| At present,people are increasingly aware of slowing down land degradation and land resource protection,and the 18th Congress of the Communist Party of China has also clarified the importance of ecological and environmental protection.1.8 billion mu of arable land red line should be kept,and 500 million mu of saline land should also be fully developed and utilized.And We should fully explore the potential of arable land in the Yellow River Delta.If saline land is fully used,it will play an essential role in safeguarding China's breadbasket and China's rice bowl.However,a quick and accurate grasp of the spatial distribution of saline soil salinity is a prerequisite for its management and utilization.Remote sensing has become an essential tool for monitoring soil salinization information.Improving the accuracy of quantitative optical remote sensing analysis of saline soil salinity is also a research hotspot.In this paper,for salinized arable land along the coast of the Yellow River Delta,the Kenli district of Shandong Province was used as the study area.Firstly,concentrated and continuous light and medium-severe saline land were selected as the test area according to the different salinity.Field soil samples were collected in mid to late April 2018 to determine the salt content(soil salinity content,SSC)of light(M)and medium-severe(S)soil samples.Near-ground multispectra images of unmanned aerial vehicle(UAV)in the test area and multispectra satellite images of Sentinel-2A MSI(MSI)in the study area were acquired;based on the UAV near-ground images of different test areas,the spectral parameters of soil salinity were screened by correlation analysis,and then the spectral parameters of varying salinity levels(light M,medium-heavy S and overall sample The inversion model of soil salinity of cropland with different salinity levels(mild M,medium-severe S and general sample I)was constructed to realize the inversion of soil salinity distribution of cropland in other salinity test areas;then the cropland areas with different salinity levels were roughly classified according to the normalized vegetation index,and the satellite spectral data were revised through the fusion of satellite and aircraft spectral data to construct the fused satellite spectral index,which was substituted into the inversion model of soil salinity of UAV in different salinity test areas,and then applied to the corresponding The quantitative inversion of soil salinity of regional cropland was achieved by using the satellite images to the corresponding cropland,and the specific research contents and results are as follows.(1)Exploring the spectral characteristics and parameters of salinity in arable land with different salinityBased on the UAV near-ground images of different salinity areas,this study used correlation analysis to screen soil salinity spectral characteristics and parameters at different salinity levels.The most robust spectral response was found for medium and heavy saline soils for different salinity levels,followed by light saline soils and the weakest for the overall group.Therefore,the degree of the spectral response of the grouped samples(M,S)was better than that of the general sample(I).(2)This paper clarifies the inverse model and spatial distribution of salinity of cultivated land in the test area with different salinityBased on the UAV near-ground images and MSI images,the SSC models of the three sets of samples were established using the Multiple stepwise linear regression(MSLR)method with the screened spectral parameters as independent variables.The R~2of the modeling set based on the UAV model was higher than 0.625,and the R~2of the validation set was more elevated than 0.645,while the R~2of the modeling set based on the MSI model was higher than 0.542,and the R~2of the validation set was higher than 0.561.Although the UAV and MSI soil salinity models for all three samples met the application requirements,the model accuracy was highest for group S,followed by group M,and lowest for group I.This indicates that the accuracy is based on different models.The grouped models based on different salinity have better accuracy and were selected as the soil salinity inversion models for other experimental areas.Therefore,modeling based on different salinity can improve the accuracy of soil salinity inversion to a certain extent.The soil salinity inversion based on the soil salinity model showed that the soil salinity in area M ranged from 0.136-5.93 g·kg-1 with a mean value of 2.158 g·kg-1,and salinization was standard and low in this area.The soil salinity in area S ranged from 0.323-21.210 g·kg-1 with a mean value of 6.871 g·kg-1,and the overall salinity content was high.Both groups were consistent with the field survey results.It is compatible with the field survey results.(3)Determine the salinity inversion model and spatial distribution of salinized arable land in the study areaBased on the relationship between normalized difference vegetation index(NDVI)and SSC,this paper first classified the cropland in the study area into mild and moderate salinization zones.Then the fusion of spectral data from UAV near-ground images and MSI images with different salinity was performed.The UAV SSC models with different salinity were substituted into the fused MSI images of lightly and moderately saline cropland and then combined to obtain the SSC inversion map of cropland in the study area.1 The UAV SSC models of the whole group were then substituted into the fused images of cropland in the study area to obtain the SSC inversion map of cropland in the study area 2.The two sets of inversion maps were compared and analyzed.The study showed that the R~2 and RMSE of the fused salinity inversion values of the study area cropland based on the grouping model were0.712 and 8.659,with absolute values of residuals ranging from 0.045 to 4.587.The R~2improved by 0.058,and the RMSE decreased by 0.561 compared with the fused image of the whole group model substitution.Therefore,group modeling based on different salinities can improve the accuracy of regional soil salinity inversions.The spatial distribution of soil salinity of cultivated land in the study area showed increasing from the southwestern region to the northeastern part,with light saline and non-saline soils mainly distributed in the southwest region,heavy saline and saline soils in the northeastern coastal area,and light and moderately saline soils in the central region,which was consistent with the field survey results.This paper thus shows that the fusion of UAV and MSI spectral data based on different salinities can improve the accuracy of soil salinity inversion in arable land and obtain better regional inversion results. |
PDF Full Text Request