不同尺度台灣森林生物量與枯落物動態之分析

Abstract

森林生態系是陸域生物圈主要的碳庫，而森林碳儲藏量與流動則是瞭解碳收支的關鍵。此論文利用遙測影像評估熱帶與亞熱帶生物量與枯落物之動態，也利用動態全球植被模式模擬颱風干擾所產生的落葉，以及颱風強度與頻率的改變對森林碳循環之影響。在第二章的研究中，此論文利用高空間解析度的QuickBird衛星影像獲取林冠面積，並利用林冠面積計算生態系碳儲藏量，以瞭解碳儲藏量在棲蘭山地形與邊際效應影響下的空間異質性情形。研究結果顯示，棲蘭山扁柏雲霧林平均每棵樹的碳儲藏量為138  118 kgC，密度為119.3 MgC ha-1。在空間分佈上，溫度與太陽輻射隨著海拔上升而遞減，而坡度則是影響扁柏林冠面積變異的重要因子。顯示地形異質性的存在，改變營養與生物氣候環境，進而影響植被結構的空間分佈情形。在第三章的研究中，此論文利用時間序列的MODIS遙測影像，萃取地表溫度(land surface temperature, LST)與可行光合作用植被覆蓋指數(photosynthetically active vegetation cover)來估算森林枯落物量。研究結果顯示，台灣森林平均年枯落物量為5.1 ± 1.2 Mg ha-1 yr-1。在枯落物的年變異上，可能受到颱風與乾旱干擾因子影響，而海拔與坡向則是影響枯落物空間變異的重要因子。在第四章的研究中，此論文利用動態全球植被模式Hybrid v4.2，並加入風速、溫度、雨量參數，以模擬福山森林生態系受到颱風干擾所產生的落葉情形。研究結果顯示，模擬與實地觀測的落葉量有顯著的關係(r2 = 0.33, p < 0.001)，但模擬結果有些微的低估現象(15.6%)。在颱風干擾的環境下，落葉量、葉面積指數、總生物量、總初級生產力、淨初級生產力皆比非颱風干擾的環境低。在提升颱風強度與頻率的模擬下，將產生更高的落葉量與土壤呼吸，使得葉面積指數、總生物量、總初級生產力、淨初級生產力減少，可能影響森林長期的碳收支能力。整體而言，利用遙測影像與動態全球植被模式監測森林生物量、枯落物、碳循環的時間與空間分佈情形，有助於瞭解與預測森林生態系在干擾環境下的碳收支動態。

Forest ecosystem is the major carbon pool in terrestrial biosphere, and understanding dynamics of forest carbon stock and flux is important for carbon budget. This dissertation used remotely sensed imagery to evaluate the pattern of biomass and litterfall in subtropical/tropical forests. The dynamic global vegetation model was also included to model typhoon-induced defoliation and carbon cycle under scenarios of typhoon intensity and frequency. Firstly, this dissertation extracts canopy areas from a high spatial resolution QuickBird image to assess the patterns of carbon stock across topographical gradients and the edge effect in a Chilan montane cloud forest. Mean ( s.d.) carbon per individual was 138  118 kgC with an estimated site carbon density of 119.3 MgC ha-1 in a false cypress (Chamaecyparis obtusa var. formosana, CHFO) cloud forest. The negative trends of temperature and solar radiation along the elevation gradient were observed, and slope played a crucial role in varying CHFO canopy area. The findings suggest that topographical heterogeneity may alter the spatial patterns of nutrients and bioclimate that influence the structure of the vegetation. Secondly, time-series MODIS LST- (land surface temperature) and PV-derived (photosynthetically active vegetation cover) metrics were used to derive litterfall. The mean annual litterfall was 5.1 ± 1.2 Mg ha-1 yr-1 in Taiwan montane forests. The temporal dynamics of the litterfall revealed that typhoons and consecutive drought events might be important for temporal variability of litterfall. The elevation gradient and aspects were critical to dominate spatial patterns of litterfall. Thirdly, a wind built-in function with wind speed, temperature and precipitation variables was developed in a dynamic global vegetation model Hybrid v4.2 to simulate typhoon-induced leaf litter in the Fushan experimental forest. The modeled and observed leaf litter was significant correlation (r2 = 0.33, p < 0.001) with moderate underestimation (15.6%). The leaf litter, leaf area index (LAI), total biomass, gross primary production (GPP) and net primary production (NPP) were higher in non-typhoon environment than in typhoon-affected environment. The raises of typhoon intensity and frequency would increase leaf litter and soil respiration resulting in the decreases of LAI, biomass, GPP and NPP. If the substantial defoliation from more intense or frequent typhoon might affect regional carbon budget at long-term time scale. Overall, using satellite images and a dynamic global vegetation model to monitor spatial and temporal patterns of living biomass, litterfall and carbon flux may facilitate monitoring and predicting of carbon budget and perturbations in forest ecosystems.