Research On Apple Oechards Information Extraction Of Fufeng County Based On Landsat8 Remote Sensing Image

In this study, we choose the Fufeng County of Baoji city in Shaanxi province as the study area, which located in the west of guanzhong plain. In this paper, we use the image of Landsat8 OLI as the basic data, and the digitized topography data and land use map as the auxiliary data. We are going to classify remote sensing image in the study area and extract the apple orchards information with the following several kinds of remote sensing image classification methods: Supervised classification, Decision tree classification, Object-oriented classification and Improved object-oriented classification. Finally, we will evaluate the accuracy of the results of the four types of classification in area and space by extracting area and confusion matrix respectively. The results as follows:(1) The best time for apple orchards information extraction in Guanzhong Area representing by Fufeng County is between the harvest of summer maize and germination of winter wheat. During this time, the farmland has no vegetation coverage in Guanzhong Area, but there are differences in reflectance characteristics of apple orchards, other orchards and woodland which have the similar growing period; Besides, the correspondence of the remote sensing image at returning green stage of winter wheat and the image(taken in 2014.3.15) used in the study can be the additional data of apple orchards information extraction.At this time, winter wheat, as the main crop of Guanzhong farmland, is at returning green stage, shows obvious green plants reflectance characteristics, while the apple orchards, other orchards and woodland are at germinating stage, shows great differences in reflectance characteristics compared with farmland, as well as a certain degree of differences among each other.(2) There are two parameters, one is the remote sensing image ratio of band 6 and band 5(band6/band5) which acquired on March 15, 2014, the other one is the RVI which extracted from the remote sensing image which acquired on October 25, 2014. We can use the two parameters as the basis to extract and identify apple orchards and other orchards in Fufeng County.(3)Between four classification results of Fufeng County based on different remote sensing classification methods, the area precision of supervised classification is 97.14% and the overall accuracy of spatial distribution is 92.58%. The area precision of object-oriented classification is 89.07%, its overall accuracy of spatial distribution reach up to 96.91%. The best method for extracting information of the apple orchards is the improved object-oriented classification method which area precision is 99.36% and has the overall accuracy of spatial distribution up to 97.33%. Besides, its Kappa coefficient is 0.9663. By contract, the decision tree method has the lowest area precision about 88.37%. The accuracy of spatial distribution is only arrive 90.36% what lead to the Kappa coefficient is as low as 0.8803. Above all, we believe that the improved object-oriented classification method will has practical significance in extracting information of the apple orchards and it can be a vital gist for apple industry planning of Fufeng County.(4) With all the samples selected by means of improved object oriented classification, we made this distribution map of apple orchards in Fufeng County. The statistics show that the total area of apple orchards of the study area of Fufeng County is 9901.328hm2 which accounts for 13.98% of the total area of this county.The apple orchards in Fufeng County are all located in 500 meters above the sea level、850 meters below the diluvial fan and the oess tableland area. Apple orchards distribute in every town of Fufeng County. Tiandu town ranks first with an area of 3366.775hm2 which accounts for 34% of the total apple orchards area of this county, and it accounts for 45.18% of the total area of this town. The apple industry is the leading industry in the town of agricultural industry. Jiangzhang town was in last place which covers only 37.139hm2, accounting for 0.38%, and it accounts for 0.44% of the total area of this town.