Inversion Of The Degree Of Soil Salinization Based On Spectra Of Soil And Vegetation In Northern Yinchuan Plain Of Ningxia

Soil salinization has become a common ecological and environmental problem all over the world,especially in arid and semi-arid areas,which seriously threatens the sustainable development of local social economy and ecological environment.In northern Yinchuan Plain of Ningxia,a large area of salinized soil is formed,the use of multi-source remote sensing accurately the distribution of soil salinization in the region,area,type and degree of information,such as governance is the important prerequisite of salinization,to prevent further soil degradation,to improve the silver north agricultural production capacity,increase reserve cultivated land resource reserves,has important practical significance to protect the safety of food.In this paper,Pingluo county in Northern Yinchuan Plain of Ningxia was taken as the research area.Use of soil science,statistics and analysis methods of remote sensing technology,with the field measured hyperspectral image data,Landsat 8 OLI and soil based on laboratory data sources,analysis of statistical characteristics of soil salinity index in the study area,the measured hyperspectral and similarities and differences among the Landsat 8 OLI images from the surface soil extract the sensitive or typical vegetation canopy spectral bands and the spectral index.Using Multiple regression methods,the author established the soil salinization index inversion model based on spectral index and sensitive bands.By analyzing the similarities and differences between the measured hyperspectral and Landsat 8 OLI images,the study used the measured spectra to correct the inversion results of remote sensing images,so as to improve the accuracy of image inversion of soil salinization degree,and realized the wide range and fast extraction of salinization information in this region.The main conclusions of this study are as follows:(1)The salt content and pH value of the surface soil in the study area are high,and the degree of salinity is heavy.In general,the spectral reflectance of the soil in the study area has no obvious rule with the variation trend of salt content,but increases with the increase of soil alkalinity.For the measured soil reflectance,the inversion model of soil salinity based on SI2 and the reflectance of sensitive bands after reciprocal differential transformation has the highest accuracy.The pH inversion model based on S3 and the reflectivity of sensitive bands after cosine first order differential transformation has a good effect.For Landsat 8 OLI image,the accuracy of the soil salinity model based on S2 and the reflectance of sensitive bands after logarithmic first order differential transformation is the highest.The pH inversion model based on NDSI and the reflectivity of sensitive bands after logarithmic reciprocal transformation has a good effect.(2)The surface soil in the study area contains high content of Cl-,SO42-and Na+,which generally belongs to the sulfate-chloride salinized soil.For the measured soil reflectance,the salinity index has the strongest retrieval ability for Ca2+(R2=0.6907),followed by Na+ and Cl-.However,the model based on sensitive bands has the best inversion ability for K+(R2=0.7505),followed by Ca2+ and CO32-.For Landsat 8 OLI image,the retrieval ability of Ca2+based on salt index was the best(R2=0.7417),followed by HCO3-,Cl-and K+.In the model based on sensitive bands,the inversion ability of Ca2+is the best,followed by Na+and Cl-.(3)Spectral characteristic curves and red edge parameters of different vegetation canopy have different responses to soil salinization degree.With the increasing degree of salinization,the canopy spectrum of Nitraria tangutorum first increases and then decreases in the visible band,then decreases gradually in the near-infrared band,and the red edge position gradually "blue shift".The canopy spectrum of Phragmites australis decreases first and then increases in the visible band.In the near-infrared band,on the contrary,the position of the red side shifts "blue" first and then "red",indicating that Phragmites australis is more salt-tolerant than Nitraria tangutorum.Under the condition of soil with moderate salinization degree,the trend of spectral characteristic curve of different overlying vegetation canopy is similar.Soil pH value and chlorophyll content of vegetation leaves were significantly correlated with the red edge parameters,but there was no correlation between soil salinity and the red edge parameters.(4)For the measured reflectance of vegetation,the inversion model of soil salinity based on EEVI and the reflectance of sensitive bands after reciprocal transformation has the highest accuracy(R2 reached 0.2929 and 0.5693 respectively);The pH inversion model based on EVI and the reflectivity of the sensitive band after smoothing and denoising is effective(R2 reached 0.6257 and 0.5975 respectively).For Landsat 8 OLI image,the accuracy of the soil salinity model based on EVI and reflectance of sensitive bands after reciprocal logarithmic transformation was the highest.The pH inversion model based on TGDVI and the reflectance of the sensitive band after the reciprocal transformation has a good effect.Based on the measured hyperspectral inversion model,the Landsat 8 OLI image model was corrected,and the R2 of the corrected vegetation index and the salinity inversion model of the sensitive bands were increased by 0.3207 and 0.3762,respectively,and the coefficient of the pH model was increased by 0.2065 and 0.2487,respectively,which can be used to extract salinization information over a large area.