Study Of Land Degradation Remote Sensing Monitoring Method In The Arid Region Using Minqin County,Gansu Province As A Case

As an important factor threating the mankind existence,social stability and sustainable development in the 21st century,land degradation in the drylands has drawn great attention of countries.How to effectively control and slow down the land degradation in the drylands has been the research emphasis and hotspot of international environmental area.Prevention measures depend on accurate extraction of the degradation information.The development of remote sensing technology greatly improves the ability of land degradation information extraction.It can effectively reveal the condition and dynamic of land degradation in the drylands and thereby provide the scientific basis for land degradation trend prediction and formulating the corresponding preventing measures.This study made the remote-sensing technical framework for land degradation monitoring based on 2010 and 2015 multi-seasonal Landsat images of spring,summer and winter,which can reflect the different landscape characteristics of surface features,and 2010 field data,using Minqin county,Gansu province as a case.We firstly realized dryland m ulti-seasonal linear spectral mixture analysis,then used the resulting endmember fractions to complete the land cover/use decision-tree classification of Minqin county in 2010 and 2015.Next,we used the land cover/use classification and endmember fraction results to evaluate the governance effect of ecological engineering for MInqin county during the second phase in the early(2010?2015).Last,we quantitatively simulated the saline and alkaline degree of agricultural soil in the cropping season.The major results of this study includes:(1)Dryland multi-seasonal linear spectral mixture analysis.The results show that sand,salt,vegetation,dark material and water can be spectral endmembers of dryland multi-seasonal spectral mixture analysis,and the representative seasons for them are different.All the mean RMSE(Root mean square residual error)values for the three seasons in 2010 as well as 2015 are less than 0.02,indicating five-endmember linear mixture model can effectively simulate spectral information of surface features in different seasons.(2)Land cover/use classification of Minqin county based on cut tree.The results show that,using multi-seasonal endmember fractions can obtain classification rule quickly and effectively.The producing tree has a simple,reasonable and clear structure,and relatively high classification accuracy.The classification results indicate that desertification land is the major cover type in 2010 and 2015,accounting for more than 80%of the study area's total area.Within it,sandy land is the major type,and the area of salinization land is relatively small.(3)The ecological management effectiveness evaluation for Minqin during the second phase in the early(2010?2015).The research finds that the land degradation area is 3.97×104hm2,and the area of restoration is 1.02×105hm2.The area of land restoration is larger than that of degradation,which indicates Minqin reached the zero net land degradation goal from 2010 to 2015,and was in a relative recovery situation.Accordingly,sand land area and sandification degree in invariant sand land are decreased.Although the area of salinized land has increased,the salinization degree of invariant region is decreased.Water area is somewhat reduced,but the water content in invariant region is increased.Forest and grass land compared to crop land,and summer crop compared to spring crop are preponderant vegetation cover.This research shows that Minqin has already enter into virtuous restoration phase after nearly 20-year ecological restoration governance.(4)Agricultural soil salinity and alkalinity quantitative inversion in the cropping season.The results show that the linear model of partial least squares regression cannot effectively describe the relationship between soil salinity/alkalinity and environmental variables,but artificial neural network(ANN)can express the complicated relationship between them by using the non-linear transfer function.Based on the inversion results of ANN,we find the quantity of salty soil is limited,and soil alkalization becomes the major threat of agricultural land.In terms of spatial distribution,salinization and alkalization degree in the lower region of Minqin are both stronger than those in the upper region.In addition,it finds that LST is important for Agricultural soil salinity and alkalinity inversion in the cropping season.