Study On Estimation Methods Of Pear Area Based On GF-1 And Landsat8 Data In Korla City

With the development of science and technology,how to timely and accurate obtain to information on crop is very important for modern agricultural production.Xinjiang pear is one of the characteristics of fruit,whose planting is mainly distributed in southern Xinjiang,and Korla is the main cultivation area with the largest production.The use of remote sensing technology to obtain the planting area,growth and other information of Korla pear,assist government management to provide data support.This research untilized GF-1 Apr-Oct and LADNSAT8 Jul-Oct of 2015,the remote sensing image filmed in Korla,using the domestic GF-1 satellite and Landsat8 satellite remote sensing data,make mask data through the GF-1 satellite remote sensing(2 m).We use ENVI5.2 identified best identification month,method and data through five supervised classification(maximum Likelihood,Parallelepiped Classifier,minimum Distance,Artificial Neural Networks,Support Vector machine).Get the best identification month,method and optimal data source of Korla pear cultivation area in 2015.Results show:(1)The overall accuracy of the Landsat8 image data is higher than the GF-1 image data,but the verification accuracy of the GF-1 satellite image data is higher than the Landsat8 satellite image data.(2)Based on the multi temporal remote sensing data,the recognition accuracy of the maximum likelihood method and the minimum distance method of GF-1 images in August is better.The maximum likelihood method of Landsat8 satellite image is October,the minimum distance method is high recognition accuracy in July.Overall analysis,GF-1 satellite image maximum likelihood classification is highest accuracy in August;Landsat8 satellite images for the maximum likelihood supervised classification of the highest accuracy in October,the analysis of the maximum likelihood method is the identification of pear planting area of the highest accuracy of the supervised classification method.(3)Use of GF-1 August 16 m satellite image data,through maximum likelihood supervised classification 2015 korla fragrant pear planting area of 2.827×104hm2,fragrant pear area extraction accuracy is 96.28%.(4)If the aim of the study is to interpret the distribution of crop space,use of the Landsat8 image data with high radiation resolution;If the aim of the study is to identify the area of the crop,use the high resolution GF-1 satellite image data to improve the interpretation accuracy.