Pith. sign in

REVIEW 2 major objections 5 minor 71 references

Small area estimation of growing stock timber volume, basal area, mean stem diameter, and stem density for mountain forests in Austria

T0 review · 2 major / 5 minor · reviewed 2026-08-07 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read The paper shows that ALS-derived height predictors, not spatial random effects or multivariate structure, are what make model-based stand-level estimates accurate in a rugged mountain forest.

desk verdict Solid applied SAE comparison with a believable headline result, but the spatial cross-validation design leaves the "spatial gains are modest" conclusion less clean than the paper implies. read the letter →

arxiv 2506.05043 v2 pith:DHEMLHGE submitted 2025-06-05 stat.AP

Arne Nothdurft , Valentin Sarkleti , Tobias Ofner-Graff , Andreas Tockner , Christoph Gollob , Tim Ritter , Ralf Kra{ss}nitzer , Philip Svazek
show 3 more authors
Martin Kühmaier Karl Stampfer Andrew O. Finley
This is my paper
classification stat.AP MSC 62M3062F1562P12
keywords Bayesianinferencesmallareaestimationforestinventoryairbornelaserscanningspatialregressiongrowingstockvolumestemdensitymountainforests
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper asks how much model complexity is actually needed to turn 146 handheld-laser-scanned field plots into reliable estimates for 824 management stands in a rugged Austrian mountain forest. It compares univariate and multivariate Bayesian regression models, with and without spatial random effects, using airborne laser scanning (ALS) height predictors, and it generates stand-level estimates by aggregating unit-level predictions on a fine grid. The central result is that the ALS predictors carry nearly all of the predictive accuracy: adding them cuts cross-validated RMSPE by about half for growing stock volume and quadratic mean diameter, while spatial random effects and multivariate structure add only modest gains. The practical conclusion is that a relatively simple, ALS-informed model can deliver uncertainty-quantified stand-level maps of timber volume, basal area, mean stem diameter, and stem density for operational planning.

What carries the argument

The load-bearing machinery is a unit-level Bayesian small area estimator that predicts each outcome on a uniform 26.36 m grid and aggregates prediction-unit posterior samples to stand level with area weights, so each stand gets a full posterior predictive distribution. The regression core is a Gaussian linear model with an exponential spatial Gaussian process for random effects; the multivariate version couples outcomes through a linear model of coregionalization in the spatial cross-covariance and an unstructured residual covariance matrix. The identity $N = \mathrm{BA} / \big((\pi/4)(\mathrm{QMD}/100)^2\big)$ converts posterior samples of basal area and quadratic mean diameter into posterior samples of stem density, giving all four outcomes a joint posterior. Predictive performance is assessed with spatially blocked cross-validation using 250 m blocks and 20 folds, intended to prevent nearby training plots from leaking information about held-out plots.

What would settle it

Re-run the 20-fold cross-validation with spatial blocks of at least 2 km (larger than the estimated effective ranges) or with distance buffers stripped out around held-out blocks. If the RMSPE gap between spatial and non-spatial models shrinks to zero, the conclusion that spatial random effects add little holds; if it widens, the modest-gains conclusion is an artifact of the 250 m blocks.

Watch

Extended reading notes

Core claim

On the paper's own terms, the discovery is that unit-level Bayesian small area estimation with ALS-derived predictors gives accurate, uncertainty-quantified stand-level estimates in the Brixen im Thale forest district, and the extra machinery of spatial random effects and multivariate outcome modeling is not what earns the accuracy. In spatially blocked cross-validation, adding ALS predictors cuts root mean squared prediction error by roughly half for growing stock volume and quadratic mean diameter and by a quarter to a third for basal area, whereas adding spatial random effects to predictor-informed models improves RMSPE only slightly. No single model wins across all outcomes and metrics, and the multivariate spatial model performs about as well as univariate spatial models, though it preserves the observed correlation between growing stock volume and basal area better. Stem density is not modeled directly; it is derived from predicted basal area and quadratic mean diameter through the identity $N = \mathrm{BA} / \big((\pi/4)(\mathrm{QMD}/100)^2\big)$, with uncertainty propagated through posterior predictive samples.

Load-bearing premise

The evaluation assumes that holding out whole 250-meter blocks keeps training and held-out plots independent, even though the estimated spatial correlation range reaches about a kilometer; if nearby blocks resemble each other, the comparison between spatial and non-spatial models is not clean.

