← 回文獻卡索引

以地面光達評估北加州多樣化森林之非破壞性地上生物量異速生長方程式Using terrestrial laser scanning to evaluate non-destructive aboveground biomass allometries in diverse Northern California forests

Krause, Forbes, Barajas-Ritchie, Clark, Disney, Wilkes, Bentley｜Frontiers in Remote Sensing 4: 1132208｜DOI: 10.3389/frsen.2023.1132208

狀態：AI_DRAFT_FROM_REVIEW｜分級：A｜閱讀深度：FULL_TEXT_CHECKED｜Jacky 審核：False

森林數位孿生底層致能

讀學習筆記 →

terrestrial laser scanningaboveground biomassallometric equationsTLSquantitative structure modelnon-destructive samplingNorthern California

專討核心文獻定位

[56] Ch2 · 地面量測 新增
Krause et al. · 2023
以地面光達對北加州5樹種282棵樹做非破壞性AGB估算，TLS體積比FIA異速方程式約高30%，並建立DBH加樹高的新異速方程式優於僅用DBH者

使用警示

本頁是文獻知識庫卡片，不等於可直接引用的最終查核稿。只有狀態升級為 CITABLE 後，才可直接進入論文引用候選。

為什麼納入這篇

這篇把地面光達定位成傳統破壞性採樣以外的非破壞性 AGB 量測途徑，量化了既有異速生長方程式與真實樹體積之間的系統性偏差，正好支撐地面量測章節對異速方程式不確定性的論述。

