Extraction And Analysis Of Rocky Desertification Information Based On NDVI-Albedo Feature Space

Rocky desertification is one of the most influential ecological and environmental problems in Southwest China,which has great influence on people’s living environment and regional sustainable development.The more accurate the information is obtained,the better the effect of rocky desertification control will be,thus providing data support for further research on rocky desertification.Rocky desertification has become a long-standing major ecological and environmental problem in karst areas of southwest China.Spatial and temporal continuous remote sensing images can obtain the multi-year variation characteristics of surface elements,and play an irreplaceable role in regional ecological environment monitoring.Establishing an efficient and accurate information extraction model of rocky desertification based on remote sensing data and quantitatively evaluating its evolution process is of great significance for rocky desertification prediction,ecological assessment and restoration in karst areas.In this paper,Dougu Township,Weining Autonomous County,Guizhou Province is selected as the research area,and Landsat TM/OLI remote sensing image data in 2001,2007,2014 and 2020 are collected.At the same time,firstly,the information of the fourth stage rocky desertification in the study area was extracted based on the traditional conventional method and NDVI-Albedo feature space.Then,the land use type data of the study area was used to mask the areas where rocky desertification was impossible.Finally,the accuracy of the two methods was verified by using Landsat8 OLI images in 2020 and GF-2 data of the same period.On the basis of the accuracy verification results meeting the research requirements,finally,the temporal and spatial change characteristics of rocky desertification in the study area are analyzed from four aspects:rocky desertification transfer change matrix,evolution rate,evolution mode and evolution frequency.The main conclusions of this paper are as follows:（1）The exposed rate of bedrock and vegetation coverage rate are important indexes for information extraction of rocky desertification.Selecting the above two indexes as the traditional conventional methods to extract rocky desertification information,the total accuracy is 69.33%,and Kappa coefficient is 0.60.Among them,the mapping accuracy of moderate rocky desertification is the lowest,which is 45.45%.The exposed rate of bedrock will increase with the aggravation of rocky desertification,that is,the exposed rate of bedrock in severe rocky desertification is the highest,while that in mild rocky desertification is the lower.The image of vegetation coverage is inversely related to the image of bedrock exposure rate.（2）The difference index model of rocky desertification（RSDDI）can accurately monitor and identify different degrees of rocky desertification.The selected NDVI-Albedo feature space index is simple and easy to obtain,and it has a good effect on extracting information of rocky desertification in the study area,with the total accuracy of 83%and Kappa coefficient of 0.78.Therefore,NDVI-Albedo feature space is an information extraction method of karst rocky desertification with great application potential,and it has important application potential in areas with similar geology to the study area.（3）The phenomenon of rocky desertification in the study area is serious,especially in mild and moderate rocky desertification areas,and the sum of the two areas will still account for 53.86%of the total rocky desertification area in the study area in 2020.The spatial distribution pattern is that the degree of rocky desertification is heavy in the northwest and southeast,and light in the east and north.The long-term trend of rocky desertification in the study area shows that the rocky desertification in most study areas has been improved and effectively curbed,but the evolution of rocky desertification in some areas is fluctuating and easy to rebound.（4）According to the indexes that can reflect the evolution of rocky desertification,such as the transformation matrix,the evolution rate,the evolution mode and the evolution frequency,the spatial and temporal changes were analyzed.From 2001 to 2020,the area of transformation and change between different degrees of rocky desertification in the study area was 64.05 km2.Different degrees of rocky desertification have different change rates.The evolution mode of rocky desertification is mainly continuous,accounting for 57.24%;The evolution frequency of severe rocky desertification is the highest,which is-3.00%?a^-1.To sum up,the correlation between vegetation,exposed rate of bedrock and surface albedo in rocky desertification areas provides a solid theoretical basis for distinguishing rocky desertification.The rocky desertification information extraction model constructed in this paper provides a new idea for accurately extracting rocky desertification information,and analyzes the temporal and spatial variation characteristics of rocky desertification from 2001 to 2020,which provides a certain scientific basis for extracting rocky desertification information and remote sensing monitoring and analysis.