Abstract: Karst rocky desertification is one of the serious ecological and environmental problems in Southwest China. As the inheritance and continuation of Landsat and SPOT (systeme probatoire d’observation de la terre) series satellite data, Sentinel data has better application prospect in ecological and environmental monitoring. By the traditional method, the Karst rocky desertification mapping extracted from Sentinel-2A is only at 20 m resolution and this method wastes the detailed information. In order to extract the Karst rocky desertification more accurately, this research constructs 10  m resolution rocky desertification index for rocky desertification mapping of Beimen River basin in Wenshan, Yunnan based on the characteristic of Sentinel-2A data and analyzes the geological causes. By comparing four fusion algorithms, à trous wavelet transform has the better effect than high pass filtering, principal component analysis and  Gram-Schmidt transform  in short-wave infrared band (B12). The results show that the 10 m  normalized differential rocky index (NDRI) and rock exposure rate (fr )  have good correlation with 20 m NDRI and fr (the correction coefficients are 0.90 and 0.81, respectively), and they can provide more information of different ground objects. It indicated that the 10 m resolution index of Sentinel-2A has better performance for Karst rocky desertification mapping in this study.  The geological causes analysis of Karst rocky desertification in  Beimen River basin shows that the Karst rocky desertification mainly consists of  potential Karst rocky desertification and light Karst rocky desertification, which account for  40.66% and 16.97% of the total  basin area.  Karst rocky desertification is distributed mostly in the Gejiu  Formation and Banna  Formation. Within slope range of 0°-25°, all types of Karst rocky desertification are concentrated development , accounting  for 82.71% of the total area of Karst rocky desertification. 

Key words: Karst rocky desertification mapping, image fusion, construction indexes, causes analysis；Sentinel-2A