Development of land use data product at provincial level based on HJ CCD data

Land use is critical for researches about carbon accounting, environmental protection, global climate change and sustainable development, etc. Recently, land use data production could not meet researches of land use and its change in administrative regions of province-level well under the rapid economic and natural disaster growth. This paper takes Guangdong province as an example, the land use classifications were produced from HJ CCD data and LandsatLook Images with supervised classification of Maximum Likelihood. Two results were compared with the second national land survey. And confusion matrix and precision of the land use result based on two different data were evaluated. The Kappa coefficient and classification accuracy (0.74 and 82.8% for HJ CCD data, 0.69 and 78.9% for LandsatLook Images) are not very high. The area of Guangdong province is about 180,000 square kilometers. The image data used in classification are in wintertime of 2013, which is a bit different from the product of second national land survey in 2010. The result is probably reasonable when taking the factors of research area and image season and land use map time into consideration. HJ CCD data have advantages over LandsatLook Images when they were used for classifying land use types. And HJ CCD data could obtain the change of land use more effective due to high time resolution and wide coverage by two CCD sensors than LandsatLook Images. For land use mapping of whole Guangdong province, four or fewer scenes of HJ CCD images were needed. But to LandsatLook Images, it needs 15 scenes of images or more. Besides, the inner consistency of HJ CCD data is better compared with LandsatLook Images because of wide cover. In the experiment, the time spans of the four HJ CCD images is seven days while the spans of 15 LandsatLook Images is 121 days for the coverage of whole Guangdong province. Land use could stay the same almost in seven days usually. However, in 121 days, there certainly be some great- change of land use happened with a vast variety of reasons.