Study On Spatial Distribution Pattern Of Organic Carbon In Maize Farmland In Zhaodong City Based On Remote Sensing

In order to curb global warming and reduce carbon dioxide emissions,China has proposed the goal of carbon peaking and carbon neutralization.Farmland ecosystem is an important part of the global carbon pool.On the one hand,soil has a strong carbon fixation capacity,and on the other hand,crop straw itself contains a lot of organic carbon.Therefore,it is of great significance to scientifically study the spatial distribution pattern of organic carbon in farmland system for carbon sequestration and emission reduction in China.In this study,using remote sensing as the main technical means,Zhaodong City,Heilongjiang Province,was selected as the study area,and GF-6 WFV,GF-3,Sentinel-1,Sentinel-2 remote sensing images were used to extract the corn planting range in Zhaodong City;Calculation of carbon content in corn straw and drawing of spatial distribution map;Remote sensing inversion of soil organic matter and mapping of soil organic carbon spatial distribution.In the extraction of corn field planting range,five GF-3,GF-6 WFV images and Sentinel-1,Sentinel-2 images in five months during the growth period were selected for the experiment.A total of 12 extraction schemes were designed,and the experiment of corn field extraction was carried out using random forest algorithm.The 12 extraction schemes include: advantages of time series image;Advantages of SAR image;GF-3/6 and Sentienl-1/2 images were compared and discussed in three aspects.Through experiments,we can find that the extraction accuracy of multi time series images considering the whole growth period of corn is significantly higher than that of single growth period;SAR images can make up for the problem of missing optical images caused by cloud cover,and improve the accuracy of multi temporal image classification to a certain extent;By comparison,GF-3/6 image is superior to Sentinel-1/2 image in terms of its higher temporal resolution.In the research work of carbon content estimation and spatial distribution pattern of corn straw,the Hierarchical Linear Model(HLM)is adopted.First,the model is used to invert the corn yield.Secondly,the corn straw yield is estimated according to the straw index and corn yield.The conversion factor of corn straw to organic carbon is0.45,so as to calculate the carbon content of corn straw and its spatial distribution characteristics.The results showed that the carbon content of corn straw in Zhaodong City was more in the east than in the west.In the inversion of soil organic matter,Sentinel-1 and Sentinel-2 images in May were selected to carry out experiments based on three machine learning algorithms:random forest regression,BP neural network regression,support vector regression.Through the experiment,it can be found that the retrieval accuracy of soil organic matter is greatly affected by the external environment,and the accuracy is relatively low;The polarization characteristics of VH and VV in SAR images are sensitive to soil water response,which can weaken the influence of water on SOM retrieval accuracy to some extent;Compared with other algorithms,stochastic forest regression model has the highest retrieval accuracy;The soil organic carbon content of maize farmland in Zhaodong City showed a trend of less in the west and more in the east.Finally,based on the extracted corn field range,the organic carbon of corn straw and soil organic carbon of corn field were added to obtain the spatial distribution map of organic carbon of corn field in Zhaodong City.The organic carbon of maize field in Zhaodong City was more in the east and less in the west.