以類別資料進行土地覆蓋時空變化熱點偵測

Abstract

隨著科技發展，如今人類能輕易改變地球上大部分地區之自然環境，以符合自身使用需求，但若僅專注於經濟發展而忽略環境保育，將會對全球的氣候條件和生態結構造成無法回復的破壞。為避免發生這樣的悲劇，政府與非政府的相關機構需要能有效監測土地利用和土地覆蓋變化，以達到有效的資源管理，並規劃或調適土地利用計畫。遙測資料是目前取得大量土地利用和土地覆蓋資料最快的方法，在多樣的遙測資料中，衛星影像因其時間解析度高與取得成本低的優勢，常被研究者用做土地覆蓋變化研究之資料來源。而在常見以衛星影像討論土地覆蓋變化的方法中，classification-based approach因能完整討論類別間變化關係，故為多數研究者採用之研究方法。然而，前人研究在以土地覆蓋類別資料討論土地覆蓋變化時，通常只比較時間序列中兩個相異時間點的土地覆蓋在數量上與空間位置上的差異，但這種方式無法完整利用多時期土地覆蓋資料中的時間序列特性。因此，本研究結合LICD與Mann-Kendall trend test，提出適用於類別資料的時空變化熱點偵測演算法，並以此演算法分析台灣宜蘭縣蘇澳鎮新馬都市計畫區2006年至2016年間的土地覆蓋變化。研究結果顯示，本研究提出的演算法能呈現出多時期土地覆蓋資料中，相異土地覆蓋類別的時空變化樣態，而此分析結果能在相關政府單位或學者進行資源管理或規劃土地利用計畫時，提供其有用的決策輔助資訊。With the development of science and technology, people can easily change the natural environment of most parts of the earth to what we want. However, if we blindly pursue economic development and ignore environmental protection, it will cause serious damage to the global climatic conditions and ecological structure. To avoid such tragedies, relevant government and non-government agencies need to be able to effectively monitor land use and land cover changes. In order to achieve effective resource management and plan or adapt land use plans. Remote sensing data is currently the fastest method to obtain a large amount of land use and land cover data. Among the various remote sensing data, satellite images are often used by researchers to derive land cover data and doing change detection due to their advantages of high time resolution and low cost. source. And among the common methods for discussing land cover change with satellite images, the classification-based approach is the method most used by researchers because it can fully discuss the relationship between categories. Previous studies are mostly only comparing the differences in land cover maps between two time periods when discussing land cover change. However, this method does not fully utilize the time series characteristics of multi-period land cover data. Therefore, this study attempts to combine the LICD and the Mann-Kendall trend test to propose a hot spot analysis algorithm for spatial and temporal changes suitable for categorical data. And this algorithm is used to analyze the land cover changes in the Xincheng and Marseille areas, Suao Town, Yilan County from 2006 to 2016. The research results show that this algorithm can present the temporal and spatial aggregation trends of various categories from the multi-period land cover category data. The results of such analysis can assist urban planners and relevant government agencies with useful information in planning.