本研究探討臺灣南部恆春半島之步道侵蝕與退化現象，研究範圍涵蓋墾丁國家公園內及其周邊區域的休閒步道系統，並整合無人機（UAV）攝影測量、地理資訊系統（GIS）、歷史衛星影像以及遊客問卷調查等方法進行分析。 本研究於2025年4月及2026年4月進行高解析度無人機航測，針對四條具代表性的休閒步道進行空拍影像蒐集。四條步道分別取得565張、786張、1，544張及537張影像。所有影像皆利用 Agisoft Metashape 軟體進行處理，以建立高解析度正射影像（Orthomosaic）與數值高程模型（Digital Elevation Model， DEM）。後續透過 QGIS 進行空間分析，製作侵蝕分布圖、等高線圖及步道剖面圖，以評估步道形態特徵與退化程度。此外，本研究利用臺灣農業部林業及自然保育署航空測量及遙測分署提供之1991年至2025年歷史衛星影像，透過影像數位化與空間比較分析，評估長期地景變化。同時，透過250份遊客問卷調查，分析遊客行為及其與步道退化之間的關聯性。 研究結果顯示，四條步道之侵蝕特徵具有明顯的空間差異。無人機產製之高解析度資料成功辨識出步道拓寬、表面切蝕、細溝侵蝕（rill erosion）、局部崖壁侵蝕以及地形不穩定等現象，並能以公分尺度呈現細微地形變化。步道橫剖面分析顯示，不同測線位置之步道寬度與深度存在顯著差異，而DEM分析則揭示了大型尺度的崖壁侵蝕以及小尺度的表層侵蝕過程。歷史衛星影像分析結果指出，研究期間內地景持續發生變化，包括裸露地面面積增加以及植被覆蓋率下降等現象。問卷調查結果進一步顯示，遊客行為，尤其是偏離步道行走以及集中於觀景點活動，會加劇局部區域的步道退化。 本研究證明，結合無人機攝影測量、歷史遙測資料與地理資訊系統，可建立一套有效的休閒步道退化監測方法，並能提供保護區內步道管理與永續環境保育之科學依據。;This research investigates trail erosion and degradation on the Hengchun Peninsula in southern Taiwan， focusing on trail systems within and adjacent to Kenting National Park by integrating unmanned aerial vehicle (UAV) photogrammetry， geographic information systems (GIS)， historical satellite imagery， and visitor questionnaires. High-resolution UAV surveys were conducted in April 2025 and April 2026 to collect aerial imagery for four representative recreational trails. A total of 565， 786， 1，544， and 537 images were acquired for Trails 1， 2， 3， and 4， respectively. The imagery was processed using Agisoft Metashape to generate high-resolution orthomosaics and Digital Elevation Models (DEMs)， which were subsequently analyzed in QGIS to produce erosion maps， contour maps， and cross-sectional profiles for evaluating trail morphology and degradation. Historical satellite imagery from 1991 to 2025， obtained from the Aerial Survey and Remote Sensing Branch of the Forestry and Nature Conservation Agency， Taiwan， was used to assess long-term landscape changes through digitization and spatial comparison. In addition， a questionnaire survey involving 250 visitors was conducted to evaluate visitor behavior and its relationship to trail degradation. The results reveal considerable spatial variability in erosion characteristics among the four trails. UAV-derived products identified trail widening， surface incision， rill development， localized cliff erosion， and terrain instability with centimeter-level detail. Cross-sectional analyses demonstrated substantial differences in trail width and depth along individual transects， while DEMs highlighted both large-scale cliff erosion and small-scale surface degradation. Historical satellite analysis indicated progressive landscape change， including an increase in bare ground and a decline in vegetation cover over the study period. The questionnaire results further suggest that visitor behavior， particularly off-trail movement and concentrated use at viewpoints， contributes to localized trail degradation. This study demonstrates that the integration of UAV photogrammetry， historical remote sensing， and GIS provides an effective framework for monitoring recreational trail degradation and supports evidence-based conservation and sustainable trail management in protected areas.
