頭前溪流域植生覆蓋變遷之研究

Abstract

頭前溪為新竹縣市主要的河川，流域內涵蓋大範圍之山區，其中大部分為林地，開發區多集中於中下游地區，人口也均多聚集於此，稻米為流域內最大宗之作物。因為流域內植生面積廣大，要瞭解該區的土地利用變化情形，可以透過植生的偵測，進而得到植生覆蓋之變遷資訊。 本研究採用1993、1995、1997、1999及2000年五個時期之SPOT多光譜影像，進行頭前溪流域植生覆蓋之研究。所選擇影像的季節皆為夏季，為植生生長茂密之季節，可以較有效的掌握該區植物的分佈。 對於植生區域的萃取，為根據植生對於不同光譜的反射特性，採用兩種植生分類方法，分別為︰直接以NDVI分類及利用NDVI輔助原始光譜分類兩種。並經由分類成果來比較不同分類方法之差異。 利用不同分類方法之成果進行變遷分析，並比較採用影像相減法所得之變遷成果。經由變遷分析得到時間序列上之頭前溪流域植生覆蓋變遷資訊，結果顯示在流域內大量之植生變動情形，僅為自然環境的改變，無大量經由人為所造成之大變化；變化情形多發生於河流兩岸之沙洲地區，推測乃為自然水位之消長所引起之植生數量的變遷。

Tao-Cheng-Hsi is Hsin-chu’s main river. Its river basin covers large mountainous areas, most of which consists of woodland. Developed areas are centralized in the middle and lower reaches, where it is most populated. Rice is the main crop in the Tao-Cheng-Hsi basin. The vegetated area is large, and so the variation of land use can be obtained from the variation of vegetated area. In this study, SPOT multi-spectral images of 1993, 1995, 1997, 1999 and 2000 are applied to the study of vegetation area in the basin. Most images are taken in July when the vegetation is in flourish. This allows the distribution of vegetation to be detected more effectively. The extraction of vegetated area is based on different reflective characteristics in the spectrum. There are two ways adopted to sort the respective areas: one is by using NDVI directly and the other by using NDVI to assist the classification with the original spectral images. The results obtained for each schemes are comparatively studied. After obtaining the classification of vegetated and non-vegetated area for each scene, the land use change is computed by image subtraction. These five classifications are then forming a time series. Studying the spatial location of those changed areas, it is found that the major variation may be caused by the change in natural environment. Most variations usually happened on the sand bar beside the river. It is assumed that the growth and decline of running water affect the quantity of vegetation.