Search

Statistical confidence in estimates of timber volume, carbon storage, and other forest attributes depends, in part, on the uncertainty in field measurements. Surprisingly, measurement uncertainty is rarely reported, even though national forest inventories routinely repeat field measurements for quality assurance. We compared measurements made by field crews and quality assurance crews in the Forest Inventory and Analysis program of the U.S. Forest Service, using data from 2790 plots and 51740 trees and saplings across the 24 states of the Northern Region. We characterized uncertainty in 12 national core tree-level variables; seven tree crown variables used in forest health monitoring; three variables describing seedlings; and 11 variables describing the site, such as elevation, slope, and distance from a road. Discrepancies in measurement were generally small but were higher for some variables requiring judgment, such as tree class, decay class, and cause of mortality. When scaled up to states, forest types, or the region, uncertainties in basal area, timber volume, and aboveground biomass were negligible. Understanding all sources of uncertainty is important to designing forest monitoring systems, managing the conduct of the inventory, and assessing the uncertainty of forest attributes required for making regional and national forest policy decisions. Key words: carbon accounting, Forest Inventory and Analysis, forest inventory, measurement uncertainty, forestry, 1. Introduction Forest inventories have historically been performed for many purposes (Labau et al. 2007), including estimating merchantable timber, determining risk of fire, evaluating habitat for wildlife, and assessing biodiversity. [...]

A system of nonlinear biomass component equations was developed for slash pine (Pinus elliottii Engelm. var. elliottii) trees using an econometric approach in which endogenous right-hand-side variables were included in some equations. The system was fitted to component biomass data from 306 slash pine trees sampled in the southeastern United States with weighted two-stage (2SLS) and three-stage (3SLS) least squares and full information maximum-likelihood (FIML) estimation methods. The predictive performances of the system fitted with these three estimation methods were ranked based on an array of statistics, and the ranking follows the order of FIML > 3SLS > 2SLS. The new system performed as well or better than previously published biomass equation systems developed using the aggregation and disaggregation approaches and fitted to the same data. The results demonstrated that the econometric approaches such as FIML and 3SLS have the potential to be useful for tree biomass modeling. Key words: southern pine, systems of equations, econometrics, carbon sequestration. Un systeme d'equations de composantes de la biomasse non lineaire a ete developpe pour les pins d'Elliot (Pinus elliottii Engelm. var. elliottii) en utilisant une approche econometrique dans laquelle les variables endogenes du cote droit ont ete incluses dans certaines equations. Le systeme a ete ajuste aux donnees des composantes de la biomasse de 306 pins d'Elliot echantillonnes dans le sud-est des Etats-Unis avec des moindres carres ponderes en deux etapes (2SLS) et en trois etapes (3SLS) et des methodes d'estimation du maximum de vraisemblance a informations completes (FIML). Les performances predictives du systeme ajuste avec ces trois methodes d'estimation ont ete classees en fonction d'un eventail de statistiques et le rang suit l'ordre du FIML > 3SLS > 2SLS. Le nouveau systeme a performe aussi bien ou mieux que les systemes d'equations de la biomasse publies anterieurement qui avaient ete developpes en utilisant les approches d'agregation et de desagregation et ajustes aux memes donnees. Les resultats ont demontre que les approches econometriques telles que le FIML et le 3SLS ont le potentiel d'etre utiles pour la modelisation de la biomasse des arbres. [Traduit par la Redaction] Mots-cles : pin du sud, systemes d'equations, econometrie, sequestration du carbone., Introduction Models for estimating the components of individual tree dry biomass generally include predictors for bole wood, bole bark, branches, and foliage. Several approaches have been used to develop prediction [...]

