臺灣高山型國家公園歷史火燒嚴重度評估及復原狀態分析

Abstract

隨氣候變遷的影響加劇，火燒頻率及強度逐漸增加，直接影響動植物多樣性及分布、生態系服務功能 (ecosystem service)等，以保護生態而成立的國家公園及保護區之經營管理策略顯得更加重要。過去在經營管理中以預測及撲滅火勢為主要行動，經常忽略掉後續生態的回復監測。然而，在生態系中，自發生干擾到植群逐漸演替至穩定狀態，才是完整的生態系循環過程。臺灣過去許多研究多針對火燒後短期的植被變化或火燒前及火燒當下引發嚴重火燒的關鍵因子，但長期以來卻鮮少有整合性的研究關於誘發嚴重火燒的風險評估及林火後長期植被復原之研究。因此，本研究以兩個章節針對 1984 – 2021年間高山型國家公園境內及周邊的歷史火燒事件，第一章節以隨機森林 (random forest)分析歷史火燒的嚴重度與火燒發生前環境因子間的關係，了解促發嚴重火燒的環境因子，並在第二章以差異常態化燃燒指標 (differenced normalized burn ratio, dNBR)監測大面積歷史火燒 (>10 ha)跡地的植被覆蓋至今是否已恢復到火燒前狀態，整合過往現地調查結果，了解火燒跡地回復過程中的自然及人為阻礙。研究結果表示，以氣候及植被為最重要的火燒嚴重度關鍵因子，其餘因子為火行為及火源供給的重要因素。火燒後的森林復原與生育地環境及火燒嚴重度具有相關性，除了其生長立地條件外，人為引發的重複火燒、復育行為及大型天然災害的發生也會影響森林覆蓋之復原速率。

The intensification of climate change has led to an increase in the frequency and severity of wildfires, resulting in significant impacts on biodiversity, species distribution, and ecosystem services. This highlights the urgent need for comprehensive management strategies in national parks and protected areas, emphasizing not only fire prediction and suppression but also the monitoring of post-fire ecological recovery. Historically, management efforts have focused predominantly on wildfire prediction and firefighting, often neglecting the critical processes of long-term ecological recovery, including vegetation succession and stabilization. In Taiwan, research has largely centered on short-term vegetation changes following wildfires or the factors triggering severe fires. However, there remains a gap in integrated studies addressing risk assessment for severe wildfires and the long-term recovery of post-fire ecosystems. This study investigates wildfire events between 1984 – 2021 within and surrounding high-mountain national parks. The research is divided into two key components: the first employs a random forest (RF) algorithm to analyze the relationship between wildfire severity and pre-fire environmental factors, identifying critical determinants of severe wildfire events. The second component assesses vegetation recovery in large burn scars (>10 ha) using the differenced normalized burn ratio (dNBR) to evaluate whether vegetation has returned to pre-fire conditions. The findings reveal that climate and vegetation are the most influential factors determining wildfire severity, with fire behavior and fuel supply also playing significant roles. Post-fire forest recovery is closely associated with site conditions and fire severity, while repeated anthropogenic fires, restoration activities, and large-scale natural disasters further affect recovery rates. By integrating historical field survey data, this study elucidates the natural and anthropogenic factors that hinder post-fire recovery processes and provides valuable insights to inform adaptive management and restoration strategies in fire-affected ecosystems.