본 연구는 우리나라 소나무, 잣나무, 낙엽송 임분의 통합적인 생장 및 수확모델을 개발하고자 수행되었다. 분석 자료는 강원 및 경북지역에 설치되어 3년 간격으로 매목조사가 이루어지고 있는 영구고정표준지로부터 수집되었다. 수종별 영구고정표준지 개소 수는 소나무 39개소, 잣나무 51개소, 낙엽송 45 개소에 설치되었으며, 각 개소마다 세 개의 시험구를 설계하여, 간벌강도에 따라 대조구 20m×20m 약도간벌구 25m×25m 강도간벌구 30m×30m의 크기로 표준지가 설치되었다. 수간석해를 위해 한 그루의 표준목이 표준지 주변에서 수집되었으며, 1차 측정 이후에 간벌이 실시되었다. 매목조사된 자료와 수간석해 자료를 이용하여 본 연구의 생장 및 수확모델이 최종적으로 개발되었다. 최대입목밀도이론을 통하여 수종별 임분밀도지수가 제시되었으며, 흉고직경, 평균재적, 우세목 수고를 이용하여 밀도와의 관계가 평가되었다. 개체목의 흉고직경과 임분흉고단면적의 생장을 예측하고자 일반화 대수차 방법을 이용하여 기준임령 불변 동적 생장식이 개발되었다. 생장모델 개발을 위해 사용된 기본 함수식은 Bertalanffy-Richards식, Korf식, and Hossfeld식 이었다. 적합 및 검증 데이터의 검정통계량 결과를 통하여 각 수종별 최적모델이 제시되었으며, 추가적인 독립변수로써 임분 흉고단면적의 로그형태를 포함한 변형모델이 제시되었다. 기준임령 40년의 지위지수 모델과 더불어 수종별로 임분단위 수확모델이 개발되었다. 임령과 지위지수만으로 구성된 정상 수확모델과 현실림 수확모델이 임분재적추정을 위하여 서로 다른 범위의 상대임분밀도를 이용하여 개발되었다. 가변밀도 수확모델이 단위면적당 흉고단면적과 입목본수, 임분밀도지수에 따른 임분밀도 인자를 적용하여 개발되었다. 그 결과 임분 흉고단면적을 이용한 가변밀도 수확모델이 세 수종에서 모두 최적 수확모델로 선정되었다. 경급분포모델개발에서는 최적 함수를 선정하기 위하여 beta, gamma, Weibull 함수가 비교되었다. 분포함수에서 moment 및 percentile 모수복구 방법이 이용되었다. 그 결과 평방평균 임분직경, 25분위수 및 95분위수 흉고직경을 통해 모수 복구된 Weibull 함수가 경급분포모델을 위한 최적 함수로 선정되었다. 더불어 평방평균임분직경, 임령, 우세목수고, 입목본수를 이용하여 모수 복구 예측을 위한 모델이 개발되었다. 간벌 전 임분상태로부터 하층간벌에 따른 경급분포를 예측하고자 경급별 간벌모델이 제시되었다. Gingrinch 형태의 입목축적차트를 이용하여 수종별로 과소, 적정, 과밀 축적 기준과 함께 입목축적 가이드라인이 제시되었다. 입목축척차트를 개발하고자 입목면적비율과 Reineke의 임분밀도지수가 최대 적정 입목축적 밀도를 위해 비교되었다. 최소 적정 입목축적 밀도를 위해 경쟁이 없는 상태에서 생육하는 수관폭과 수관면적이 이용되었다. ha당 입목본수, ha당 흉고단면적 평균직경을 기반으로 하는 수종별 입목축적차트가 최종적으로 제시되었다. 본 연구에서 개발된 통합적 산림생장 및 수확모델들은 현장에서 실용적인 이용을 위하여 잘 적용될 수 있다고 판단된다.
