• 标题：Improving estimation of soil organic matter content by combining Landsat 8 OLI images and environmental data: A case study in the river valley of the southern Qinghai-Tibet Plateau
• 关键词：Environmental factors; Landsat 8 OLI images; Precision agriculture; Qinghai-Tibet Plateau; Soil organic matter
• 作者：

全文下载

We presented a methodology to improve estimation of the SOM content. ? The addition of environmental factors improved the SOM content estimation accuracy. ? The first-order derivative of band reflectance was superior in estimating SOM content. ? Non-destructive monitoring of the SOM content was carried out in the QTP. The Qinghai-Tibet Plateau (QTP) is a typical ecologically fragile area. Once the surface vegetation degenerates, it may not be restored. This requires the development of soil organic matter (SOM) monitoring method without destroying the surface, so as to ensure the sustainable development of plateau agriculture. This work investigated the environmental factors that are significantly related to SOM content in the river valley of the southern QTP. These environmental factors include soil hydrothermal factors (soil moisture content and soil temperature), topographic factors (elevation and slope) and vegetation factor (NDVI). The original band reflectivity (OR) of Landsat 8 OLI images and the band reflectivity after the first-order derivative (FDR) and the second-order derivative (SDR) processing were combined with the above environmental factors to estimate SOM content. The results showed that the accuracy of the model was improved obviously by adding environmental factors. The estimation effect of back propagation neural network (BPNN) model was better than that of geographically weighted regression (GWR) model, partial least squares regression (PLSR) model and multivariable linear regression (MLR) model. GWR model can also meet the estimation requirements, while PLSR and MLR models cannot achieve effectively the estimation of SOM content. FDR-BPNN model considering environmental factors was the best model for estimating SOM content, with R2 being 0.947, RMSEC being 4.701 g·kg?1 and MAEV being 5.485 g·kg?1. Moreover, the model had the lowest uncertainty and the highest stability. This study will provide a good insight for the monitoring of SOM content in the future, and provide basic data support for the implementation of precision agriculture in the QTP.