English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Tools annual

Global: Zendy Free Plan

Global: Zendy Tools monthly

Global: Zendy Plus monthly

The accuracy of ten growth and yie ld models for Southern Appalachian upland hardwood forests and southern bot tomlandforests was evaluated. In technical appl icat ions, accuracy is the composi te both bias (average error) andprecis ion. Resul ts indicate that GHAT, NATPIS, anda local ly cal ibrated version of NETWIGS may be regarded as being operat ional ly val id growth and yield models for Southern Appalachian yellow-poplar (Liriodendron tulipifera) and mixed oak (Quercus spp.) forests that fall within the range of characteris t ics of the test data set . No publicly avai lable growth and yield models speci f ical ly developed f o r southern bot tomland hardwoodforests exis t , Four general models that contain most of the applicable species to predic t growth of these forests were tested. SETWIGS was found to be the most accurate of the four models tested and is recommended for use if the reported level of accuracy is acceptable and the target stand character is t ics fa l l wi th in the range of our tes t data set . Resul ts indicate that the growth and densi ty dynamics of dense, young stands of both upland and bot tomland hardwoods were poorly predicted by the models . Models predicted basal area and densi ty changes in yel low-poplar s tands more accurately than mixed hardwoods. Predict ions for upland hardwoods were more accurate than those for bot tomland hardwoods. Model accuracy uni formly decreases wi th increasing length of the projection period. South. J . Appl. For. 24(3):176-l&5. B ecause the accuracy of growth and yield model projections is l ikely to affect the quali ty of forest management decisions, models should be evaluated in order to build confidence in their validity (Vanclay et al. 1996). Rykiel(1996) identified three types of model evaluation: (1) operational validation, (2) conceptual validation, and (3) data validation. Operational validation tests whether the model output conforms with its stated purpose. Conceptual validation evaluates NOTE: H. Michael Rauscher is the corresponding author, and he can be reached at (828) 667-5261 ext. 102; E-mail: mrauscher@fs.fed.us. Longterm forest research plots tend to have a high front end cost with little immediate reward to individuals who design the study and establish the plots. Benefits are often reaped long after the initiators have retired. This study could not have been done without the vision and hard work of Lino Della-Bianca (retired USDA Forest Service), Donald E. Beck (retired USDA Forest Service), Ralph M. Hooper (retired USDA Forest Service), Robert Kellison (retired NC-State Hardwood Research Cooperative),Virginia Gibbs, Tracy Roof and Julia Murphy, forestry research technician, Bent Creek Experimental, Asheville, NC. The plot network within the NC State Hardwood Research Cooperative could not have been possible without the support, and commitment of its industrial and state agency members throughout the South since 1963. We thank Carole Beam, research statistician with the NC State-Hardwood Research Cooperative, for her hard work in the data analysis phase of this project. Our thanks to the following reviewers: Marilyn Buford, Harold Burkhart, George Gertner, Jeff Goelz, Thomas Lynch, Ralph Meldahl, Richard Oderwald, EdwardRykiel, John Stanturf, Susan Stout, and four anonymous reviewers. Manuscript received September 13, accepted January 3 1,200O. whether the model provides a scientifically acceptable explanation of the causal relationships included in the model. Finally, data validation checks whether the test data accurately represent the system of interest. Forest growth and yield models, unlike many ecological simulation models, are primarily intended for use by forest managers in the decisionmaking process. Therefore, operational validation is of primary importance and will be considered in this article. Conceptual val idat ion is , in this context , a secondary consideration and is not addressed here. The validity of data is particularly important in the context of model evaluation. The operational validation process for growth and yield models may be viewed from two perspectives: (1) hypothesis testing and (2) confidence interval estimation (Reynolds 1984, Rykiel 1996). Hypothesis testing procedures are the appropriate tools to determine whether the model mimics the real world by passing some predefined level of accuracy. Confidence interval estimation establishes the degree of confidence that can be placed in model predictions through estimates of the direction, magnitude, and variability of the prediction error (Reynolds 1984). Because generally accepted standards of accuracy for growth and yield models do not exist and vary from one situation to another, the confidence interval perspective to operational validation is fre176 Reprinted from the Southern Journal @Applied Forestry, Vol. 24, No. 3, August 2000. Not for further reproduction. quently more practical than the hypothesis test ing perspective and is the approach used in this study (Goodall 1972). Objectives of this article are to test the accuracy of predicting the changes in basal area growth and stand density in: (1) ten growth and yield models for Southern Appalachian upland hardwood forests, and (2) four growth and yield models for southern bot tomland hardwood forests . The Southern Appalachian region covers approximately 37 million ac in parts of seven states: Virginia, North Carolina, Tennessee, Kentucky, South Carolina, Georgia, and Alabama (SAMAB 1996). Forests currently cover 70% of the region. The Southern Appalachian upland hardwood part of this study concentrates on the 16 million ac of the region in either mixed oak (Quercus spp.) or yellow-poplar (Liriodendron tul ipiferu> forest. The southern bottomland hardwood part of this study concentrates on the approximately 23 million ac in the eight Southern states (Alabama, Arkansas, Georgia, Louisiana, Mississippi, North Carolina, South Carolina, and Virginia) wetland hardwood forest types (USDA 1988). Bottomland sites are considered those in which 50% or more of the forest s tand is comprised of tupelo (Nyssa spp.), sweetgum (Liquidambar styraciflua), oak (Quercus spp.), and cypress (Taxodium spp.), singly or in combination, with pines (Pinus spp.) contributing less than 25% of the stand. There is, however, a good deal of discrepancy in how wetland forests are defined and measured (Cubbage and Flather 1993).

Testing the Accuracy of Growth and Yield Models for Southern Hardwood Forests