This study was carried out to develop integrated growth and yield models of Pinus densiflora, Pinus koraiensis, and Larix kaempferi stands in Korea. Data for developing models was collected from the permanent monitoring plots, which were established in Gangwon and Gyeongsang provinces of Korea and repeatedly remeasured at 3 year interval. The total number of study sites were 39, 51, and 45, respectively by species, and the three plots were installed by each study sites. Those plots were designed for thinning study with the different plot size: 20m×20m for control plot, 25m×25m for light thinning plot, and 30m×30m for heavy thinning plot. One standard tree for stem analysis was collected near permanent plots, and thinning was operated after 1st inventory. Using forest measurements data and stem analysis data, growth and yields models were developed in this study. Stand density index for each species was suggested through the maximum stem number theory and size-density relationship was evaluated using diameter, dominant height, and volume. To predict the growth of individual tree diameter and stand basal area, base-age invariant dynamic equations were applied with generalized algebraic difference approach. Base equations used for developing growth models were Bertalanffy-Richards, Korf, and Hossfeld models. The best growth models for each species were provided and modified models with logarithm form of stand basal area as an additional independent variable were suggested by examining the fit statistics of fitting and validation data set. In whole stand level, three types of yield models were developed by species with site index model of base age 40. Normal and empirical yield model, comprised of only age and site index, were developed using different relative stand density to estimate stand volume. Variable-density yield model was lastly developed applying three kinds of stand density variables: basal area per ha, number of trees per ha, and stand density index. As a result, variable-density yield model with basal area per ha was selected as the best yield model of each species. In diameter distribution model development, three kinds of distribution functions were compared to select the best function: beta, gamma, and Weibull functions. In distribution functions, moment and percentile methods were used for parameter recovery. As a result, Weibull function with parameter recovery using quadratic mean diameter, 25th, and 95th percentile diameter was selected as the best function. In addition, the model for prediction of parameter recovery were developed using quadratic mean diameter, age, dominant height, and number of trees. In order to predict the diameter class distribution based on low thinning from the stand condition before thinning, finally, diameter class removal model was suggested for the stands with low thinning. Stocking guideline was suggested with fully-, over-, under-stocked criteria for each species using Gingrich-style stocking chart. To develop the stocking chart, tree area ratio and Reineke’s stand density index were used and compared for the maximum fully-stocked density. Crown width and area in open-growing trees were used for minimum fully-stocked density. Stocking charts by species were finally provided based on number of trees per ha, basal area per ha, and average tree diameter. The integrated growth and yield models developed in this study are highly considered to be applied for practical use in the field.
목차
Chapter 1. Introduction 11.1. Background 11.2. Problem statement and motivation 31.3. Objectives and scope 41.4. Significance of this study 5Chapter 2. Estimation of maximum stem number and evaluation of size-density relationship 62.1. Introduction 62.2. Materials and methods 82.2.1. Data 82.2.2. Size-density relationship 92.2.3. Maximum stem number theory 122.3. Results and discussion 172.3.1. Maximum stand number theory 172.3.2. Size-density growth prediction 232.4. Conclusion 28Chapter 3. Development of base-age invariant growth models using generalized algebraic difference approach 293.1. Introduction 293.2. Materials and methods 313.2.1. Data 313.2.2. Base models and GADA formulations 323.3. Results and discussion 353.3.1. General growth models with age 353.3.2. Growth models combined with stand density 413.3.3. Calibration and evaluation 463.4. Conclusion 49Chapter 4. Comparison of yield models in whole stand level 504.1. Introduction 504.2. Materials and methods 524.2.1. Data 524.2.2. Growth and yield model prediction 544.3. Results and discussion 574.3.1. Site index development for yield model 574.3.2. Normal yield model 594.3.3. Empirical yield model 624.3.4. Comparison of normal and empirical yield model 654.3.5. Variable-density growth and yield model 664.4. Conclusion 70Chapter 5. Prediction of diameter distribution models 715.1. Introduction 715.2. Materials and methods 735.2.1. Data 735.2.2. Functions for diameter distribution model 755.2.3. Prediction of parameter recovery 815.2.4. Diameter class removals model 815.3. Results and discussion 825.3.1. Ranking of distribution functions and parameter recovery methods 825.3.2. Prediction of quadratic mean diameter and percentile diameter equations 875.3.3. Diameter class removal model 905.4. Conclusion 92Chapter 6. Development of Gingrich-style stocking chart 936.1. Introduction 936.2. Materials and methods 956.2.1. Gingrich style stocking chart composition 956.2.2. Data 976.2.3. Developing stocking equations 996.3. Results and discussion 1036.3.1. Parameter prediction for Gingrich-style stocking charts 1036.3.2. Gingrich stocking and chart comparison 1066.3.3. A-level and B-level stocking 1116.4. Conclusion 112Chapter 7. General conclusions and recommendation 1137.1. Conclusions 1137.2. Contributions 1157.3. Recommendation 117Literature cited 118국문요약 154