Study On Remote Sensing Monitoring Method Of Multi-source Satellite For Rainstorm And Flood In Farmland

It is a gradual process for the impact of rainstorm disaster on the growth stage of farmland crops.Not only to obtain disaster of the spatial distribution information on the farmland,but also to obtain the time characteristic information of the disaster start and the impact of the disaster.The multi-source satellite remote sensing observation technology has the advantage of capturing the instantaneous state of the ground and depicting the process.The paper used Terra/MODIS,Landsat and Sentinel satellite observation data,developed multi-source satellite remote sensing observation data.In this paper,a method for extracting dynamic information from the multi-temporal NDVI based on satellite remote sensing observation is proposed.Combined with the satellite data of high spatial resolution,the spatial distribution of the more delicate rainstorm flood disaster is obtained.And the application and discussion of the method in Chaohu basin,which is the main experimental area of the 2016 rainstorm disaster.First,the area of farmland affected by flood.Based on conventional methods of flood monitoring,extraction of water information before and after disaster is carried out.And the GIS spatial analysis technology was used to obtain the disaster area.Extraction water information method based on Landsat8 data and Sentinel-2A data,which used the characteristic values of water information in the green band,red band,near-infrared bands,normalized difference water index NDWI,normalized difference vegetation index NDVI and modified water index MNDWI.Then different threshold values based on above characteristic values were selected for satellite data at different periods,and the classification decision tree method was used to extract water.The Sentinel-1A data obtain the flood disaster of distribution,which used the threshold method.MODIS NDVI data is to obtain the distribution of disaster affected areas by extracting the pixels which theirs NDVI difference value in continuous phase and continuously lower than the selected threshold.MODIS data has low spatial resolution,Combined with the results of other three kinds of high resolution satellite data extraction,more accurate flood disaster of distribution on farmland was obtained.Secondly,identification of the disaster beginning period and extraction of the duration based on multi-temporal MODIS NDVI.The analysis of the NDVI difference value shows that the NDVI value of 2016 affected by rainstorm is less than 0.15 compared with the same phase in 2015.This threshold of 0.15 was used to identify the beginning period when NDVI begins to decrease,and the duration which were below 0.15 values.Based on these characteristics,we can assess the degree of the impact of rainstorm and the recovery of crops affected by rainstorm.The main steps of the research method include abnormal data removal and interpolation processing,multi-temporal NDVI change detection and disaster dynamic information extraction to identify the beginning period and the duration.Finally,the identification of cultivation system based on multi-temporal NDVI.The unsupervised classification of multi-temporal NDVI data is carried out by using IsoData technology(Iterative Self-organizing Data Analysis),which obtain the land use type of the research area before and after the disaster.The changing information of the land use type after the disaster was obtained by comparison of the results of the unsupervised classification in the year of 2015,and verified the identification results of beginning period and the duration affected by rainstorm.The results show that the beginning period and the duration affected by rainstorm are identified by the changes characterized values of multi-temporal MODIS NDVI.Based on these information,areas of crop fields affected by rainstorm are obtained.On the other hand,combining the flooded lands extracted from Landsat and Sentinel data of 30 m and 10 m respectively,which provides more accurate areas affected by rainstorm even if they have only a few temporal data available.As a conclusion,multi-satellites remote sensing were as one of the measures assessing the influences of disaster,which can obtain dynamic information of disaster impacts on crop land which could provide scientific data supports for the recovery of farmland,the loss evaluation and relief policy after disaster.