Flood Disaster Based On Vegetation Index Time Series Remote Sensing Monitoring Model

Heilongjiang province,as a major agricultural province,is Chinese important grain output base.Because of its unique geographical location and climate factors and soil types,drought,floods,hail,wind and other agricultural disasters occur frequently.In recent years,due to the deteriorating climatic conditions,more and more extreme weather emerged,and the incidence of agricultural disasters increased significantly,which had a significant impact on the agricultural development of the province.Flood disaster has become one of the most frequent agricultural disasters in Heilongjiang province and the severed flooding has a great impact on the cultivated land and the crops.Therefore,timely and accurate monitoring of agricultural flood disasters and analysis of the spatial and temporal characteristics of agricultural flood disasters in Heilongjiang province can not only guarantee food security,but also provide reliable data and theoretical support for pre-disaster early warning,disaster monitoring,post-disaster rescue and disaster damage assessment.Traditional flood disaster monitoring methods are mainly based on field investigation and sampling,which will consume a lot of manpower and material resources,and has some certain subjective factors.The assessment of disaster information is not accurate,and the acquired information is seriously lagging behind.With the emergence of remote sensing technology,it is possible to monitor large-scale disasters timely and accurately,which is more suitable for agricultural disaster monitoring.This article selects three typical area of Heilongjiang province to build Sanjiang,Jiusan and Daqing to the research area.Using the 16 days synthetic of 250-meter spatial resolution MODIS vegetation index data,the vegetation index time series model is constructed.The disaster discrimination threshold is determined based on the time series variation law of the vegetation index.And extracting the scope of agricultural disasters,and verify the extraction results with high-resolution image data and ground measured data and insurance company reporting data.Finally,the time series curve is used to detect the growth period and extract the flood disaster range.And analyze the characteristics and causes of the spatiotemporal pattern of flood disasters in Heilongjiang province in 2016.The main conclusions are as follows:(1)The precision of disaster range extraction based on time series of vegetation index is high.By extracting typical research area disaster area after SG filter,the median vegetation index(the median vegetation index time series curve)time series curve,as the area of crops standard growth curve,and looks like the actual curve contrast,building disaster monitoring vegetation index time series model(SAUC).10 selected as the best threshold,the extracting range of disasters,and verifiesthe accuracy of the results.In 16 times the verification results,the average relative error is 13.26%,verify the accuracy reached 86.74%,and the practicability is strong.The model used in Heilongjiang province agricultural disaster monitoring,the extraction accuracy reached 81.4%compared with the high resolution image,and the similarity was high,which could be applied to the dynamic monitoring of agricultural disasters in a wide range and for a long time.(2)Based on the analysis and discrimination of the vegetation index time series can effectively monitor flood disaster.Use the difference in time sequence of flood disasters and other types of disasters,it is found that crops will have a sudden shape on the curve after flood disasters.According to the basis of the growth period of mutation detection,extract the flood information.Compared with high-resolution simultaneous image data,the data has an extraction accuracy of81.7%.Among the 16 known disasters,all the flood disasters were extracted.Two of the remaining non-flood disaster types were classified as flood disasters,with a correct rate of 87.5%.The extraction accuracy of flood disaster was higher,which could be applied to the monitoring of agricultural flood disaster in practice.(3)The spatial and temporal distribution characteristics of flood disaster in Heilongjiang province are affected by climatic factors,topographic factors and soil factors.According to the scope of the extraction of the flood in Heilongjiang province in 2016,the results show that the flood disaster of 2016 in Heilongjiang province is mainly occurred in late June,mainly distributed in Jiamusi,Hegang,Shuangyashan,and Qiqihaer.Combining with climate data in 2016 and soil,terrain data,it found that the correlation between the three factors and floods is higher.The results show that the agricultural disaster monitoring based on MODIS vegetation index time series model can be timely and effective agricultural disaster monitoring and extract the affected scope.According to the regional disaster information can have a wide range of agricultural disaster information.According to the mutation detection of the vegetation sequence time series growth period,the flood disaster range can be extracted with high precision,which has applicability and practicability.It can be used as one of the basis for flood disaster monitoring and disaster assessment,and is provided by the meteorological department of Heilongjiang province.Flood information statistics provide more scientific and timely information for agricultural flood warnings.