Estimating total area of paddy fields in Heilongjiang, China, around 2000 using Landsat thematic mapper/enhanced thematic mapper plus data

Agricultural statistics are a fundamental reference for evaluating damage caused by natural disasters, estimating food supply and demand, and framing policies. A statistical table is usually prepared by an administrative district. Unfortunately, the Heilongjiang Statistical Yearbook of China was not completely prepared by such a district. Therefore, remote sensing technology is necessary for estimating the total area of agricultural lands in each administrative district. The test area is the Heilongjiang Province in China. The Landsat Thematic Mapper (TM) and Enhanced Thematic Mapper Plus (ETM+) data acquired during and immediately after the rice-planting season around 2000 (1999-2002) were used for the land-use/land-cover classification. All possible data during or immediately after the rice-planting season (from the beginning of June to the beginning of July) were selected so that paddy fields could be detected accurately. Borders of prefecture-level cities were generated using borders of cities and prefectures derived from the Digital Map Database of China 1:1,000,000 International Version. The TM/ETM+ data with bands 3, 4 and 5 (red, near infrared and middle infrared, respectively) were prepared for the land-use/land-cover classification. These data were classified by the unsupervised method Iterative Self-Organizing Data Analysis Technique. Land-cover classes were identified and reclassified into 12 classes, i.e., submerged paddy fields, overgrown paddy fields, bare dry cropland, overgrown dry cropland, bare ground, grassland, woodland, wetland, water, built-up, shade and shadow, and clouds. Some scenes including clearly misclassified paddy fields were image-processed or reclassified to reduce misclassification. The accuracy of detecting paddy fields was estimated to be in the range of 86.2-94.6%. The area of paddy fields in Heilongjiang Province was estimated to be 19.4x10(3) km(2) and overestimated by 17.7% for the Heilongjiang Statistical Yearbook.