When conducting a forest inventory, sometimes portions of plots cannot be measured due to inaccessibility. Two primary methods have been presented to account for partial nonresponse in the estimation phase: (i) use a ratio-to-size estimator or (ii) apply an adjustment factor to all plot observations in proportion to the missing area. Both approaches provide identical estimates of the population mean, but the estimates of variance differ when partial nonresponse is present. The performance of variance estimators was examined for a range of population forest area and partial nonresponse proportions in the sample. The ratio-to-size variance estimator performed unbiasedly with respect to simulation results, but the adjustment factor variance estimates were biased, with magnitude and direction dependent upon the forest area proportion and amount of partial nonresponse. The bias is relatively small when the partial nonresponse is small, which is often the case; however, the ratio-to-size method is preferred to ensure accurate variance estimation for a wide range of circumstances. Key words: nonforest, ratio estimate, bias, correlation. Lors de la realisation d'un inventaire forestier, il arrive parfois que des portions de placettes ne puissent pas etre mesurees parce qu'elles sont inaccessibles. Deux methodes principales ont ete presentees pour tenir compte de l'absence partielle de resultat pendant la phase d'estimation : (i) utiliser un estimateur quotient selon la taille, ou (ii) appliquer un facteur d'ajustement proportionnel a la zone manquante a toutes les observations de la placette. Les deux approches produisent une estimation identique de la moyenne de la population, mais l'estimation de la variance est differente lorsqu'il y a une absence partielle de resultat. La performance de l'estimateur de la variance a ete examinee en fonction d'une gamme de superficies forestieres et de proportions d'absence partielle de resultat dans l'echantillon. L'estimateur quotient selon la taille de la variance produisait des resultats de simulation non biaises, mais les estimations de la variance provenant du facteur d'ajustement etaient biaisees en fonction de l'amplitude et de la direction des resultats qui variaient selon la proportion de la superficie forestiere et de la quantite d'absence partielle de resultat. Le biais est relativement petit lorsque l'absence partielle de resultat est de petite taille, ce qui est souvent le cas. Malgre tout, la methode de l'estimateur quotient selon la taille est preferee pour s'assurer d'une estimation precise de la variance pour une large gamme de circonstances. [Traduit par la Redaction] Mots-cles: absence de foret, estimation quotient, biais, correlation., Introduction Various factors affect the specification of field plot measurement protocols in forest inventories. For large-area inventories in which many different land characteristics are likely to be encountered, a mapped-plot [...]

Post-stratification is often used to increase the precision of estimates arising from large-area forest inventories with plots established at permanent locations. Remotely sensed data and associated spatial products are often used for developing the post-stratification, which offers a mechanism to increase precision for less cost than increasing the sample size. While important variance reductions have been shown from post- stratification, it remains unknown where observed gains lie along the continuum of possible gains. This information is needed to determine whether efforts to further improve post-stratification outcomes are warranted. In this study, two types of optimal post-stratification were compared with typical production-based post-stratifications to estimate the magnitude of remaining gains possible. Although the optimal post-stratifications were derived using methods inappropriate for operational usage, the results indicated that substantial further increases in precision for estimates of both forest area and total tree biomass could be obtained with better post-stratifications. The potential gains differed by the attribute being estimated, the population being studied, and the number of strata. Practitioners seeking to optimize post-stratification face challenges, such as evaluation of numerous auxiliary data sources, temporal misalignment between plot observations and remotely sensed data acquisition, and spatial misalignment between plot locations and remotely sensed data due to positional errors in both data types. Key words: relative efficiency, sampling error, remote sensing, optimal stratification. La stratification a posteriori est souvent utilisee pour augmenter la precision des estimations qui decoulent des resultats d'inventaires forestiers a grande echelle realises a partir de places echantillons permanentes. Des donnees de teledetection et des produits spatiaux associes sont souvent utilises pour elaborer la stratification a posteriori, laquelle offre un