Prediction And Mapping Of Spatial Distribution Of Soil Organic Matter-study In Putian

As one of the main circle layers of the earth,the soil circle carries people’s production and life behaviors and is an important material basis for crop growth.In the 21 st century,the development of modern agriculture has placed new demands on the accurate and rapid understanding of the spatial distribution of soil organic matter content in the region.In May 2017,a total of 43 soil samples of soil surface(0-20cm)were collected in the Mulanxi River Basin in Putian.The carbon and nitrogen contents of the soil samples were determined.Combined with the Landsat 8 images of January 3,2017(Winter)and August 15,2017(Summer),the soil organic carbon,nitrogen content and spectral variables of the study area were discussed.The relationship between landscape factors and the correlation between spectral variables,landscape factors and soil organic carbon and nitrogen content were compared and analyzed.Referring to the current mainstream spatial distribution prediction method of soil organic matter,landscape theory,geostatistics,remote sensing and GIS technology are applied to the spatial distribution model prediction of soil organic carbon and nitrogen content in the study area,and the advantages and disadvantages of the models constructed by different methods are compared.It is of great significance to use the remote sensing image to predict the spatial distribution of soil organic matter content in the southern region,the sustainable development of cultivated land resources,and the effective management and rational planning of soil resources.The main conclusions are as follows:(1)The spatial heterogeneity of soil organic carbon and nitrogen content in the study area is obvious,and the prediction effect of cultivated land area is better: the non-cultivated land in the study area is mainly forest land,orchard,grassland,etc.,and the land is complex and vegetation.High coverage has a greater impact on the final modeling effect;(2)There are seasonal differences in predictive variable screening.The correlation between spectral variables and soil organic carbon and nitrogen content is more direct: the correlation between spectral variables,landscape factors and soil organic carbon is higher in winter images,and soil nitrogen The correlation of the content is higher in the summer image.Seasonal differences are mainly due to changes in surface cover in different seasons.(3)Scientific planning modeling samples and variables have an important impact on improving the model effect: In the modeling results of three methods based on OLS,GWR and GWRK,the GWR method considering spatial heterogeneity has a certain effect on the modeling effect.Improvement,especially when the modeling effect of the OLS method is poor,the effect of the model built based on the GWR method is significantly improved.The GWRK method has a good improvement on the phenomenon of over-prediction of the model.GWRK interpolates the residuals and cannot stably improve the final modeling results.After refining the modeling samples and scientifically planning the variables needed to build the model,the OLS method can also achieve good results.