結構式摘要｜中英文對照

研究問題
在缺乏在地破壞性採樣資料的北加州多樹種森林，能不能用地面光達非破壞性地準確估算樹高、胸徑與地上生物量，並據此建立比既有公式更可靠的物種別異速生長方程式？
In diverse Northern California forests that lack local destructive sampling data, can terrestrial laser scanning non-destructively and accurately estimate tree height, DBH and aboveground biomass, and be used to derive species-specific allometric equations that are more reliable than existing published equations?
資料來源
原文確認：在北加州三個樣區（Pepperwood Preserve、Saddle Mountain、Latour Demonstration State Forest）對 5 樹種共 282 棵樹做地面光達掃描；另以傳統森林調查蒐集 550 棵樹的胸徑與 291 棵樹的樹高作為驗證資料。木材密度取自文獻。
The text confirms terrestrial laser scanning of 282 trees from 5 species across three Northern California sites (Pepperwood Preserve, Saddle Mountain, and Latour Demonstration State Forest); traditional forest inventory provided DBH for 550 trees and height for 291 trees as validation data. Wood density values were taken from the literature.
方法
原文方法確認：使用 RIEGL VZ-400i 光達於每樣區 10×10 公尺網格設 9 個掃描站，於 RiSCAN PRO 完成多站平差套合並以 2013 年機載光達校正建立數值高程模型；在 Lidar360 半自動分割單株、量測 DBH 與樹高，於 CloudCompare 清點雲並以 TLSeparation 去除葉片，再用 TreeQSM 量化結構模型估算莖與枝體積，結合文獻木材密度推算 AGB。AGB 與 FIA、Jenkins et al. 2003、Chojnacky et al. 2014 及 Sillett et al. 2019（僅紅杉）四套公式比較，並以 RMSE、CV RMSE、bias 與 percent bias 評估；最後以對數轉換最小平方迴歸為每樹種建立僅用 DBH 與用 DBH 加樹高的兩式新異速方程式。
The text confirms use of a RIEGL VZ-400i scanner at nine scan positions on a 10×10 m grid per plot; scans were co-registered in RiSCAN PRO via multi-station adjustment and aligned to 2013 airborne LiDAR to build a digital elevation model; individual trees were semi-automatically segmented in Lidar360, with DBH and height measured, cleaned in CloudCompare, leaf-separated with TLSeparation, and modeled with TreeQSM to estimate stem and branch volume; AGB was computed by combining QSM volume with literature wood density. TLS AGB was compared against FIA, Jenkins et al. 2003, Chojnacky et al. 2014, and Sillett et al. 2019 (redwood only) equations using RMSE, CV RMSE, bias and percent bias; finally, log-transformed ordinary least squares regression was used to derive two new allometric equations per species (DBH-only and DBH-plus-height).
主要結果
原文確認：TLS 推算的胸徑與樹高與野外實測無顯著差異（DBH R2 = 0.98、樹高 R2 = 0.95）。除 P. ponderosa 外，跨所有樹種 TLS QSM 體積約比 FIA 異速方程式高 30%。在 AGB 比較上，與 Jenkins et al. 2003 公式的 CV RMSE 最低（36.10%）、bias 約 1%，與 Chojnacky et al. 2014 與 FIA 公式則為 39.09%、40.34% CV RMSE 與 20%、11% bias。整體 TLS AGB 平均比既有公式高約 10%，但偏差因樹種而異，闊葉樹（Quercus 屬）偏差大於針葉樹；S. sempervirens 與 Q. agrifolia 偏離最大。納入樹高的新異速方程式（Eq. 2）對除 Q. garryana 外的所有樹種都比僅用 DBH（Eq. 1）擬合更好。
The text confirms that TLS-derived DBH and height did not differ significantly from field measurements (DBH R2 = 0.98, height R2 = 0.95). Across all species except P. ponderosa, TLS QSM volumes were about 30% greater than FIA-equation estimates. For AGB, comparisons with Jenkins et al. 2003 equations had the lowest CV RMSE (36.10%) and ~1% bias, while Chojnacky et al. 2014 and FIA had CV RMSE of 39.09% and 40.34% and bias of 20% and 11%. Overall, TLS AGB averaged about 10% greater than published equations, with species-dependent deviation; hardwoods (Quercus) deviated more than conifers, and S. sempervirens and Q. agrifolia deviated the most. The new height-inclusive equations (Eq. 2) fit better than the DBH-only equations (Eq. 1) for all species except Q. garryana.
限制
原文確認：研究未做破壞性採樣驗證，故新方程式是否因在地差異而偏離仍待確認；木材密度取自文獻並假設物種內一致，但心材與邊材、莖與枝之間其實有差異；樹體大小未納入交互作用項，較大樹偏離 1:1 線較多；QSM 對點雲不清或下層密集樹冠之個體常失敗，常綠林帶葉期遮蔽更增困難；且在低密度或小尺度森林，TLS 資料處理時間可能超過直接野外量測。
The text confirms no destructive-sampling validation, so whether the new equations deviate due to local variation remains to be verified; wood density was taken from the literature and assumed constant within species despite heartwood-sapwood and stem-branch differences; tree size was not included as an interaction term and larger trees deviated more from the 1:1 line; QSMs failed for trees with unclear point clouds or dense understory, with leaf-on occlusion adding difficulty in evergreen forests; and at small scales or low stand density, TLS processing time may exceed direct field measurement.

Key Findings

發現證據確定性
Terrestrial laser scanning can non-destructively recover tree DBH and height as accurately as field inventory and can estimate AGB via QSM volume and literature wood density.Original Results 3.1: TLS DBH and height did not differ significantly from field measurements (R2 = 0.98 for DBH, R2 = 0.95 for height) across 550 DBH and 291 height observations.checked_against_original_txt
Published allometric equations systematically underestimate biomass relative to TLS-measured trees, with TLS QSM volumes about 30% greater than FIA estimates and TLS AGB about 10% greater than published equations.Original Results 3.2-3.3 and Discussion: across all species except P. ponderosa, TLS QSM volumes were ~30% greater than FIA; TLS AGB averaged ~10% greater than published equations, with the largest deviation for S. sempervirens and Q. agrifolia.checked_against_original_txt
New TLS-derived allometric equations that include both DBH and height fit better than DBH-only equations for almost all study species.Original Results 3.4 and Table 1: height-DBH equations had better R2 and CV RMSE than DBH-only equations for all species except Q. garryana, where R2 did not differ.checked_against_original_txt

