基於遙測技術於越南湄公河三角洲的稻作監測

Remote Sensing-based Rice Crop Monitoring in Mekong River Delta, Vietnam

稻米是世界僅次於玉米與小麥的第三大糧食作物，為三分之二世界人口的主要食物來源。以稻米產量估測所進行的稻區耕作與生長區域的監測，是稻米作物監測的重要作為。本研究主要目的是發展一套以MODIS影像應用稻米物候期資訊進行稻米耕作系統區域的分類方法。研究使用2002到2012年的資料，並依序進行三個主要研究步驟：（1）資料預處理；（2）稻區分類（3）精度評估。研究成果顯示分類結果與地面參考資料的總體精度高於80.6%，Kappa值高於0.74；比較政府統計資料，則R2皆大於0.8。然而，分類資料還是有略高的相對誤差，為0.9~13.6%。分析結果顯示，從2002到2012年三季稻的區域增加了17.8%，二季灌溉稻區域與二季雨季稻區域減少約6.%與13.9%。

Rice is the world’s third largest crop behind maize and wheat and the major food source of two-thirds of the world’s population. Monitoring rice cropping practices and growing areas for estimating crop production is an important activity for crop management. The main objective of this study to develop a mapping approach for delineating rice cropping systems Mekong River Delta (MRD) from MODIS data using phenological information of rice crop phenology. The data were processed for 2002 to 2012 following three main steps: (1) data pre-processing, (2) rice crop classification, and (3) accuracy assessment. The comparisons between the mapping results and ground reference data indicated overall accuracies and Kappa coefficients generally higher than 80.6% and 0.74. The mapping results and the government’s rice area statistics also examined, indicating the consistency between these two dataset (R2 ≥ 0.8). The results were, however, slightly overestimated with a relative error in area (REA) from 0.9%-13.6%. From 2002 to 2012, the area of triplecropped rice increased 17.8%, while those of the double-cropped irrigated rice and double-cropped rain-fed rice decreased roughly -6.1% and -13.9%, respectively.