mecanisme pour augmenter la precision a un cout moindre que l'augmentation de la taille de l'echantillon. Bien que d'importantes reductions de la variance dues a la stratification a posteriori aient ete demontrees, on ne sait toujours pas ou se situent les gains observes le long du continuum des gains potentiels. Cette information est necessaire pour determiner si des efforts pour ameliorer davantage les resultats de la stratification a posteriori sont justifies. Dans cette etude, deux types de stratification a posteriori optimale ont ete compares a des stratifications a posteriori typiques fondees sur la production pour estimer l'ampleur des gains potentiels restants. Bien que les stratifications a posteriori optimales aient ete derivees a l'aide de methodes inappropriees pour un usage operationnel, les resultats indiquent que d'importants gains supplementaires de precision dans l'estimation de la superficie forestiere et de la biomasse totale des arbres peuvent etre obtenus avec de meilleures stratifications a posteriori. Les gains potentiels different selon l'attribut estime, la population etudiee et le nombre de strates. Les praticiens qui cherchent a optimiser la stratification a posteriori font face a des defis tels que l'evaluation de nombreuses sources de donnees auxiliaires, le defaut d'alignement temporel entre les observations des places echantillons et l'acquisition des donnees de teledetection, ainsi que le mauvais alignement spatial entre la localisation des places echantillons et les donnees de teledetection a cause d'erreurs positionnelles dans les deux types de donnees. [Traduit par la Redaction] Mots-cles: efficacite relative, erreur d'echantillonnage, teledetection, stratification optimale., Introduction Most forest inventory programs, including both national forest inventories (NFI) and greenhouse gas inventories, seek explicit means of increasing the precision of estimates they report. Specifically, because the combination [...]

Both aggregative and disaggregative strategies were used to develop additive nonlinear biomass equations for slash pine (Pinus elliottii Engelm. var. elliottii) trees in the southeastern United States. In the aggregative approach, the total tree biomass equation was specified by aggregating the expectations of component biomass models, and their parameters were estimated by jointly fitting all component and total biomass equations using weighted nonlinear seemingly unrelated regression (NSUR) (SUR1) or by jointly fitting component biomass equations using weighted NSUR (SUR2). In an alternative disaggregative approach (DRM), the biomass component proportions were modeled using Dirichlet regression, and the estimated total tree biomass was disaggregated into biomass components based on their estimated proportions. There was no single system to predict biomass that was best for all components and total tree biomass. The ranking of the three systems based on an array of fit statistics followed the order of SUR2 > SUR1 > DRM. All three systems provided more accurate biomass predictions than previously published equations.Key words: nonlinear seemingly unrelated regression, Dirichlet regression, biomass additivity, heteroscedasticity.Des strategies d'agregation et de desagregation ont ete utilisees pour developper des equations de biomasse non lineaires additives pour le pin d'Elliott typique (Pinus elliottii Engelm. var. elliottii) dans le sud-est des Etats-Unis. Dans l'approche agregative, l'equation de la biomasse totale des arbres a ete etablie en agregeant les previsions des modeles des composantes de la biomasse, et leurs parametres ont ete estimes en ajustant conjointement toutes les equations de la biomasse totale et de ses composantes a l'aide de regressions non lineaires ponderees apparemment non reliees (NSUR) (SUR1) ou par ajustement conjoint des equations des composantes de la biomasse a l'aide du NSUR pondere (SUR2). Dans une autre approche, par desagregation (DRM), les proportions des composantes de la biomasse ont ete modelisees a l'aide de la regression de Dirichlet et la biomasse totale estimee des arbres a ete desagregee en composantes de la biomasse en fonction de leurs proportions estimees. Aucun des systemes permettant de predire la biomasse s'est avere le meilleur pour toutes les composantes et la biomasse totale des arbres. Le classement des trois systemes, sur la base d'un ensemble de statistiques d'ajustement, suivait l'ordre suivant : SUR2 > SUR1 > DRM. Les trois systemes ont fourni des previsions de biomasse plus precises que les equations publiees precedemment. [Traduit par la Redaction]Mots-cles : regressions non lineaires apparemment non reliees, regression de Dirichlet, additivite de la biomasse, heteroscedasticite., IntroductionSlash pine (Pinus elliottii Engelm. var. elliottii) is an important commercial timber species in the southeastern United States (US). Slash pine has been planted on more than 4.2 million ha, [...]

