Sandy desertification monitoring with the Relative Normalized Silica Index (RNSI) based on SDGSAT-1 thermal infrared image

Abstract

The Silica Index (SI) obtained from thermal infrared bands is an effective way for monitoring sandy desertification. Sustainable Development Goals Science Satellite 1 (SDGSAT-1) enriches the thermal infrared imagery globally with three thermal infrared bands. However, the use of SDGSAT-1 for monitoring sandy desertification remains underexplored, especially with radiance instead of emissivity. This study firstly explored the capacity of SDGSAT-1-based SI for large-scale sandy desertification monitoring and its performance was evaluated through the comparison with ASTER-based SI and related studies. Taking the Mu Us sandy land as the study area, a key technique was developed to generate the Relative Normalized Silica Index (RNSI) to remove the land surface temperature effects on multi-temporal SI. Then the annual composite RNSI was acquired by averaging all available non-growing season RNSI scenes and used to classify the sandy desertification with a simple threshold method. The results show that the relative normalization approach could effectively remove the effect of land surface temperature difference in the original SI, ensuring that the average composite RNSI has good spatial consistency. The proposed composite RNSI could effectively classify different sandy desertification lands, with an overall accuracy of 85.7%. Moreover, the RNSI has greater advantages in discriminating moderate and severe sandy desertification lands compared to commonly used NDVI. The proportions of light, moderate, severe, and extreme severe sandy desertification lands account for 21.4%, 28.7%, 29.2%, and 20.7%, respectively. Severe and extremely severe sandy desertification covers nearly half of the study area. Therefore, the protection and recovery of the Mu Us sandy land still require continued attention. Since SDGSAT-1's thermal infrared sensor has a higher spatial resolution (30 m) and wider swath (300 km), the initial findings from this study demonstrate that SDGSAT-1 has great potential for rapid monitoring of sandy desertification with high precision at a large scale.