Short-interval land use change detection

Abstract

As the fast economic development and urban expansion,it is difficult for traditional methods to monitor land use changes in short time interval.Moreover,remotely sensed data acquired by optical sensors is often limited by bad weathers and cloud cover.SAR images,such as Radarsat-1,are an ideal tool for weather-proof observation on ground surface.This paper analyzed the results of land use change detections with time lags of 24 days,48 days and 72 days according to the period of acquisition dates of Radarsat-1. The results need a compromise between accuracies and efficiencies related to the time lags. For most of the situation, it is sufficient of using a time lag of 24 days to obtain accuracy of 60% or above. In some cases of months, a time lag of 48 days is needed. For obtaining higher accuracies, longer time lage such as 72 days are needed.

傳統的土地利用變化監測方法已無法完全滿足當前經濟的快速發展和城市急劇擴張的要求,利用光學遙感數據對地觀測也受到天氣和云層覆蓋等諸多因素的限制。而Radarsat-1等雷達遙感數據可以在'全天候'的條件下對地表進行觀測。論文以Radarsat-1圖像獲取的24 d時間周期為最短時間間隔,分析了24 d、48 d和72 d時間間隔下的土地利用變化監測,對短時間間隔內的土地利用變化監測的結果進行了總結。監測的結果需要在精度與時間間隔之間做出取舍。對于大部分的情況,24 d的時間間隔已可以得到60%以上精度的結果,部分月份的監測需要48 d的監測時間間隔。若需要更高的精度,則需要72 d甚至更長。