手持光達於都市林胸徑與樹高調查之研究

Abstract

隨著都市林木資源調查自動化需求提升，手持式光達（Handheld LiDAR）因具操作靈活、效率高等優勢，逐漸應用於都市林調查領域，惟其量測精度與誤差來源尚待評估。本研究旨在評估手持式光達於都市林木性態調查中的應用可行性及其測量精度表現。研究內容包含三大部分：都市林木調查資料蒐集、手持式光達與傳統人工量測數據之比對分析，以及影響測量誤差因素之探討。研究地點於臺北市文山區元利建設四季莊園二期基地內進行樹木性態調查，針對胸徑（DBH）及樹高（Tree Height）兩項指標進行手持式光達掃描與人工量測，建立對應之資料庫；資料分析採用成對樣本t檢定比較兩種量測方法的差異，並運用單因子變異數分析（ANOVA）及相關性分析探討誤差值與樹木尺寸、林分密度等因子的關聯性，並根據分析結果提出於都市林調查應用手持式光達之實務建議。 研究結果顯示，手持式光達在量測胸徑與樹高時的精確度會受到樹木尺寸及林分密度等因素的影響，隨樹木尺寸增加測量誤差有逐漸增加的趨勢，但胸徑與樹高的誤差解釋力均偏低，顯示除樹木大小外尚存在其他重要影響因子。此外林分密度對測量誤差亦有顯著影響，高密度區（16-20株/400m²）的測量誤差明顯高於低密度區（1-4株/400m²），尤其在樹高量測上表現更明顯，推測與枝葉遮蔽及光束遮斷效應有關。整體而言，手持式光達於林分密度較低、主幹形狀規則且無分枝遮蔽的樹木，配合地面空曠且平坦的環境時，量測精度較高，但即便精度較高仍須進行人工複查。若於高密度或樹木結構複雜的地區，或遇胸高位置遮蔽等特殊情境，仍建議採取人工量測，以提升數據可靠性與調查結果之精確度。

This study aims to evaluate the feasibility and measurement accuracy of handheld LiDAR in urban tree attribute surveys. The research consists of three main components: urban tree data collection, comparative analysis between handheld LiDAR and traditional manual measurements, and the exploration of factors influencing measurement errors. The study was conducted at the Yuanli Construction Four Seasons Manor Phase II, located in Wenshan District, Taipei City. Two key attributes—diameter at breast height (DBH) and tree height—were measured using handheld LiDAR and manual methods to build corresponding datasets. Statistical analyses were conducted to compare error characteristics between the two methods and further investigate the relationship of measurement errors to tree size and stand density. Practical recommendations for applying handheld LiDAR in urban forestry surveys were then proposed based on these findings. The results indicated that measurement errors of handheld LiDAR for both DBH and tree height exhibited significant yet weak positive correlations with tree size. Specifically, the explanatory power of tree size for DBH measurement error was low, and even lower for tree height, suggesting other significant influencing factors beyond tree size. Additionally, stand density had a significant impact on measurement accuracy; measurement errors were notably greater in high-density plots (16-20 trees per 400 m²) compared to low-density plots (1-4 trees per 400 m²), especially pronounced in tree height measurements, likely due to foliage obstruction and beam interception effects. Overall, handheld LiDAR performed well for measuring trees with medium to low DBH and height, but caution and careful scanning strategies, along with point cloud processing adjustments, are recommended for use with large-diameter, tall trees or high-density environments to enhance measurement accuracy.