Editorial extensions

If this is right

  • Forest managers can rely on ALS-informed univariate spatial models, and in some cases non-spatial models, to produce stand-level estimates for all 824 stands from just 146 plots.
  • Collecting and processing airborne laser scanning data should be prioritized over adding spatial structure to the model, because predictors cause most of the RMSPE reduction.
  • The identity linking basal area, quadratic mean diameter, and stem density lets practitioners report stem density with quantified uncertainty without fitting a separate count model.
  • The mostly sub-20% coefficients of variation in the prediction maps show which stands are estimated precisely enough to support silvicultural and harvest planning.
  • When all outcomes are observed, multivariate spatial models are not needed for point predictions; they remain useful mainly when preserving outcome correlations matters or when data are spatially misaligned.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • If the estimated effective spatial ranges (up to about 1.1 km) are correct, the 250 m blocks may not fully isolate held-out blocks, so the measured advantage of spatial models could be biased; repeating the cross-validation with larger blocks would clarify whether spatial effects are truly negligible.
  • The aggregation-plus-identity pipeline should transfer to other ALS-covered forest districts with sparse plots; a direct test would compare model-based stand estimates against independent field measurements in a new district.
  • In settings where some outcomes are unobserved at some plots, multivariate spatial models should show their value because they can borrow strength across outcomes; the paper notes this promise but does not test it.
  • The two plot radii (20 m and 10 m) mean small plots may sample stands unevenly; weighting plot outcomes by effective sampled area, beyond using a common predictor-extraction radius, is a natural robustness check the paper leaves open.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

2 major / 5 minor

Summary. The paper develops Bayesian unit-level small area estimation models for stand-level growing stock volume (GSV), quadratic mean diameter (QMD), basal area (BA), and stem density (N) in a 548 ha Austrian mountain forest district, using ALS-derived predictors and field measurements from 146 mobile personal laser scanning plots. Four candidate model classes are compared: univariate and multivariate regressions, each with and without spatial random effects, and each in intercept-only and all-predictors versions. Predictions are made on a fine grid and aggregated to 824 stands, with N derived deterministically from BA and QMD via the identity N = BA / ((pi/4)*(QMD/100)^2). Predictive performance is assessed through 20-fold spatially blocked cross-validation with 250 m hexagonal blocks. The main empirical finding is that ALS-derived predictors drive predictive performance, while spatial random effects and multivariate structure yield only modest gains; the authors therefore recommend simpler univariate spatial, and sometimes non-spatial, models for operational use.

Significance. If the central claim holds, the paper offers practically useful guidance for precision forestry in mountainous terrain: high-resolution ALS predictors are more valuable than complex spatial or multivariate random-effect structures when all outcomes are observed at all plots. The study is valuable as an applied comparison of Bayesian hierarchical spatial models for small area estimation, and it is strengthened by the availability of code and data on GitHub, by the explicit treatment of posterior predictive distributions for derived quantities such as N, and by the use of spatially blocked cross-validation to avoid optimistic random-fold comparisons. The main caveat is that the cross-validation design may not faithfully represent the distance regime and aggregation scale that matter for stand-level operational predictions, which weakens the force of the 'modest spatial gains' recommendation.

