利用GIS与TM数据资料集成技术估算中国南方早稻面积——以龙游县为例 (英文 )

Integration of GIS and TM Data in Extraction of Early Rice Planted Area of South of China: A Case Study of Longyou County

提出利用 GIS与 TM数据集成技术估算中国南方丘陵山地早稻种植面积的方法 .该方法首先利用 ARC/INFO对土地利用现状图进行数字化 ,建立拓朴关系后将其转化为栅格 ,然后进行投影变换 ,使土地利用现状图、行政图、TM数据具有相同的坐标 ,最后利用土地现状图 ,提取水田分布图 ,对水田分布图进行分类估算早稻种植面积 .不同方法比较结果表明 :非监督分类法不能用于提取丘陵山区的水稻种植面积 ;只用 TM资料估算龙游县早稻面积 ,与统计数据相比 ,平行六面体分类法、最大似然分类法分别达到 82 .83 %和 59.95% ;而用 GIS与 TM数据资料集成技术对水田分布图进行分类估算早稻面积 ,平行六面体分类法、最大似然分类法的估算精度分别达到 93 .98%和 60 .65% ,所以利用平行六面体分类法对南方丘陵山地早稻种植面积估算是可行的 .

This paper introduced the methodology of early rice planted area estimation by integration of GIS and TM data. The methodology enhanced the classification precision of TM image in both plain and mountainous areas. Land use and town boundary maps were digitized and transferred to raster format and opened in ENVI image analysis system. TM image, land use map and town boundary map were registered at the same projection. Utilizing the mask function in ENVI image analysis system, the study area, different land use types and towns TM images were established. The early rice planted area was calculated from Longyou TM image and paddy field TM image using supervised classification methods. Application of Maximum likelihood and Parallelepiped methods on TM image for early rice area estimation enabled the achievement of 59.95% and 82.83% classification precision, whereas applying the same methods on paddy field TM image enhanced the classification precision up to 60.65% and 93.98% respectively. Thus, in this study, Parallelepiped method applied to paddy field TM image was proposed as the best method for early rice area estimation in Longyou County. In addition the early rice information derived from this methodology can be used in any rice monitoring model or yield estimation studies with high accuracy and less susceptibility to error. The availability of GIS database makes the method easily applicable to new TM images. [WT5”HZ]