Understanding the patterns of tree recruitment is essential to quantifying the future health and productivity of forest ecosystems. Using national forest inventory information, we incorporated browse impact measurements into models of sapling (2.5-12.7 cm diameter at breast height (DBH)) and overstory tree (>12.7 cm DBH) ingrowth across the northern United States. Ingrowth was modeled with standard and zero-inflated techniques using discrete Poisson and negative binomial distributions. Zero-inflated models using stand attributes and browse impacts provided the best fit statistics for modeling the occurrence and frequency of ingrowth over a 5-year time period. Results indicate that stands with very high browse impact would contain 50.0% fewer ingrowth saplings compared with stands with no browse impact. Greater browse impacts similarly yielded a negative effect on overstory tree ingrowth, but to a lesser degree than saplings. Despite the stochastic nature of ingrowth observations, incorporating browse impacts may be essential in determining accurate levels of ingrowth in forests where herbivory constrains forest regeneration objectives. Key words: herbivory, white-tailed deer, regeneration, ingrowth, Forest Inventory and Analysis. Il est essentiel de comprendre les patrons de recrutement des arbres pour quantifier l'etat de sante et la productivity futures des ecosystemes forestiers. A partir des informations tirees de l'inventaire forestier national, nous avons incorpore des mesures d'impact du broutage dans des modeles de recrutement des gaules (2,5- 12,7 cm de diametre a hauteur de poitrine (DHP)) et des tiges ([greater than or equal to] 12,7 cm au DHP) dans l'etage dominant a travers le nord des Etats-Unis. Le recrutement a ete modelise a l'aide de techniques standards et avec une surrepresentation de zeros en utilisant les distributions de Poisson et binomiale negative. Les modeles avec une surrepresentation de zeros utilisant les attributs du peuplement et les impacts du broutage fournissaient les statistiques les mieux ajustees pour modeliser l'occurrence et la frequence du recrutement sur une periode de 5 ans. Les resultats indiquent que les peuplements severement impactes par le broutage contiendraient 50,0 % moins de recrues comparativement aux peuplements exempts de broutage. Des impacts plus importants dus au broutage ont de facon similaire eu un effet negatif sur le recrutement des tiges dans l'etage dominant, mais a un degre moindre que dans le cas des gaules. Malgre la nature stochastique des observations au sujet du recrutement, l'introduction des impacts du broutage peut s'averer essentiel pour determiner le degre exact de recrutement dans les forets ou le broutage menace les objectifs de regeneration forestiere. [Traduit par la Redaction] Mots-cles: broutage, cerf de Virginie, regeneration, recrutement, Analyse et Inventaire Forestier., Introduction Tree recruitment is the accretion of newly established trees of a certain size in forest stands. Quantifying tree recruitment is essential to forest managers and is an integral component [...]