major comments (2)
  1. [Section 2.5.1, Table 5] The spatially blocked cross-validation uses 250 m hexagonal blocks, justified as the average distance between stand centroids and nearest plots, but the estimated effective spatial ranges in Tables 3 and 4 reach about 0.34-1.1 km for the all-predictors models and the posterior intervals are wide. Under an exponential correlation function, a 1.1 km effective range implies correlations of roughly 0.5 at 250 m and above 0.7 at 125 m, so held-out blocks are not spatially independent of the training data. This can inflate the apparent performance of spatial models in the cross-validation. More importantly, the target stands are much smaller than the blocks (median stand area 0.36 ha, about 60 m across), and operational predictions for stands that contain or are adjacent to inventory plots require spatial borrowing at distances well below the 125-250 m minimum separation enforced by the block design. The reported conclusion that spatial random effects provide only modest gains may therefore be an artifact of evaluating spatial models in their least favorable distance regime. I recommend reporting cross-validation at multiple block sizes (e.g., 125 m, 500 m, 1 km) and/or stratifying predictive errors by the distance from each prediction unit to the nearest training plot, so that the distance regime relevant to actual stands is tested.
  2. [Section 2.2] The predictor set was selected using stepAIC on the full dataset before cross-validation. This means information from the held-out folds is used to fix the model structure for the all-predictors models, which can bias the cross-validated RMSPE and coverage statistics and overstate the predictive advantage of the ALS-derived predictors relative to the intercept-only models. Because the comparison between spatial and non-spatial models uses the same selected predictors, this issue does not by itself overturn the 'modest spatial gains' conclusion, but it weakens the strength of the model-comparison evidence. The authors should either repeat the predictor selection inside each cross-validation fold or provide a sensitivity analysis showing that the conclusions are unchanged under a pre-specified or fold-internal selection.
minor comments (5)
  1. [Section 2.5.1] The cross-validation aggregates predictions to 20 block-level 'stand analogues', but each block contains about 7 plots and is far larger than the median stand of 0.36 ha. The block-level RMSPE therefore measures performance for aggregates of several plots, not for the 824 target stands; the unit-level results in Table A.2 partly mitigate this, but the authors should explicitly discuss the mismatch between block size and stand size when interpreting the stand-level performance claims.
  2. [Section 2.1] The plot measurements are treated as error-free response variables, although the PLS pipeline has documented detection rates of 96-98.5% and a height RMSE of 2.6 m near treetops. Measurement error in the responses does not bias point predictions under the Gaussian model, but it should be discussed as a source of over-optimistic predictive uncertainty, particularly for the derived outcome N.
  3. [Section 2.1] The inventory plots have two different radii (59 plots at 20 m, 87 plots at 10 m), while the predictor variables are extracted over a constant circular area with radius 14.87 m. This change-of-support mismatch between response and predictor footprints, combined with the varying plot sizes, creates heteroscedastic measurement error that is not accounted for in the models; a brief discussion or sensitivity analysis would help.
  4. [Figure 3] The caption of Figure 3 describes the panels as 'residuals' but the diagonal panels are labeled as distributions of spatial random effects; Figure A.1 uses the same caption structure for actual residuals. The captions should be clarified to distinguish these two quantities.
  5. [Section 3.3] The 100% empirical coverage rates for several models in Table 5 and the very wide 95% CI ranges for N suggest that the interval estimates for the derived outcome are conservative; the paper notes this only implicitly, and a short discussion of why N intervals are so wide would be informative.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the central model-comparison conclusion is supported by held-out spatial-block cross-validation, and the N derivation is an explicit algebraic identity applied after prediction rather than an input to fitting.

full rationale

The paper's main claim—that ALS-derived predictors dominate predictive performance while spatial random effects and multivariate structure add only modest gains—is an empirical outcome of comparing candidate models under 20-fold spatially blocked cross-validation. The RMSPE, bias, and coverage metrics in Table 5 are computed from held-out folds in which all plots within a block are withheld, so the comparison between spatial and non-spatial models is not forced by construction. The derivation of stem density N from BA and QMD in Eq. (4) is a transparent algebraic identity: N = BA / ((pi/4)*(QMD/100)^2). This identity is applied only after posterior predictive samples of BA and QMD are drawn, and it is not used to estimate BA or QMD; therefore it does not make the N predictions circular, even though N is a derived rather than directly modeled outcome. Self-citations to the spBayes package (Finley et al. 2007, 2015), to Finley et al. (2008) for the multivariate coregionalization construction, and to the authors' treeX package are standard methodology and tooling citations; the associated code and data are provided in a public repository, and no load-bearing argument reduces to an unverified self-citation. The only notable concern is whether 250 m spatial blocks are large enough to prevent spatial leakage given estimated effective ranges up to about 1.1 km, but that is a validity threat to the evaluation design, not a circularity in the derivation chain. No fitted parameter is renamed as a prediction, no uniqueness theorem is imported from the authors' prior work, and no ansatz is smuggled in via self-citation. The paper's conclusions therefore stand as an independent empirical finding rather than a restatement of its inputs.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

The central claim rests on standard Bayesian spatial regression machinery and on assumptions about data quality and CV design; no new entities are introduced. The most load-bearing assumptions are error-free PLS measurements and leakage-free spatial blocking.

free parameters (3)
  • Spatial decay parameters (effective range) = 0.34-1.2 km across outcomes
    Estimated in models; determines spatial borrowing strength and conclusion that spatial effects are modest.
  • Regression coefficients for DVHM/DTM predictors = e.g., beta_Mean = 0.080 (GSV), beta_Elev = 0.064 (GSV)
    Fitted via MCMC; quantify the predictive contribution of ALS predictors.
  • Variance components (tau^2, sigma^2, Psi) = e.g., GSV all-predictors spatial: tau^2=0.047, sigma^2=0.056
    Fitted residual and spatial variances; used in R2 and uncertainty intervals.
assumptions (4)
  • domain assumption Outcomes are Gaussian after log transformation
    Underlies Eqs. (5)-(6); supports PPD sampling but not verified for all outcomes.
  • domain assumption Spatial random effects follow stationary isotropic exponential GP
    Section 2.3; stationarity checked only indirectly via coverage rates.
  • domain assumption Plot-level PLS measurements are error-free
    Section 2.1 reports detection rates and height RMSE but these are not propagated.
  • ad hoc to paper 250 m spatial blocks remove cross-validation leakage
    Section 2.5.1; effective ranges exceed block size, so leakage is possible.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Small area estimation of growing stock timber volume, basal area, mean stem diameter, and stem density for mountain forests in Austria." pith.science (2026). https://pith.science/paper/DHEMLHGE

