Dynamic Monitoring Of Soil Salinity In The Yellow River Delta Based On Measured Spectrum And Time-series Remote Sensing Data

In the process of accelerating socialization and deepening globalization,soil system has suffered different types of damage,among which soil salinization is the most extensive soil damage type restricting agricultural development.The study of soil salinization by remote sensing can provide technical and data support for soil ecological restoration,which is of great significance for environmental rational exploitation and social sustainable development.In this thesis,the Yellow River Delta is taken as the research area,spectral correction is carried out on the time-series remote sensing image by using the measured hyperspectral data,and soil salt influence factor database is constructed based on a variety of image factors.Combined with the time-series remote sensing data,soil salt monitoring model is constructed to realize dynamic monitoring.On this basis,measured spectral data and Sentinel-2 remote sensing images of different phases in March,May,September and December were used as data sources.By spectral correction method,the influence of environmental factors such as water classification on reflection spectra in remote sensing images was removed,and spatial heterogeneity,distance and other factors were comprehensively considered.The correlation between soil salinity and modified spectrum,salinity index,vegetation index,surface moisture index and spatial relative distance was established.Three different soil salt inversion models were constructed by partial least square regression,multiple linear regression and BP neural network.The accuracy was tested using measured salt data,and the model with the best performance was selected.The soil salt inversion calculation was carried out on the sentinel remote sensing data of different phases in four periods,and the dynamic monitoring was realized.The results are as follows:(1)The measured spectral data was used for radiation normalization correction of the image spectrum,and the average fitting accuracy of each band of the measured spectrum and the image spectrum was improved from 0.474 before correction to 0.740after correction,which could remove a certain degree of interference of environmental factors and significantly improve the correlation between the image spectral data and soil salinity.(2)Considering the influence of multiple factors and the integrity and spatial heterogeneity of geographical environment,the modified image spectrum,vegetation index,salinity index,surface moisture index and spatial location information were used as the influencing factors of soil salinity.The correlation of various factors was calculated,and a total of 22 soil salt sensitive factors were extracted,including 12 band spectra,3 vegetation indices,3 salinity indices,1 surface moisture index and 3 spatial location information.A long time-oriented soil salinization influence factor database was constructed.(3)Partial least squares regression,multiple linear regression and BP neural network models were constructed to consider the linear and nonlinear relationship between soil salinity and influencing factors.The highest fitting accuracy R~2=0.845 for partial least square regression model,0.779 for multiple linear regression model and0.598 for BP neural network model were obtained.Compared with the accuracy of the model,partial least squares regression model based on soil salinization influence factor database was selected as the best model for soil salt dynamic monitoring.(4)In terms of time distribution,soil salinization was more severe in March and December,and less severe in May and September,which was mainly affected by seasonal precipitation and crop growth.The rainfall in May and September is abundant,which has a certain dissolving and buffering effect on the salt in the soil,carrying the salt underground or flowing into rivers and oceans with the current.Moreover,the vegetation grows luxuriant and can absorb part of the salt.Precipitation in March and December is low,and evaporation is greater than precipitation.When water evaporates,a large amount of salt will be brought to the surface to form condensation,which increases the salt content of the surface soil and leads to the difference of soil salinization degree at different times.(5)In terms of spatial distribution,the degree of salinization in the study area is larger in the west than in the east,in the south than in the north,and in the southwest than in the northeast,and the difference is larger in spring and winter,but smaller in summer and autumn,which is mainly affected by urban and human activities.The soil ecological improvement measures taken since 2009 in the Yellow River Delta have improved the soil salinization in the study area.However,the southwest region is close to the urban agglomeration,and urban development and human activities bring exogenous pollutants to the soil,affecting the improvement effect of soil salinization,resulting in the soil salinization degree in the southwest region is heavier than that in the northeast region.(6)Considering the spatial distance factor,the results showed that the closer the distance to the Yellow River,the less salinization degree,and the closer the distance to the aquaculture water body and the coastline,the more salinization degree.Farmland irrigated with fresh water from the Yellow River can buffer soil salinity to a certain extent,diluting and taking away part of salt.However,near the coastline and aquaculture water,there are salinization plants and saline-alkali land with serious soil degradation.Coupled with the influence of seawater intrusion,soil salinity around the coastline and aquaculture water will increase,leading to the distribution difference of soil salinization in space.However,the image coverage of soil salinization by three kinds of ground objects is limited.Research on dynamic monitoring of soil salinization can provide support for government decision-making,so as to take targeted measures according to the time and space location,maximize the improvement of soil ecology,and help the development of agricultural industry in our country.