Inversion Of Soil Salinity In The North And South Hills Of Lanzhou City Based On Multi-Source Remote Sensing Fusion

Soil salinization is a serious constraint on the sustainable use of soil resources,while the north and south hills of Lanzhou City are located in the Gansu,Mongolia and Xinjiang arid-desert realm salinity zones,where salinization is a particularly serious problem in China,making it urgent to carry out monitoring of them.In this paper,based on the Landsat series multispectral data(Landsat5,Landsat8,Landsat9)and MODIS(Moderate Resolution Imaging Spectroradiometer)hyperspectral data product MOD09 GQ,a spatio-temporal fusion algorithm is used to High spatial and temporal resolution fused images were generated.The optimal model for soil salinity inversion was constructed using field sampling experimental data and fused images to quantitatively invert the soil salinity in the north and south hills of Lanzhou from 2000 to 2022.The spatial and temporal dynamics of saline soils of different grades were analysed.The main findings of the study are as follows:(1)The STARFM(Spatial and Temporal Adaptive Reflectance Fusion Model)was used to fuse the multispectral Landsat and hyperspectral MODIS data,and the obtained fused images with high spatial and temporal resolution can significantly improve the accuracy of soil salinity inversion,with a correlation of The correlation is above 0.9 and the root mean square error of surface reflectance is within 0.7,which provides data guarantee for the subsequent soil salinity inversion.(2)The optimal model for soil salinity inversion was constructed.The correlation analysis between the spectral reflectance of each band and the soil salinity showed that the sensitive bands for soil salinity were blue,green,red,short-wave infrared 1 and short-wave infrared 2.Subsequently,using the sensitive bands as the independent variables and the soil salinity as the dependent variable,the inversion models for soil salinity were constructed using partial least squares regression,BP network neural model and a combination of the two.After accuracy evaluation,the combined model was selected as the optimal model to invert the soil salinity content in the north and south hills of Lanzhou from 2000 to 2022.(3)Spatial and temporal dynamics of different grades of saline soils.The soil salinity inversion results were graded,and the spatial and temporal dynamics of soil salinity were analysed using the dynamic attitude and transfer matrix.The results show that: spatially,from 2000 to 2022,soil salinization in the north and south hills of Lanzhou City decreases from north to south;temporally,it first increases,then decreases and then slightly increases,and the total area of saline soils shows a decreasing trend.2000-2022,the interconversion between different grades of saline soils is frequent,and from 2000 to 2005 and 2005 to 2010,it mainly converts to heavily saline soils;from 2010 to The stability of each class of saline soils is strongest in 2015;from 2015 to 2020,the conversion is mainly to non-saline soils;from 2020 to2022,the conversion is mainly to light saline soils.(4)Drought is the main cause of increased soil salinity in the study area,while the study area is dominated by chalky-sand soils,where soil texture and capacity affect the permeability of the soil,resulting in a spatially differentiated distribution of soil water content and limited salinity leaching and washing.Among the human activities,environmental greening works are effective in improving saline soils.