以常態化差異植生指標預測植生復育邊坡之地上部碳存量

氣候變遷及全球暖化是目前人類需共同面對的問題。全球暖化是由於溫室氣體的增加，當中又以二氧化碳為主，對於制定緩解全球暖化的策略，方可從了解碳匯量著手。森林生態系對於碳的吸收具極大貢獻，為貢獻陸域生態系中非常重要的碳庫，為探究這些問題，森林調查是最常進行的工作之一，然而森林幅員廣闊，實地調查耗時耗力，僅能以取樣方式進行，對於空間及時間仍存在侷限性，且臺灣氣候及地形相對複雜，坡地災害造成地表變化非常迅速，相較實地調查，遙感探測技術能於短時間提供地表資訊，以較為經濟的方法推估林分材積生物量與推估森林碳存量。本研究於南瀛天文館之不同年份崩塌地植生復育林內設置17個10m×10m的樣區進行實地調查，量測材積以估算碳存量，經估算結果顯示，復育年份2年、7年、9年、12年、15年及20年之碳存量分別為0.85±0.78 Mg/ha、19.6±5.54 Mg/ha、25.7±8.64 Mg/ha、13.5±9.76 Mg/ha、14.3±8.33 Mg/ha及60.3±16.4 Mg/ha，並分析不同年份植生復育邊坡之碳存量變動情形，而後以多光譜無人飛行載具於研究範圍內進行拍攝，利用多光譜空拍影像之近紅外光與紅光計算常態化差異植生指標 (NDVI)建置模型以推估碳存量，經驗證後，兩者具有顯著性相關 (R2=0.897, p<0.001)，顯示可有效推估地上部碳存量之空間分布，另以光能利用模型推估淨初級生產量 (NPP)，並與各工區之碳吸存率進行相關性分析，結果顯示，利用推估之NPP結果與實地調查之碳吸存率並無相關性 (R2=0.22, RMSE=3.61)，因此後續探討各項因子對於模型推估能力的影響，以評估其對於推估植生復育林地碳吸存率的適用性。

關鍵字

碳存量；遙感探測；常態化差異植生指標；淨初級生產量

並列摘要

Climate change and global warming are the issues that humans must face. The cause of global warming is greenhouse gas emissions, mostly carbon dioxide. Realizing the carbon sink is the strategy for alleviating global warming. The forest ecosystem is the most important carbon pool for contributing terrestrial ecosystem and absorbing the carbon. Forest investigating is one of the most typical works to estimate the carbon pool in the forest ecosystem. Since the forests are spacious and investigating personally usually takes much time and effort, sampling in setting plots is generally selected for investigation. However, plot sampling is still limited in space and time, the weather and topography are complicated in Taiwan, and slope land disaster changes land cover quickly, increasing the investigation difficulties. Compared with the personal investigation, remote sensing is more efficient in providing information on the land surface and more economical the biomass and volune of forest stand forest carbon stocks. Seventeen plots with a space 10 m by 10 m were set up in an area reforested after landslids occurred in different years at Tainan Astronomical Education Area. After measuring the volume to estimate carbon stocks, the result shows that their’s Year regeneration 2 years, 7 years, 9 years, 12 years, 15 years and 20 years were 0.85±0.78 Mg/ha, 19.6±5.54 Mg/ha, 25.7±8.64 Mg/ha, 13.5±9.76 Mg/ha, 14.3±8.33 Mg/ha and 60.3±16.4 Mg/ha, respectively. Analyzing the changes of carbon stocks on landslides of different-years reforestation. Using a multispectral UAV to photograph within the field of research. Making use of multispectral UAV images in Near Infrared and Red to calculate NDVI, also, building up the model to estimate carbon stocks. The verification shows that a significant correlation (R2=0.897, p<0.001) between them. It can effectively estimate spatial distribution of overground carbon stock. Moreover, using model to estimate NPP based on luminous energy and conducting correlation analysis according to carbon sequestration rate of worksite. As a result, it shows no correlation (R2=0.22, RMSE=3.61) between estimated NPP and carbon sequestration rate in field survey. Thus, keep discussing the influence of model estimating ability by each factors to evaluate applicability and its carbon sequestration rate at restored slopelands.