Key Figures and Tables

公開網站原則：未確認授權前，不直接複製原文圖表；優先使用自製圖表導讀或重繪圖。

項目內容關鍵數字Jacky 判讀重用策略
Figure 2, Figure 3, Figure 4, Table 1Figure 2 regresses traditional vs TLS DBH and height; Figure 3 compares TLS QSM volume against FIA volume by species; Figure 4 compares TLS QSM AGB against AGB from FIA, Jenkins et al. 2003, Chojnacky et al. 2014 and Sillett et al. 2019; Table 1 reports the new per-species allometric equation coefficients with R2, RMSE and CV RMSE.550 trees for DBH and 291 for height; DBH R2 = 0.98, height R2 = 0.95; 282 successful QSMs across 5 species; TLS QSM volume ~30% greater than FIA; Jenkins 2003 lowest CV RMSE 36.10% and ~1% bias; Table 1 new-equation R2 ranges 0.78-0.96.把這組圖表當成地面量測階段異速方程式不確定性的實證證據，說明同一批樹用不同公式可差出兩到三成，是 FDT 碳儲量層必須吸收而非忽略的誤差來源。本文為 CC BY，圖表可註明出處重製；發布時仍偏好自繪簡化版聚焦在 30% 與 10% 偏差訊息。

Extracted Evidence Table

可支撐主張指標或結果原文位置可引用備註
既有異速生長方程式相對於 TLS 量測會系統性低估樹體積與生物量。TLS QSM 體積約比 FIA 高 30%（除 P. ponderosa 外）；TLS AGB 平均比既有公式高約 10%，與 Jenkins et al. 2003 公式比對 CV RMSE 最低為 36.10%。原文 Results 3.2-3.3，p.05；Discussion p.08；Figure 3、Figure 4、Table 1。True可引用作為地面異速方程式不確定性的具體數字；30% 與 10% 為本研究 5 樹種的結果，非全球或台灣通用值。

Critical Appraisal

Strengths

Weaknesses

Validation quality對 TLS 維度量測驗證強（與野外實測 R2 0.95-0.98）；但 AGB 本身缺破壞性採樣真值，屬間接驗證
Transferability to Taiwan中等；TLS 非破壞性建立在地異速方程式的方法可移轉，但台灣樹種與木材密度需以本地資料重新校正
Risk of overclaiming不可把 30% 體積差或 10% AGB 差當成已用破壞性採樣證實的真實偏差，原文亦明言需後續破壞性採樣驗證。

與 Jacky 博論 / Review 的用途

博士論文支撐博論主張，地面量測與異速生長方程式本身帶有系統性且因樹種而異的不確定性，FDT 的碳儲量層必須把這層誤差顯式建模，而非套用單一既有公式。
TJFS Review強化 TJFS review 地面量測章節，提供 TLS 作為非破壞性異速方程式來源的具體案例與可引用的偏差數字。
可引用句候選2023 年，Krause 等人發表的文獻中指出，以地面光達非破壞性估算的地上生物量約比既有異速生長方程式高 10%、其結構模型體積更比 FIA 公式高約 30%，顯示傳統公式對部分樹種會系統性低估生物量。
不可用來主張不可用本文作為台灣特定碳儲量數字，或作為空載與星載光達生物量精度的證據。

授權與圖表重用

Article licenseCC-BY-4.0
Figure reuse policyREUSE_ALLOWED_WITH_ATTRIBUTION_CC_BY_4.0
Notes首頁版權聲明確認本文為 Creative Commons Attribution License (CC BY) 開放取用文章，圖表在註明出處後可重製；發布前仍建議自繪簡化版以聚焦地面量測訊息。

待查核清單