HR: 17:30h
AN: B24B-07    [Abstracts]
TI: Cropland Area Extraction in China with Multi-Temporal MODIS Data
AU: BAGAN, H
EM: hasi@iis.u-tokyo.ac.jp
AF: Institute of Industrial Science, The University of Tokyo, Komaba 4-6-1, Meguro-ku, Tokyo, 153-8505, Japan
AU: * Baruah, P J
EM: pjbaruah@iis.u-tokyo.ac.jp
AF: Institute of Industrial Science, The University of Tokyo, Komaba 4-6-1, Meguro-ku, Tokyo, 153-8505, Japan
AU: Wang, Q
EM: wangqx@nies.go.jp
AF: Asian Environmental Research Group, National Institute for Environmental Studies, 16-2 Onogawa, Tsukuba-City, Ibaraki, 305-8506 Japan., Tsukuba, 305-8506, Japan
AU: Yasuoka, Y
EM: yyasuoka@iis.u-tokyo.ac.jp
AF: Institute of Industrial Science, The University of Tokyo, Komaba 4-6-1, Meguro-ku, Tokyo, 153-8505, Japan
AB: Abstract: extracting the area of cropland in China is very important for agricultural management, land degradation and ecosystem assessment. In this study we investigate the potential and the methodology for the cropland area extraction using multi-temporal MODIS EVI data and some ancillary data. A 16-day composite EVI time-series data for 2003 (6 March 2003 - 2 December 2003) with a spatial resolution of 500 m, and the ancillary data included Land-use GIS data, Landsat TM/ETM, ASTER data, and county-level cultivated land statistical data of year 2000. The Self-Organizing Map (SOM) neural network classification algorithm was applied to the EVI data set. To focus on agricultural and desertification, we designed 9 land-cover types: 1) water, 2) woodland, 3) grassland, 4) dry cropland, 5) sandy, 6) paddy, 7) wetland, 8) urban/bare, and 9) snow/ice. The overall classification accuracy was 85% with a kappa coefficient of 0.84. The EVI data sets were sensitive and performed well in distinguishing the majority of land cover types. We also used county-level cultivated land statistical data from the year 2000 to evaluate the accuracy of the agricultural area from classification results, and found that the correlation coefficient was high in most counties. The result of this study shows that the methodology used in this study is, in general, feasible for cropland extraction in China. Keywords: MODIS, EVI, SOM, Cropland, land cover.
DE: 0402 Agricultural systems
DE: 0430 Computational methods and data processing
DE: 0480 Remote sensing
DE: 0540 Image processing
DE: 0555 Neural networks, fuzzy logic, machine learning
SC: Biogeosciences [B]
MN: 2007 Fall Meeting