Determining the desertification intensity based on spectral indices using Sentinel-2 images (Case study: Sistan and Baluchestan province)

Different vegetation covers have different albedo levels. On the other hand, surface albedo is one of the most important components of surface radiation balance, which can be used to identify severely degraded and desertified regions. Vegetation can be considered as one of the most important key components in arid regions to reduce the effects of erosion and desertification due to the effects of vegetation for land surface stability. Expansion of desertification and also changes in vegetation cover, could be change the surface Albedo. The purpose of this study is to determine the desertification intensity based on spectral indices, Albedo, Topsoil Grain Size Index (TGSI) and Normalized Difference Vegetation Index (NDVI) using remote sensing technology. Identification the damaged areas with the lowest cost in the shortest time, using Sentinel-2 images with a spatial resolution of 10 meters is one of the objectives of this study. Also, this study will introduce the best indicator for monitoring desertification intensity in arid regions for the first time in the Sistan and Baluchestan region based on spectral indices using Sentinel-2 images.

 The following steps were performed to evaluate the intensity of desertification and identify the appropriate indicator in order to mapping the desertification intensity: 1) Selection the images and perform image preprocessing operations using SNAP software; 2) Calculation of TGSI, NDVI and Albedo indices; 3) Investigation the correlation between indices using SPSS®24 software. 4) Preparation of desertification intensity map of the region and obtaining the equation of desertification intensity using ArcGIS®10.3 software. In the first step of this research, Sentinel-2A satellite data related to MSIL-1C sensor was selected on August 20, 2020. The images were selected in such a way that the growing season of the plants is not annual and temporary, and also the day was selected when there is no cloud cover. The required images were downloaded and used from the URL address: http://scihub.copernicus.eu/.

 The results of linear regression between NDVI and Albedo indices showed that, these two indices had negative correlation, and the correlation coefficient in Souran and Zabol was 0.76 and 0.63, respectively. The results showed that with increasing NDVI, decreased of the albedo index occurred. Also, the results of linear regression model showed strong and positive relationship between TGSI and Albedo indices, as the correlation coefficient of Souran and Zabol was 0.78 and 0.81, respectively. The results showed that the TGSI and the albedo simultaneously decreased or increased. Desertification intensity in the study areas was determined based on the equation I= a × Index ± Albedo and also by using Natural Breaks (Jenks) method in ArcGIS software, desertification intensity of study areas classified to 5 degrees, 1. Without desertification, 2. Low desertification, 3. Moderate desertification, 4. Severe desertification, and 5. Extremely desertification. In this study Albedo, NDVI and TGSI indices were extracted based on Sentinel-2 satellite data. The results of linear regression between NDVI and Albedo showed that there is strong negative relationship between these indices that was consistent with the results of similar studies. The high and negative correlation, means that any increase in the vegetation cover will lead to decrease the Albedo. On the other hands the areas with high Albedo, indicate degradation of vegetation cover and bare soil. In the regions with sever desertification intensity, the value of surface Albedo was high and the vegetation cover was low. Classification of desertification intensity in Sistan region based on Albedo-NDVI model showed that 27.73% of the area were in the class of without desertification intensity, 18.03% in the low class, 32.92% in the moderate class, 20.3% were in the severe class and only 1.02% of the area were in the very severe desertification intensity class. Also, the classification of desertification intensity in Souran based on Albedo-NDVI model showed 4.82% of the area without desertification, 8.44% in low class, 50.97% in moderate class, 34.48% in severe class and 1.3% of the area were in very severe desertification class. The highest percentage of desertification intensity of the area were in the moderate class. The results of linear regression between TGSI and Albedo indices also showed that there is a positive and strong relationship between these indices. The results showed that the relationship between TGSI and Albedo indices was stronger than the relationship between NDVI and Albedo indices and in both regions the correlation coefficient was higher. One of the main reasons for this is the dispersion of vegetation cover in arid areas. The relationship between TGSI and Albedo better shows the spatial characteristics of vegetation-free areas as well as areas with very low vegetation cover to determine the intensity of desertification. The TGSI index reflects the coarse particle size of the topsoil, which has a positive relationship with the fine sand content of the topsoil. Whatever the larger particle size of the topsoil, will have the greater desertification intensity. In the areas where the content of fine sand in the topsoil is high, the high range of TGSI index will be seen.

 In this study, using Sentinel-2 multispectral images and remote sensing technique, we extracted the intensity of desertification in different arid regions of the Sistan and Baluchestan province, for the first time in Iran. Based on the spectral reflection that occurred from the ground and the spatial resolution of 10 meters, we studied the intensity of desertification in two areas. Based on the results of this research, we suggest to use the combination of Albedo-NDVI and Albedo-TGSI models in order to monitoring the desertification intensity in arid regions of Iran. The results of this study showed that areas without desertification and low intensity of desertification are better identified based on Albedo-TGSI model.