以MODIS卫星光谱资料推估不同乾旱时期植群冠层之水分含量.doc

以 MODIS 衛星光譜資料推估不同乾旱時期植群冠層之水分含量Estimating the Vegetation Water Content on Different Drought Condition with MODIS Reflectance Data研究生：顏瑋利 Wei-Li Yan指導教授：陳朝圳 Chaur-Tzuhn Chen【摘要】常態化植生指標 (NDVI) 雖然廣泛使用於植群監測之指標，但對於乾旱監測而言，因乾旱對於植群影響具有延遲之現象，NDVI 無法監測即時性的乾旱事件本研究先以控制性試驗，利用手持光譜儀及葉綠素測定器，收集不同乾旱狀態下之葉綠素變化及光譜反應特性，探討乾旱發生之植物葉片之生物物理反應並利用 MODIS 衛星影像之紅光、近紅外光、短波紅外光和熱紅外光波段，進行 NDVI、水勢指標 (WI)、常態化差異水指標 (NDWI) 和總體植群水分指標 (GVMI) 之計算，探討不同類型指標對於乾旱反應之敏感度，而乾旱程度之空間分布則以地表溫度(T s)和NDVI，建立植生指標與最高地表溫度之乾燥邊界，推估不同季節溫度植生乾燥指標 (TVDI) 之空間性變異。

結果顯示，在乾旱逆境下，植物葉綠素在未達到永久傷害時，葉綠素和光譜反射率可以建立起相關性五種指標乾季和濕季在不同土地利用類型，皆有顯著差異(p<0.05)，推估TVDI 分布圖能呈現清楚的乾旱空間變異，都市開發地區之土壤水分含量，顯然低於森林覆蓋區域，TVDI 確實能反應土壤的濕潤狀態，可作為監測乾旱之指標草生地對於乾旱的反應比森林敏感，WI 和植群水分含量有很高的相關性，但會受到冠層結構的影響；草生地之 GVMI 比 NDVI提前 3 個月偵測到植群含水量之變化(p<0.05)，結合近紅外光和短波紅外所推估之指標僅能表示植群含水量之狀態，無法以度量單位表示量的多寡，而 NDVI 無法表示植群的水分含量，但能提供地表植群之植生量關鍵詞：地表溫度、植生指標、近紅外光、短波紅外光、乾旱監測【Abstract 】The Normalized Difference Vegetation Index (NDVI) is a well-known measure for biophysical variables that has been widely used for vegetation monitoring. However, NDVI is not always an appropriate tool for real-time drought monitoring. Due to a lagged vegetation response to drought, NDVI cannot detect drought events instantaneously. In this study, we used the data of the CI-700 and CM-1000 experiment for chlorophyll variation and reflectance on different drought conditions, and discussed biophysical leaf properties under drought stress. The R, NIR, SWIR and thermal bands of MODIS images were used as data for calculating NDVI, Water Index (WI), Normalized Difference Water Index (NDWI) and Global Vegetation Moisture Index (GVMI). The land surface temperature and NDVI was assessed in order to estimate a drought index of Temperature-Vegetation Dryness Index (TVDI) in different seasons. The TVDI was based on an empirical parameterization of the relationship between land surface temperatures (Ts) and NDVI. The results showed that there were significantly different (p<0.05) in five land use indexes. As expected, there is gradual drought in the dry season in developed areas as well as in forest areas. From this result, we can conclude that the TVDI reflects the soil moisture status, and that it can be used as an index in future drought monitoring. The drought response of the grassland is more sensitive than forests. WI has a high correlation with vegetation water content, but affected by canopy structure and viewing geometry. GVMI of grassland detected changes in vegetation water content three months prior to NDVI. A combination of the SWIR and the NIR is required to calculate vegetation water content. GVMI is not related to the vegetation moisture content expressed as a percentage of water per quantity of biomass. NDVI provides different information (e.g., vegetation greenness), which is not directly related to the quantity of water in the vegetation.Keywords: Land surface temperature (Ts), Vegetation index (VI), Near Infrared (NIR), Shortwave infrared (SWIR), Drought monitoring。