Investigation Of Remote Sensing Inversion And Spatio-Temporal Variations Of Soil Salinization In The Yellow River Delta Based On Fusion Data

Soil salinization is the most common land degradation problem in coastal areas of China.The Yellow River Delta is a typical coastal saline soil distribution area in China.The groundwater in this area is shallow and highly mineralized.Affected by land subsidence,climate change,seawater intru-sion and human activities,the ecological and environmental problems are becoming increasingly prominent.Soil salinization restricts the sustainable development of local agriculture and economy.Therefore,remote sensing technology is used to timely monitor soil salinization and grasp the tem-poral and spatial evolution characteristics of saline soil distribution on different time scales,which is of great significance to formulate soil salinization mitigation strategies.In the dynamic monitoring of soil salinization,the remote sensing images with medium and high spatial resolution generally have the disadvantages of low temporal resolution and easy to be dis-turbed by clouds and rain,which greatly affects the frequency of quantitative remote sensing inver-sion and is difficult to meet the needs of studying the dynamic change characteristics of soil salinity.For the Yellow River Delta,there is a lack of research on the temporal and spatial changes of soil salinity on multiple time scales.Taking the Yellow River Delta as the study area,this paper uses data fusion technology to obtain multi temporal remote sensing image data of the study area based on Landsat series data and MODIS data from 2005 to 2021.Based on the modeling methods of multiple linear regression,partial least squares regression and random forest,the Quantitative Inversion Mod-eling of soil salt was studied.Based on the soil salt inversion data,the temporal and spatial variation characteristics of soil salt in the Yellow River Delta were analyzed.The main research contents and conclusions of this paper are as follows:（1）High spatial-temporal resolution time series image generation.Based on Landsat series data and MODIS data,ESTARFM model is used to fuse and obtain high temporal and spatial resolution surface reflectance data to generate relatively complete interannual time series images and short time scale（8 days）time series images in spring.Comprehensively evaluate the accuracy of the fused data and its application potential in soil salinity inversion.The results show that from the subjective eval-uation,the fused data is basically consistent with Landsat observation data in terms of texture struc-ture and feature information.From the objective evaluation,the scatter points of pixels with the same name in each band are concentrated around the diagonal,and the error is less than 4%.The error of visible near-infrared band is less than 3%,which is better than the 5%uncertainty of Landsat-7 ETM+surface reflectance data.ESTARFM fusion data has high accuracy and can provide data support for long-term high-frequency dynamic monitoring of saline soil.（2）The construction of soil salt prediction model and the selection of optimal soil estimation model.Full band reflectance is used as input variable.Based on multiple linear regression（MLR）,partial least square regression（PLSR）and random forest（RF）algorithm,salt prediction model is constructed and its accuracy is evaluated and compared.The results show that the accuracy of RF salt prediction model is the highest,with modeling set R²=0.917,RMSE=0.727 g kg^-1,validation set R²=0.893,RMSE=0.743 g kg^-1,especially in the range of non saline soil.At the same time,the verification results of independent sampling data in 2012 and 2021 also show that the prediction accuracy of RF model is better than MLR and PLSR model,R² is 0.867 and 0.860,and RMSE is0.817 g kg^-1and 1.420 g kg^-1.（3）Analysis of temporal and spatial variation of soil salinity in the Yellow River Delta on mul-tiple time scales.Based on the high-frequency soil salt data in the Yellow River Delta in recent 20years,starting from the interannual and intra annual short-time scales,the temporal and spatial vari-ation characteristics of soil salt content are analyzed by means of area statistics,calculated dynamic degree,saline soil change trend mapping,calculated transfer matrix and so on.The results show that from the short time scale in spring,the overall salinization degree has an obvious downward trend.During this period,the area of non saline soil and mild saline soil increases,and the area of moderate saline soil decreases.From late March to early April,there was a short-term rebound in salt content.From the dynamic attitude of different types of saline soil,the change rate of mild saline soil is the fastest from February to March,and the area of moderate saline soil decreases faster from March to April.The most stable type of saline soil in the study area from February to April is saline soil.From the perspective of spatial change and transformation structure of saline soil types,the degree of sali-nization showed an overall decreasing trend from February to March.All types of saline soil in the study area mainly transformed to adjacent low-grade saline soil（mild–non saline soil,moderate–mild,severe–moderate,saline soil–severe）.The distribution of heavy saline soil is the most unstable,and it is easy to transform to medium saline soil.From March to April,the degree of soil salinization in the middle of the study area（Lijin County and Kenli District）showed an aggravating trend.On the interannual scale,the soil salt content in the Yellow River Delta increased first,then decreased,and then remained unchanged from 2005 to 2021,with the highest value in 2008(4.930 g kg^-1)and the lowest value in 2015(3.299 g kg^-1).During this period,the area of non saline soil showed an increas-ing trend,and the change of slightly saline soil fluctuated greatly,and the change speed was the fastest.The area of moderately saline soil shows a decreasing trend,while the area of severely saline soil and saline soil is relatively stable.Compared with the intra annual scale,the proportion and area of soil with changed salinization degree in the Yellow River Delta are larger on the interannual scale.From2005 to 2021,the proportion of saline soil with decreasing trend is always higher than that with in-creasing trend.From 2005 to 2021,the area distribution of saline soil in the northern part of the study area and the eastern coastal area of Kenli District decreased significantly,mainly to heavy saline soil,water area and construction land.The area of non saline soil and slightly saline soil in inland and along the Yellow River has increased significantly,and the degree of salinization has been signifi-cantly improved.