre Melo Demattê, Lucas Rabelo Campos, Mahboobeh Tayebi, Merilyn Taynara Accorsi Amorim, Jorge Rosas, Fellipe Alcântara de Oliveira Mello, Lucas Tadeu Greschuk, Songchao Chen, Shamsollah Ayoubi,
Tópico(s)Remote Sensing and LiDAR Applications
ResumoSoil depth is one of the most critical factors which impact on culture productivity and makes difficult appropriate management decisions. However, assessing this parameter is also the most challenging tasks in the agronomic field. The objective of this work was to predict the spatial distribution of soil depth using remote sensing data and machine learning techniques. 292 sites were allocated (based on the toposequence approach) and perforated (from 0 to 2 m) at three different locations in Brazil. Based on these, in-situ traditional depth maps (denominated empirical) were elaborated as for future validation. Afterwards, we elaborated a strategy to achieve these different depths by remote sensing (RS) approach. Landsat 8 OLI bands, Land Surface Temperature (LST), Normalized Difference Vegetation Index (NDVI) and emissivity in dry and rainy seasons as well as terrain attributes were applied to predict soil depth. For this purpose, the most important covariates were selected using Recursive Feature Selection (RFE) based on Random Forest (RF) and Support Vector Machine (SVM). Afterwards, the application of RF and SVM by selected covariates were compared based on ten-fold cross validation for each location. The best model was selected based on R 2 (coefficient of determination), RMSE (Root Mean Square Error) and MAE (Mean Absolute Error) were used to assess the accuracy of developed models and bootstrapping approach was applied to make uncertainties in each location. Finally, the predicted map was compared with empirical one. The results in all locations indicated that the terrain attributes were the most important factors and RF had the highest performance in predicting soil depth.
Referência(s)