- 學位論文
柳杉人工林林分生長機制模型之建立與應用
Development and Applications of a Mechanistic Stand Growth Model for Japanese cedar (Cryptomeria japonica) Plantations
摘要
氣候變遷不可逆的後果促使各行各業實現碳中和的目標。森林則被用作吸收二氧化碳的自然解方(nature-based solution)。儘管如此，缺乏在氣候變遷下，能可靠預測生長的生長機制模型(mechanistic model)，阻礙了臺灣制定更好的碳吸存(carbon sequestration)森林經營策略。為了結合氣候變遷對林分生長的影響，本研究為臺灣大學實驗林的柳杉(Cryptomeria japonica)人工林開發了一林分生長混合模型(hybrid model)，即結合動態冠層孔隙與泊松死亡之生理原理生長預測模型(Physiological Principles for Predicting Growth with Dynamic Canopy Opening and Poisson Mortality, 3-PGDCOP)，以應用於森林經營及林學研究。3-PGDCOP根據氣候因子模擬動態林分生長，透過異速生長方程式(allometric equations)估計生物量分配(biomass allocation)，以零膨脹泊松模型(zero-inflated Poisson modeling)量化死亡率，並根據林分狀態變數(stand state variables)模擬動態冠層孔隙(canopy opening)。為了對模型進行參數化和校正，本研究回顧現有文獻和植物性狀開放資料庫，並統整臺灣大學實驗林二十三個柳杉試驗地之長期氣象及林分生長資料，林齡69至107年不等(2019年)。模型驗證結果顯示模型表現良好，林分密度(stand density)的方均根誤差(root mean square error, RMSE)及平均絕對百分比誤差(mean absolute percentage error, MAPE)分別為235 st ha-1及18.4%，方均根胸徑(quadratic mean diameter at breast height)則分別為2.3 cm及6.3%。模型對觀察值變異量(variation)的解釋能力高，對林分密度及方均根胸徑的決定係數(coefficient of determination, R2)分別達0.943及0.943。 異速生長方程式的參數化結果顯示植物根莖葉之生物量與方均根胸徑之間的異速生長關係存在地域差異。雙層泊松死亡率模型(2-layer Poisson mortality model)能有效模擬自我疏伐(self-thinning)，其模型誤差暗示在老齡林中存在非依密度之死亡機制(density-independent mortality process)。冠層地面覆蓋比例(fractional ground cover by the canopy)模擬了不同栽植密度下之林分冠層發展情形。 為協助柳杉人工林經營者應對氣候變遷，本研究將3-PGDCOP應用於氣候變遷及人為林分更新情境分析(scenario analyses)、栽植密度之決定、立地肥沃度(fertility rating)之估計，以及葉部生物量異速生長分析。分析結果建議臺灣大學實驗林人為更新老齡柳杉人工林林分，並將未來種植之柳杉林輪伐期(rotation age)定為35-48年，疏伐年限(age limit for thinning)定為18-23年，栽植密度定為3000 st ha-1，以在未來氣候中獲得更好的碳儲存能力和更高的木材產量。在上述建議範圍內，若未來溫室氣體集體輻射強迫力(radiative forcing)較高，輪伐期和疏伐年限應隨之加長，反之亦然。立地肥沃度之估計值與霧頻度(fog frequency)的相關性顯示霧對林分生長有正面影響，相關係數(correlation coefficient)達0.72。模型模擬之葉部異速生長方程式呈現右移現象而非恆定，因此異速生長方程式應建立於具相近林齡之林分生長資料。
並列摘要
The irreversible outcomes of climate change have urged all sectors to attain the goal of carbon neutrality. Forests are used as a nature-based solution to absorbed carbon dioxide. Nonetheless, the lack of a mechanistic growth model for reliable predictions under the changing climate hinders the formulation of better forest management strategies for carbon sequestration in Taiwan. To incorporate the influence of climatic variations on stand growth, this study developed a hybrid stand growth model, the Physiological Principles for Predicting Growth with Dynamic Canopy Opening and Poisson Mortality (3-PGDCOP), for the Cryptomeria japonica plantations in the National Taiwan University Experimental Forest (NTUEF) for applications in forest management and research. The 3-PGDCOP simulates dynamic stand growth from climatic variables, estimates biomass allocation by allometric equations, evaluates mortality via zero-inflated Poisson modeling, and simulates dynamic canopy opening from stand state variables. To parameterize and calibrate the model, I reviewed the existing literatures and open database of plant traits, and assembled long-term meteorological and stand growth data across 23 C. japonica sites in the NTUEF with stand ages ranging from 69 to 107 years in 2019. The validation results showed a good model performance, with RMSE and MAPE of 235 st ha-1 and 18.4% for stand density and 2.3 cm and 6.3% for quadratic mean diameter at breast height (DBH), respectively. Model explanatory power on the observed variation was high with a determination of coefficient (R2) of 0.943 for stand density and 0.943 for quadratic mean DBH. The parameterization results of the allometric equations revealed location variations in the allometries between plant part biomasses and the quadratic mean DBH. The 2-layer Poisson mortality model well predicted the self-thinning, and the prediction error implied the existence of a density-independent mortality process in old-growth stands. The fractional ground cover by the canopy simulated the canopy development in stands of various planting densities. To assist C. japonica plantation managers coping with climate change, I applied the 3-PGDCOP in scenario analyses on climate change and artificial regeneration, planting density determination, site-specific fertility rating estimation, and the analysis on foliage biomass allometry. Management recommendations for the C. japonica plantations in the NTUEF are regeneration of the old stands, setting the rotation age to 35-48 years and the age limit for thinning to 18-23 years, and adopting a planting density of 3000 st ha-1 to attain better carbon storage capacity and higher timber production in the future climate. Within the above recommended ranges, the rotation age and the age limit for thinning should be set longer under the conditions of higher collective radiative forcing of greenhouse gases in the future, and vice versa. The correlation between the calibrated fertility rating and fog frequency consented with the view of positive impact of fog on stand growth, with a correlation coefficient of 0.72. The right shift phenomenon in the simulated foliage allometric equation recommended the building of allometric equations on inventory data of similar stand age.
參考文獻
延伸閱讀
- 陳柏蓉（2020）。探討各種生長收穫模式對柳杉人工林蓄積量模擬結果之分析〔碩士論文，國立臺灣大學〕。華藝線上圖書館。https://doi.org/10.6342/NTU202003825
- 王松永（1985）。柳杉及杉木之材質與利用（Ⅰ）。林產工業，4(4)，79-88。https://doi.org/10.6561/FPI.1985.4(4).9
- 孫義方、黃正良、陳明杰、曹崇銘、黃志堅、傅昭憲、黃騰禾（2012）。柳杉人工林不同疏伐強度對原生樹種更新之影響。中華林學季刊，45(1)，13-29。https://doi.org/10.30064/QJCF.201203.0002
- 王松永（1986）。柳杉及杉木之材質與利用（II）。林產工業，5(1)，66-76。https://doi.org/10.6561/FPI.1986.5(1).8
- 陳淯婷、顏添明、李介祿（2010）。Crown Characteristics and Crown Shape of Japanese Cedar (Cryptomeria Japonica D. Don) in an Even-Aged Stand。中華林學季刊，43(2)，213-222。https://doi.org/10.30064/QJCF.201006.0003