Response Relationship Study On Soil Organic Matter And Related Inlfuencing Factors In Ebinur Lake Basin

Soil organic matter (SOM) is an important source and sink of carbon cycling interrestrial ecosystems, which is an important attribute of the soil, it can improve soilfertility and productivity and fixed carbon response to elevated atmospheric CO2concentration. SOM content is an important basis for evaluation of ecologicalrestoration effect, to curb carbon emissions and maintain soil quality. And it is plays akey role in desertification, grassland degradation and regional ecological environmentdeteriorating in environmental issues of wide public concern. In view of theimportance of soil organic matter, it is very necessary to study the spatial distributioncharacteristics and response relationship between soil organic matter and influencingfactors.Currently, SOM and its influencing factors analysis is limited to the classicalstatistical analysis tools, failed to soil organic matter and many influencing factors inthe multivariate statistical analysis framework, and revealed that response relationshipis linear and smooth space. It will provide refence and studies for other regions.However, a large number of studies have shown that soil organic matter and itsinfluencing factors are generally spatial variability, thus the response relationship ofsoil organic matter and soil factor may also have spatial variability, which is spatialnon-stationary and nonlinear. Although the individual reports in the literature relatedoutcomes, it is the lack of systematic research.Ebinur Lake Basin soil organic matter content and soil texture type, vegetationcommunity types, the depth of the soil profile, soil pH, soil conductivity and soilheavy metal content and other factors response relationship of nonlinear and spatialnon-stationary would be included in multivariate statistical analysis framework. Firstly, I have analyzed the distribution of the horizontal and vertical space variabilityfor the Ebinur Lake watershed soil organic matter; Secondly, multiple linearregression model and spatial expand model have been made between soil organicmatter and soil factors, aimed to analysis the linear response and spatial variationcharacteristics of soil organic matter and soil factors with multivariate statisticalanalysis framework. Then, it is a purpose to explore the nonlinear responserelationship of soil organic matter and soil factors, and established the additive model;Finally, I have applied the spatially varying coefficient model to study spatialnon-stationary.The main conclusions are as follows:(1)The results show spatial variability of SOM content exhibits differentcharacteristics with changes in soil depth; SOM display a patchy pattern in the1–80cm soil layer in the Ebinur Lake Basin. However, the SOM content changes arerelatively continuous in the81–120cm soil layer, and the SOM content is higher in theeastern and western parts of the region than in the central region of the Ebinur LakeBasin. The spatial variation patterns satisfy the hole model in the1–80cm soil layer,while the exponent model is fitted well in the81–120cm soil layer in thesemi-variogram. Also, the results of the EOF analysis illustrate the vertical spatialdistribution of SOM shows different characteristics with the soil depth across theEbinur Lake Basin. The SOM content decreases as the soil depth increases in thecentral part of the Basin. However, the SOM content displays an increasing trend withincreasing soil depth in the eastern and western regions of the Ebinur Lake Basin. Inthe Ebinur Lake Basin, the patterns of spatial variation in the SOM are extremelysignificantly different between the shallow soil layers when compared to the deepersoil layers.(2)The soil pH, conductivity and SOM are linear negatively correlated, and theheavy metal content and SOM are linear positively correlated, while the degree of correlation is varying across soil depths. The results of additive model analysis showthat there are nonlinear relationship between the SOM content and soil factors, andthe nonlinear patterns are dramatic changes in different soil depths. The regressionrelationships of SOM and soil factors are spatial non-stationary and the patterns arevarying across the soil depths.(3)The global and local diagnostic results of the model residual indicate thatlinear regression model fitting residual with significant nonlinearity and spatialcorrelation. The fitting residual of Additive model residual was not significantnonlinear; and the fitting residual spatial correlation of spatially expand model andspatially varying coefficient model was not significant. Multiple linear regressionmodel are less effective, additive model was able to effectively explore the nonlinearresponse relationship between the variables; spatially varying coefficient model canfully reveal the spatial variability of response relationship; In addition, spatiallyvarying coefficient model based on local linear fitting method can effectively analysisthe local nonlinear response relationship of the variables.(4)This characteristic is strongly relevant to the patches of the arid oasisecological landscape and the features of the vegetation rhizodeposition. Also, thespatial heterogeneity during evolution of the soil profiles in the Basin suggests asignificant restraining effect on vertical spatial variability of the SOM with soil depth.Papers have some innovations in analytical methods and modeling:(1) The correlation and variability analysis method of vertical distribution of thesoil elements proposed vertical lag correlation coefficient to measure the strength ofthe soil elements along the soil depth vertical direction correlation.(2)Additive model and spatially varying coefficient model based on localregression techniques between SOM and soil factor objected to analyze nonlinear andspatial non-stationary of the regression relationship.(3)Local regression analysis method of spatial data analysis model residuals had been used to diagnosis trend and nonlinear of the residuals, the paper applied locallinear method to fit nonparametric regression model of the residuals, and based on theresidual series of smooth curves had been used to evaluate the analysis effect of thenonlinear response relations on linear regression model, spatially expand model,additive model and spatially varying coefficient model between soil organic matterand soil factor in the Ebinur Lake Basin.