In forest inventories, it is often of interest to calculate amounts of usable wood volume in trees. This usually requires knowledge of how much of the total volume is unusable (cull) due to form or decay deficiencies. This information is primarily obtained when collecting data on sample plots, although the assessments are often difficult and subjective. To provide an alternative, methods were developed to estimate individual-tree cull attributes. The procedure initially involves classification using logistic regression to assign trees to one of three categories (no cull, intermediate cull, entirely cull) based on probability cut points. Subsequently, trees classified as having intermediate cull are assigned a cull amount predicted from a generalized linear regression model. The best results for cull prediction were obtained using cut points that minimized absolute prediction error; however, better prediction of net cubic volume of trees was realized when the errors were weighted by tree size. The model-based approach may be particularly useful in obtaining temporal consistency, such that trend estimates may better reflect the actual change in forest resources. Resume: Dans les inventaires forestiers, on est souvent interesse a calculer le volume de bois utilisable dans les arbres. Cela exige habituellement de connaitre la proportion du volume total qui n'est pas utilisable (rebut) a cause de la forme ou de la carie. Cette information est obtenue principalement lors de la collecte de donnees dans des placettes echantillons mais l'evaluation est souvent difficile et subjective. Dans le but d'offrir une alternative, des methodes ont ete developpees pour estimer les attributs du rebut d'arbres individuels. La procedure implique d'abord une classification a l'aide d'une regression logistique pour repartir les arbres dans trois categories (pas de rebut, partiellement rebut, totalement rebut) sur la base de points de decoupage des probabilites. Par la suite, les arbres places dans la classe intermediaire se voient attribuer une quantite de rebut predite par un modele de regression generalise. Les meilleurs resultats pour la prediction de la quantite de rebut ont ete obtenus en utilisant des points de decoupage qui minimisent l'erreur de prediction absolue. Cependant, la prediction du volume net des arbres etait meilleure lorsque les erreurs etaient ponderees par la dimension des arbres. L'approche fondee sur une modele peut etre particulierement utile pour assurer une coherence temporelle de telle sorte que les estimations de la tendance puissent mieux refleter les changements reels dans les ressources forestieres. [Traduit par la Redaction], Introduction One of the most important aspects of forest inventory is estimating tree volume. More specifically, of interest is determining the amount of usable material for specified wood products. Two [...]

Post-stratification is used to reduce the variance of estimates of the mean. Because the stratification is not fixed in advance, within-strata sample sizes can be quite small. The survey statistics literature provides some guidance on mini-mum within-strata sample sizes; however, the recommendations and justifications are inconsistent and apply broadly for many different population structures. The impacts of minimum within-strata and total sample sizes on estimates of means and standard errors were examined for two forest inventory variables: proportion forestland and cubic net volume. Estimates of the means seem unbiased across a range of minimum within-strata sample sizes. A ratio that described the decrease in variability with increasing sample size allowed for assessment of minimum within-strata sample requirements to obtain stable estimates of means. This metric indicated that the minimum within-strata sample size should be at least 10. Estimates of standard errors were found to be biased at small total sample sizes. To obtain a bias of less than 3%, the required minimum total sample size was 25 for proportion forestland and 75 for cubic net volume. The results presented allow analysts to deter-mine within-stratum and total sample size requirements corresponding to their criteria for acceptable levels of bias and variability. La post-stratification est utilisee pour reduire la variance des estimations de la moyenne. Comme la stratification n'est pas fixee a l'avance, la taille de l'echantillon dans les strates peut etre tres faible. La litterature qui porte sur les resul-tats d'enquete fournit des indications sur la taille minimale de l'echantillon dans les strates. Mais les recommandations et les justifications sont inconsistantes et s'appliquent de faCon generale a plusieurs structures de population differentes. Les impacts de la taille minimale de l'echantillon dans les strates et de la taille totale de l'echantillon sur les estimations de la moyenne et de l'erreur-type ont ete examines pour deux variables d'inventaire forestier: la proportion de terres forestieres et le volume net de bois. Les estimations de la moyenne semblent non biaisees pour un eventail de tailles minimales de l'e-chantillon dans les strates. Une metrique qui decrit la diminution de la variabilite avec l'augmentation de la taille de l'echantillon a permis de determiner la taille minimale de l'echantillon dans les strates pour obtenir des estimations stables de la moyenne. La metrique a montre que la taille minimale de l'echantillon dans les strates devrait etre d'au moins 10. Les estimations de l'erreur-type etaient biaisees lorsque la taille totale de l'echantillon etait petite. La taille totale minimale de l'e-chantillon requise pour obtenir un biais inferieur a 3 % etait de 25 pour la proportion de terres forestieres et de 75 pour le volume net de bois. Ces resultats permettent aux analystes de determiner la taille totale de l'echantillon et la taille de l'echantillon dans la strate qui correspondent a leurs criteres pour obtenir des niveaux acceptables de biais et variabilite. [Traduit par la Redaction], Introduction Stratification is a technique used in sample survey design to improve the precision of population estimates (Holt and Smith 1979). In stratified sampling, units of the population having similar [...]