Spatial And Temporal Evolution Pattern And Driving Mechanism Of Rocky Desertification In Typical Karst Areas

Rocky desertification in China is mainly concentrated in the southwest region.Rocky desertification is due to natural factors and human unreasonable development and utilization of rock exposed.Rocky desertification directly caused a series of ecological and environmental problems such as regional soil erosion,ecosystem function degradation,rivers,lakes,reservoirs silting.Since the 1980 s,the climate has gradually warmed up,human activities have been strengthened and unreasonable development and utilization have led to frequent natural disasters such as landslides and debris flows.In previous studies,single index method,comprehensive index evaluation method and image classification method were mostly used to monitor the spatial distribution of rocky desertification,while various influencing factors of rocky desertification and their interactions were ignored,and the degree of rocky desertification at different levels could not be distinguished.In this paper,Bijie City is taken as the research object.Based on Landsat8 OLI and Landsat5 TM remote sensing images,typical surface parameters such as Bare Soil Index(bare soil index),Dryness Index(dry index),LST(surface temperature),Albedo(surface albedo),RBI(bedrock bare rate),NDVI(normalized difference vegetation index),MSAVI(modified soil adjusted vegetation index)are inverted.Taking Dafang District as an example,two kinds of feature space model methods,point-point model and point-line model,are introduced and the accuracy is verified.The optimal remote sensing monitoring model of rocky desertification was proposed.The model was applied to Bijie City,and different indexes were inverted to explore the spatiotemporal change and driving mechanism of rocky desertification in Bijie City from multiple perspectives.The results are as follows :(1)The applicability of the feature space model is high,and the average accuracy is82.5 %.Among the point-point model and point-line model,the point-point model(BLIDI)based feature space model has the highest accuracy,with the accuracy of 91.3 %.This model has good applicability in Bijie City.Among the ten rocky desertification monitoring index models,RBI-DI feature space model based on point-line mode(90.4%)and DINDVI feature space model(87.5%)were second only to BLI-DI feature space model based on point-point mode(91.3%),while Albedo-BLI feature space model based on point-point mode had the worst monitoring accuracy(72.4%).In the two types of models,the overall accuracy of the point-point model is generally higher than that of the point-line model.(2)From the time series analysis,the overall rocky desertification showed volatility ’weak-strong-weak ’ change from 1985 to 2020.Rocky desertification showed a serious trend from 1985 to 1995.Rocky desertification showed a weakening trend from 1995 to2005.Rocky desertification showed a serious trend again from 2005 to 2010.Rocky desertification showed a weakening trend from 2010 to 2020.Rocky desertification finally showed a significant improvement trend,but the evolution intensity of each period was different.From the migration trajectory analysis,the center of gravity of rocky desertification is mainly distributed in Tianba Town,Weixin Town and Miao Bai Township of Yi Nationality in Qixingguan District.The overall center of gravity of rocky desertification moved westward,indicating that the degree of rocky desertification in the east was higher than that in the west.(3)From the spatial distribution analysis,the rocky desertification in Bijie City is generally moderate rocky desertification.In the distribution of different grades of rocky desertification,severe and extreme rocky desertification are mainly distributed in the Hui and Miao Autonomous Region of Yi Nationality in Weining,and mild rocky desertification and non-rocky desertification are mainly distributed in Qianxi County and Dafang District.From different land use types,no rocky desertification and mild rocky desertification are mainly distributed in forest land,moderate rocky desertification and severe rocky desertification are mainly distributed in cultivated land,and extreme rocky desertification is mainly distributed in unused land.From the perspective of different slopes,no,mild and moderate rocky desertification are mainly distributed between 8° –15° on gentle slopes,and severe and extreme rocky desertification are mainly distributed in gentle areas.From different altitudes,no rocky desertification and mild rocky desertification are mainly distributed in 550-1100 m,severe rocky desertification and extreme rocky desertification are mainly distributed in high altitude and low altitude areas of 0-550 m.(4)In the driving mechanism analysis,different factors have different effects on rocky desertification from 1985 to 2020.In the single factor detection analysis based on geographical detector,the dominant factor with the strongest explanatory power for rocky desertification has changed from vegetation coverage and land use types to human activity factors(population density and GDP).In the interaction dominant factor analysis,the interaction factor changed from vegetation coverage∩land use type to vegetation coverage∩population density and vegetation coverage∩GDP,and the interaction factor showed a linear enhancement relationship with two factors.