Response Of Soil Quality Degradation To Land Use/Cover Changes In The Yellow River Delta

Land Use/Cover Change is a hot topic in global change research.It is one of the core contents of resource and environment dynamic research.At present,the degradation of land and land resources is a serious problem for human survival and development.Soil quality has become the focus of attention all over the world.The Yellow River delta is a region with the largest reserve of land resources and the greatest potential for exploitation in the eastern coast of China.It is also a typical ecological fragile area,and the soil salinization is common.It is of great significance to explore the degradation characteristics of soil quality in this area and to find out the effect of land use on soil quality degradation,which is of great significance to the utilization of resources and the management of ecological environment protection.Taking the Kenli District of Dongying city of the Yellow River Delta as the research area,this paper takes the Landsat satellite remote sensing images in 2007 and 2013 and the soil surface sampling data of 0~20 cm as the main data sources to study the remote sensing extraction method of land use / cover information in this area and monitor its spatiotemporal dynamic changes.5 models of multiple linear regression,partial least squares regression(PLSR),BP neural network,support vector machine(SVM)and random forest(RF)were established by using Landsat8 OLI multispectral image,HJ-1A HSI hyperspectral image and the fusion image in 2013.The remote sensing inversion method of soil salt and alkali degradation information was explored.On this basis,the main indexes of soil fertility quality are systematically analyzed by GIS spatial interpolation and geostatistics,and the spatial and temporal distribution and variation characteristics are explored.Then,the soil quality evaluation indexes were constructed.Under the support of GIS,the soil quality was evaluated with the comprehensive index method,and the soil quality degradation in central and western region of the study area was quantitatively evaluated.Finally,the response of soil quality degradation to Land Use/Cover Change is systematically analyzed.The main results are as follows:1)Remote sensing monitoring of Land Use / Cover ChangeThe remote sensing images of two groups of 2007 and 2013 two phase remote sensing data are used to classify the remote sensing images.The results show that the imaging time of remote sensing images has influence on the classification results.Because the spectral of the study area is complex,the remote sensing image of early spring should be used as much as possible to distinguish different classes.The result of machine learning method classification is better than the traditional maximum likelihood method.As far as machine learning methods are concerned,the support vector machine classification is better than neural network classification.Compared with the two time phase,the main Land Use/ Cover Changes in the area are as follows: forest grassland to saline alkali land,saline alkali land to water,saline alkali land to forest grassland,forest grassland to cultivated land and saline alkali land to construction land.The main land use / cover changes in the target area are: saline alkali land to water area,forest grassland to saline alkali land,cultivated land to saline alkali land,saline alkali land to cultivated land,saline alkali land to construction land.2)Soil quality degradation assessment and remote sensing retrieval(1)On the inversion of soil salinity by fusion of multispectral remote sensing images and hyperspectral remote sensing images.The Landsat8 OLI multispectral image and the HJ-1A HSI hyperspectral image are fused.On the basis of the classical statistical analysis model,typical algorithms of machine learning are used to establish the retrieval model,which have obtained good results.The research shows that the fusion image has high spatial resolution of multispectral image and hyperspectral resolution of hyperspectral image,and the inversion accuracy is better than the hyperspectral image,and the precision of hyperspectral image inversion is better than the multi spectral image.The machine learning model is better than the statistical analysis model,which can better simulate the complex nonlinear relationship between the soil salt and the image characteristics,and the appropriate data preprocessing can help to improve the retrieval accuracy.(2)On the spatial and temporal analysis of soil quality index.Six soil quality indicators,including alkali hydrolyzed nitrogen(AN),available phosphorus(AP),available potassium(AK),organic matter,pH value and salt content,were selected for evaluation of soil fertility quality.The analysis showed that the soil AN in the study area was normal distribution in 2007,AP and AK were lognormal distribution,and all had medium intensity spatial variability and medium intensity spatial autocorrelation.The range of variation,the coefficient of variation and the base effect are AP,AK and AN from large to small.The exponential and spherical models are better than other variogrammodels.Moran's I is a robust and effective way to measure spatial autocorrelation of soil nutrients.The high contents of AN in the study area mainly concentrated in the cultivated land in the southwest,the paddy field in the southern part and cultivated land in the northern area.The high content of AP mainly concentrated in the cultivated land in the southwest and the northern cultivated land.The high content of AK mainly concentrated in the northern,southwestern and yellow areas.The content of SOM in the study area is low,but pH and soil salinity are higher.Most of the soils are alkaline and salinization is serious.Although the field sampling data in 2013 is less than the first phase,the spatial distribution law can be effectively revealed by the IDW method.The two phases of data are comparable.The comparison of the two periods showed that the content of AN in the study area decreased obviously,the content of AP decreased significantly,the content of AK increased significantly,the content of SOM decreased slightly,the pH value increased,and the salt content in the soil increased significantly.Based on the comprehensive analysis of soil quality indicators,it is preliminarily judged that the quality of cultivated land in the study area has a general downward trend;especially the degree of salinization is aggravating.At the same time,the availability of the spatial interpolation method is explored.Based on AN data,the data sets of three different spatial distribution patterns are designed,which are discrete,random and aggregated.Using the automatic optimization of each model,the selfadaptability of different interpolation methods in soil nutrient space prediction is compared and analyzed,and its precision is explored.The results showed that sample size,spatial distribution pattern,spatial autocorrelation and spatial aggregation degree all affect interpolation accuracy.(3)On the evaluation of soil quality and assessment of degradation.Comprehensive index method was used to evaluate the soil quality of two phase data,and on this basis,soil quality degradation assessment was carried out in typical areas of central and Western Kenli.The results showed that from 2007 to 2013,the soil quality of the study area degraded significantly.The area of soil quality degradation is mainly located in the part of Xing Long Street,Kenli street,Yongan town and Huanghekou Town.The area of soil quality improvement is very few and scattered in southwest Haojia Town,Dongji Town,Shengtuo town and Kenli street.Other areas are basically unchanged.3)The response of soil quality degradation to Land Use/Cover ChangeThe relationship between land use / cover change and soil quality degradation was quantitatively analyzed and the relationship between different LUCC types and soil quality changes was discussed.The results showed that the corresponding LUCC types in the degraded areas were mainly saline alkali land to cultivated land,cultivated land to saline alkali land,forest and grassland to cultivated land,forest and grassland to saline alkali land,cultivated land to forest and grassland.In the target area,land reclamation,forest and grassland destruction,cultivated land destruction and abandon,are important factors leading to the degradation of soil quality.Then,the relationship between the 5 LUCC types and the soil quality change grade was analyzed by modeling,and the fitting degree of the functions was generally high.The contribution rates of 2 conversion types of cultivated land to saline alkali land,forest and grassland to saline alkali land have negative correlation with the change of soil quality.The greater the contribution,the higher the degree of soil quality degradation is.The other 3 LUCC types,which are conversion of saline alkali land to cultivated land,forest and grassland to cultivated land,cultivated land to forest and grassland;all have complex nonlinear relationship with the grade of soil quality,which is a parabolic function.When they resulted in soil degradation,the greater the contributions of the LUCC types are,the greater the degradation of soil quality.When they improved soil quality,the greater the contribution of the LUCC types,the more likely soil quality will be improved.To sum up,this paper uses remote sensing and GIS technology to study the Land Use/Cover Change process and soil quality degradation of central and western region in Kenli District of the Yellow River Delta,and explores soil quality degradation and its response to Land Use/Cover Change,quantitatively analyzes the relationship between the two,provides the basis for the rational planning and utilization of land in the Yellow River Delta,and will promote the sustainable development of the area.