Modeling Diameter Distributions of Loblolly Pine Plantations in Western Gulf Coastal Plain

Author:

Lou Xiongwei123ORCID,Weng Yuhui3ORCID,Fang Luming21,Grogan Jason3

Affiliation:

1. School of Information Engineering, Zhejiang A&F University, Lin’an, Zhejiang, China

2. Key Laboratory of State Forestry and Grassland Administration on Forestry Sensing Technology and Intelligent Equipment, Zhejiang A&F University, Lin’an, Zhejiang, China

3. Arthur Temple College of Forestry and Agriculture, Stephen F. Austin State University, Nacogdoches, TX, USA

Abstract

Abstract Diameter distribution models are useful tools for forest management planning, in particular for even-aged plantations of important commercial species such as loblolly pine. Using data collected from loblolly pine plantations across East Texas, two diameter distribution model systems were developed, with the first being a conventional, Weibull-form statistical model system and the second being developed using gradient boosting (GB) technique. Both models were tested using an independent data set and compared with the regional model currently being used, which was developed by Lee and Coble (2006). Compared with Lee and Coble (2006), the Weibull-form model of this study had 66.7% smaller prediction bias, 27.2% lower mean absolute error (MAE), and 18.9% smaller root-mean-square error (RMSE). Compared with the Weibull-form model of this study, the GB model had 33.9% lower MAE, 39.5% smaller RMSE, and greater R2. Thus, the GB model greatly outperformed the Weibull-form model, which, in turn, was greatly improved over the Lee and Coble (2006) in prediction accuracy. By combining a regional volume or weight equation, both proposed diameter distribution models can be used to predict stand wood volume or weight by diameter class. Both models, in particular the GB model, are recommended for use in predicting stand structures and developing stand and stock tables for loblolly pine plantations in the region. Study Implications Knowing future stand tree size distributions is important for forest management planning. This study developed two quantitative tools to predict diameter distributions for loblolly pine (Pinus taeda) plantations in the Western Gulf Coastal Plain, with one based on the Weibull function (Weibull-form model) and the other developed using the gradient boosting technique (GB model). For the Weibull-form model, using current stand information, readers can manually calculate future stand trees per acre by diameter class. Importantly, the Weibull-form model provides more accurate (less bias and more precise) prediction than any currently available models for loblolly pine in the region. The GB model, which uses the same predictors as the Weibull-form model, can achieve even better (similar bias but more precise) prediction than the Weibull-form model. However, no equations and model coefficients for the GB model were provided, and use of the GB model relies on computer programming. A computer program was developed to implement the GB model. We recommend use of both models, in particular of the GB model, in managing loblolly pine in the region. The results aid our understanding in loblolly pine stand structure development and management in the region.

Funder

Zhejiang Provincial Department of Science and Technology

Publisher

Oxford University Press (OUP)

Subject

Plant Science,Forestry

Reference32 articles.

1. Fertilized midrotation-aged slash pine plantations-stand structure and yield prediction models;Bailey;South. J. Appl. For,1989

2. Base-age invariant polymorphic site curves;Bailey;For. Sci.,1974

3. Quantifying diameter distributions with the Weibull function;Bailey;For. Sci.,1973

4. Loblolly Pine Growth and Yield Prediction for Managed West Gulf Plantations

Cited by 1 articles. 订阅此论文施引文献 订阅此论文施引文献,注册后可以免费订阅5篇论文的施引文献,订阅后可以查看论文全部施引文献

同舟云学术

1.学者识别学者识别

2.学术分析学术分析

3.人才评估人才评估

"同舟云学术"是以全球学者为主线,采集、加工和组织学术论文而形成的新型学术文献查询和分析系统,可以对全球学者进行文献检索和人才价值评估。用户可以通过关注某些学科领域的顶尖人物而持续追踪该领域的学科进展和研究前沿。经过近期的数据扩容,当前同舟云学术共收录了国内外主流学术期刊6万余种,收集的期刊论文及会议论文总量共计约1.5亿篇,并以每天添加12000余篇中外论文的速度递增。我们也可以为用户提供个性化、定制化的学者数据。欢迎来电咨询!咨询电话:010-8811{复制后删除}0370

www.globalauthorid.com

TOP

Copyright © 2019-2024 北京同舟云网络信息技术有限公司
京公网安备11010802033243号  京ICP备18003416号-3