Estimation of rice field area using two-date Landsat TM images in Korea

To estimate the area of rice crop fields in central part of Korean peninsula, two sets of Landsat TM data were obtained during the growing season of May 31, 1991 (transplanting stage) and August 19, 1991 (heading stage). Four different methods were applied: (1) comparison of vegetation indices, (2) maximum likelihood classification, (3) unsupervised classification based on ISODATA algorithm, and (4) the same unsupervised classification with different number of clusters. For the classification methods, ten bands of image data that include TM band 3, 4, 5, RVI, and wetness were extracted from each data set. Accuracy of area estimating was assessed by using topographic maps and field collected data using GPS. All four methods were very effective for extracting the area of rice fields using two-dates TM data with accuracy ranging from 82% to 95%. In particular, the simple comparison of vegetation indices and DN values was relatively easy and fast to process the digital data to estimate the rice crop fields as compared to the classification methods.