A Study On Methods Of Remote Sensing Survey And Performance Evaluation Of Crop Spatial Distribution In Tianjin City

Recognition of major types of crop is the fundamental work for estimating its distribution area,monitoring its growing condition,and forecasting its productivity,and is the foundation of agricultural remote sensing.Thus,timely and accurately acquiring its types,growth,and spatial distribution at large scale can provide relevant information for food production and agricultural planning.It has a great significance on the agricultural production management and governmental food policy decision-making.With the rapid development of remote sensing technology and the continue increasing of data sources,spectral and spatial resolution of remotely sensed image data are increasing,and the application of remote sensing in agriculture is deepening and refining,make it possible to obtain crop area and spatial distribution quickly and accurately.In this essay,Tianjin was used as a study area,the GF-1 WFV was used as the remotely sensed data sources,and the image data,acquired during the critical growing periods of the main crops in 2016,was combined with the ground survey data,using the technologies of RS,GIS,and etc.The decision tree classification method was used to extract the spatial distribution data of winter wheat,rice,corn in Tianjin City.The main achievements of this research are as follows:(1)The image data that cover the whole study area were obtained by orthorectification,radiometric calibration,atmospheric correction,mosaic,and cutting of the remotely sensed data acquired during the critical growing periods of main crops.(2)The vegetation index(NDVI)and water index(NDWI)were extracted and constructed on the basis of the remote sensing images for each phase,and combined with the ground survey data.The seasonal variation curves of the NDVI and NDWI of the main crops,such as winter wheat,rice,and corn,were plotted.(3)Based on the phonological data of main crops in Tianjin,the seasonal variation of the NDVI and NDWI of winter wheat,rice and maize were analyzed,and the classification indices and classification criteria of common objects in the study area were defined,the time nodes that differentiate different types of crops were determined,and the appropriate thresholds was selected through repeat tests.(4)The decision tree classification models for summer harvest-crops---winter wheat and autumn harvest-crops---rice and maize were established in the ENVI 5.1 software platform,and the spatial distribution of these main crops in Tianjin was revealed.(5)According to the agricultural statistical data and ground survey data of the crop spatial distribution in the study area,the accuracy of remote sensing survey result of the crop spatial distribution in different seasons was statistically analyzed using the confusion matrix method.The statistical analysis results show that(a)the overall accuracy of the summer crops is 95% and Kappa coefficient is 0.89 with respect to 95% and 91% of the producer’s accuracy and user’s accuracy of winter wheat,and(b)the overall accuracy of autumn crops is 90% and Kappa coefficient is 0.82 with respect to 86% and 85% of the producer’s accuracy and user’s accuracy of rice as well as 93% and 87% of the producer’s accuracy and user’s accuracy of corn.It can be concluded that the prediction accuracy of the remote sensing survey method for crop spatial distribution described in this paper satisfies the precision requirement of the Third National Agricultural Census and it is a feasible remote sensing survey method for crop spatial distribution.