Study On The Information Extraction Of Citrus Field In South Of Jiangxi Province And The Estimation Of Chlorophyll By Remote Sensing

Citrus is one of the important fruit crops in China,and plays an important role in the national agricultural product economy.Gannan region has unique conditions for citrus cultivation.As the main production area of citrus production in China,it is known as the "World Orange Township".As of 2019,the planting area of navel orange in Ganzhou has reached 1.62 million mu,with an output of 1.22 million tons and an output value of 12.9 billion yuan.It has become a local pillar and characteristic industry in Ganzhou and an important economic source for local farmers.However,in recent years,Gannan citrus has been repeatedly affected by extreme weather and plant diseases and insect pests.The planting area and growth conditions and their instability have affected the income of fruit farmers in the entire region.It has also resulted in the inability to accurately estimate output and large fluctuations in market prices.Therefore,remote sensing and GIS technology can monitor the planting area and growth status of citrus in Gannan in a large area.With the support of GIS and RS technology,this paper takes Chongyi County of Ganzhou City as the research area,and GF-1WFV and HJ-1A / B CCD and other highresolution data sources as data sources to form a time series image set and realize the county-based Rapid extraction and analysis of large citrus planting areas,and on this basis,one of the sample areas was selected to analyze the relationship between the chlorophyll and the spectral curve of citrus leaves in the citrus orchard,in order to analyze the health and growth status of citrus.This move has an important role in understanding the spatial distribution and growth status of citrus cultivation in the study area and ecological monitoring,and provides a convenient way for the citrus industry to scientifically and accurately estimate yields,help farmers in the Soviet Union to get out of trouble,and make decisions by government departments and related enterprises.The research results are as follows:(1)Based on the regression analysis method,the conversion equation between the NDVI,DVI,and RVI indexes of GF-1WFV and HJ-1A / 1B CCD was constructed and verified.The results showed that the conversion accuracy of the three vegetation indexes between different sensors was More than 85%,the error is small.It shows that the conversion equation has high reliability and feasibility,which lays the foundation for the construction of multi-source time series image data sets.This method can also be applied to the conversion between other sensors.(2)In-depth excavation of phenology information and image texture information of citrus growth,combined with NDVI,DVI,RVI-based spectral information,to construct a 12-month time series image set during citrus growth period,and screened out through comparative analysis such as JM distance method The most suitable time node for citrus information extraction.Citrus growth is closely related to phenology,and reflects the differentiated characteristics in remote sensing images.In addition,citrus images have large artificial planting traces.When a single index or spectrum cannot be used to distinguish vegetation,combining phenology and texture can effectively distinguish.It can also avoid image redundancy,overcome the defect of less data,and reduce the workload.(3)Based on the combination of citrus phenology and texture information,CART decision tree classification and nearest neighbor classification method are used to extract citrus information.The accuracy is more than 75%,especially the former is more than 88%.It shows that CART decision tree is more suitable for extraction of citrus woodland,and the high score and environmental series data are based on their own advantages,and have a more potential data source for extracting large crops for market.(4)On the basis of object-oriented classification,collect measured spectrum data of citrus leaves,comprehensively analyze the correlation between various vegetation indexes and SPAD values,and then select an index with a high correlation with SPAD values to build a regression model equation,And comparative analysis,it is concluded that the combination of various indexes and parameters has a good predictive ability on chlorophyll,and the measured and predicted chlorophyll content are above 50,indicating that the citrus is in good growth condition and has not been affected by pests and diseases.The research laid the foundation for the monitoring of citrus growth status.