Technology Study On Near-real-time Remote Sensing Agricultural Information Monitoring Based On Filed Parcel

With the development of the times and the progress of science and technology,the development of modern agriculture urgently needs the information of agricultural situation monitoring with high precision,near-real-time limitation and facing farm and other basic agricultural production units.Crop growth and agricultural disasters are the more important part of agricultural information and the more concerned content in agricultural production,among which agricultural disasters are the most common drought and flood disasters,so it is very important to monitor agricultural information quickly and accurately.Remote sensing technology,as one of the important technologies of agricultural information acquisition,has the advantages of high timeliness,fast speed and high accuracy.It can not only meet the needs of obtaining agricultural information,but also greatly save time cost and improve working efficiency.Most of the existing monitoring methods use satellite data with low spatial resolution,which not only has low accuracy,but also has poor timeliness,and the monitoring scope is too broad.In this paper,the methods of classification of cultivated land,near-realtime monitoring of agricultural conditions and the near-real-time monitoring platform of agricultural conditions under complex background conditions are studied,and the main results are as follows:(1)In view of the problem that the classification accuracy of cultivated land decreases due to complex background,A method for extracting cultivated land area with complex background was established,sentinel-2 satellite images,digital elevation model(DEM)models data and normalized difference vegetation index(NDVI)build multisource data,And using object-oriented segmentation method to eliminate the influence of mixed pixels on classification accuracy,Combined with cart decision tree method,To realize the accurate extraction of cultivated land information.The accuracy of this method is evaluated by using mixed samples obtained from field investigation and high resolution image interpretation,The results show that the overall classification accuracy of the proposed method is as high as 96.1%,kappa coefficient is0.94.Compare a single cart decision tree taxonomy,Overall classification accuracy increased by 25%,kappa coefficient increased by 0.54.The classification accuracy of cultivated land under complex background is improved.(2)In view of the lack of timeliness and accuracy of the existing monitoring methods,A near-real-time monitoring method was established.Based on sentinel-2satellite data with near-real-time ageing,Calculation of vegetation condition index(VCI),To realize the near-real-time monitoring of crop growth;Adoption of the normalized drought index normalized difference drought index(NDDI)and NDVI to obtain the degree of water stress and growth of crops,To realize the near-real-time monitoring of crop drought;Calculate the normalized water body index normalized difference water index(NDWI)before and after the disaster NDWI To realize the identification of floods,By comparing the difference NDVI time series curves in disaster years,This paper describes the situation of crop waterlogging disaster in real time.To verify this method,According to the actual situation,five plots of single cotton crop in Bole area of Xinjiang,five plots of single sugarcane crop in Wuming and Longan area of Guangxi,three adjacent major rivers in Wuming and Longan area of Guangxi and three plots of sugarcane planting in Longan County of Guangxi are used as experimental objects of growth monitoring,drought monitoring and flood monitoring,The results show that the results of crop growth,drought and waterlogging are obtained by this method,And verify that the monitoring results are aging,The results of near-real-time monitoring of the fastest 1 day were obtained.(3)Based on the remote sensing computing cloud platform GEE(google earth engine),the near-real-time agricultural monitoring platform is constructed,and the near-real-time agricultural monitoring method established in this paper is integrated.The real-time acquisition of satellite data,the automatic mask of cloud region,the automatic calculation of various indexes,human-computer interaction and visual display can quickly and accurately monitor crop growth,drought degree and flood disaster.The platform can provide timely crop growth information and guidance for farms and farmers,and provide scientific and effective basis and help in disaster reduction and relief of agricultural disasters.