Estimation Of Cultivated Land Area In Taishan City, Guangdong Province With Remote Sensing Data

Cultivated land is the basis for development of agriculture. So it is necessary to accurately grasp the number and distribution of cultivated land for carrying out the macro-management of agriculture. Taishan City is in economically developed areas of Pearl River Delta in Guangdong Province, although the cultivated land resource is limited, but modern agriculture is developed. It is the main agricultural supply base for Pearl River Delta and Hong Kong and Macao. In recent 10 years, the change of cultivated land is larger with the economic development, especially urbanization. The traditional method of cultivated land is mainly for ways to get through statistical reporting area of the township's arable land changes. By a variety of reasons, especially infrastructure of land, cultivated land area statistical data is inaccurate. How to accurately inform of the number of cultivated area and spatial distribution has become the urgent need. Earth observation remote sensing technology in the surface area measurement accuracy has advantages. That effective using remote sensing to measure cultivated land can confirm the accuracy of reported data, which can provide another independent data of number and spatial distribution of land resources for the local agricultural departments.The main research goal is to chart cultivated land map using remote sensing data, calculate exact amount of cultivated land area, draw the spatial distribution of cultivated land, analyze spatial distribution of cultivated land in Taishan City, and provide reference for the local agriculture sector. At the same time, explore the using feasibility of accurately measuring and calculation of cultivated land with remote sensing data, provide information for remote sensing regions mapping in similar regions.In this paper, TM data is the main data source and the ground reference information is the secondary research data. The pattern recognition PNN model is introduced to the classification of remote sensing images against the complex feature type and low classification for the Taishan region. The different surface feature use different discriminat function, based on multi-scale segmentation, combined with Object-oriented thinking, and to establish PNN model. By the first PNN operation, the water, woodland and shrubs were removed from the image. The cultivated land is get from the remaining image information through the second PNN operations.Then check on the correction information extracted, modified classification image area which is too small or the wrong or the missed points, thematic maps of land distribution of Taishan City can be get.Meanwhile, this paper tests and verifies the feasibility in calculation of cultivated land with remote sensing technology. The data by improved PNN algorithm is compared with the data by visual interpretation, the maximum likelihood classification and ISODATA algorithm. The result shows that the overall accuracy by improved PNN algorithm is 88.77%, and the accuracy of cultivated land is 89.37%. This verifies the feasibility of the method. Therefore, we can believe that the regional land mapping with remote sensing technology has a very wide range of applications. Land remote sensing mapping results in Taishan City showed that the cultivated land area in Taishan City is 91.79 million mus, in line with local conditions.The main innovation of this study is to identify of the cultivated land number and spatial distribution with remote sensing technology, and discuss the feasibility in the calculation of cultivated land with remote sensing data.