@misc{pith2026250605043,
  author       = {Pith},
  title        = {Pith review of: Small area estimation of growing stock timber volume, basal area, mean stem diameter, and stem density for mountain forests in Austria},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/DHEMLHGE}},
  note         = {Machine review of arXiv:2506.05043}
}
read the original abstract

Regression models were evaluated to estimate stand-level growing stock volume (GSV), quadratic mean diameter (QMD), basal area (BA), and stem density (N) in the Brixen im Thale forest district of Austria. Field measurements for GSV, QMD, and BA were collected on 146 inventory plots using a handheld mobile personal laser scanning system. Predictor variables were derived from airborne laser scanning (ALS)-derived normalized digital surface and terrain models. The objective was to generate stand-level estimates and associated uncertainty for GSV, QMD, BA, and N across 824 stands. A unit-level small area estimation framework was used to generate stand-level posterior predictive distributions by aggregating predictions from finer spatial scales. Both univariate and multivariate models, with and without spatially varying intercepts, were considered. Predictive performance was assessed via spatially blocked cross-validation, focusing on bias, accuracy, and precision. Despite exploratory analysis suggesting advantages of complex multivariate spatial models, simpler univariate spatial -- and in some cases, non-spatial -- models exhibited comparable predictive performance.

Figures

Figures reproduced from arXiv: 2506.05043 by the authors.

Figure 1
Figure 1. Location and extent of the 824 forest stands (grey boundaries) in the forest district Brixen im Thale and locations of the 146 sample points (red dots). ner et al. (2022), and Ritter et al. (2017, 2020). Stem volume was calculated using a traditional stem-form function (Pollansch¨utz, 1965). Individual tree detection and tree segmentation from ground-based LiDAR was conducted using the treeX R software package (Tock… view at source ↗
Figure 2
Figure 2. Hexagonal blocks used to partition the study area into spatially distinct folds for 20- fold spatial cross-validation. Plots shown in [PITH_FULL_IMAGE:figures/full_fig_p019_2.png] view at source ↗
Figure 3
Figure 3. Summaries of residuals from univariate spatial all-predictors candidate models. Di￾agonal panels show the distribution of each model’s spatial random effect; lower triangle panels display pairwise scatter plots between models; and upper triangle panels report the corresponding Pearson correlation coefficients. Asterisks next to the correlation estimates indicate frequentist statistical significance: ** p < 0.01, and… view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: Posterior predictive distribution (PPD) mean and coefficient of variation for the four outcomes: growing stock timber volume (GSV), quadratic mean stem diameter (QMD), basal area (BA), and stem density (N) for forest stands in Brixen im Thale. Panels in the top row sho…

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

71 extracted references · 70 canonical work pages

  1. [1]

    Ashton, M. S. and Kelty, M. J. (2018). The Practice of Silviculture: Applied Forest Ecology . John Wiley & Sons

    work page 2018

  2. [2]

    Assmann, E. (1950). Grundflächen- und Volumenzuwachs der Rotbuche bei verschiedenen Durchforstungsgraden . Forstwiss. Centralbl. , 69:256--286

    work page 1950

  3. [3]

    Assmann, E. (1954). Grundflächenhaltung und Zuwachsleistung Bayerischer Fichten-Durchforstungsreihen . Forstwiss. Centralbl. , 73:257--271

    work page 1954

  4. [4]

    O., Andersen, H.-E., Pattison, R., Cook, B

    Babcock, C., Finley, A. O., Andersen, H.-E., Pattison, R., Cook, B. D., Morton, D. C., Alonzo, M., Nelson, R., Gregoire, T., Ene, L., Gobakken, T., and N sset, E. (2018). Geostatistical estimation of forest biomass in interior alaska combining landsat-derived tree cover, sampled airborne lidar and field observations. Remote Sensing of Environment , 212:212--230

    work page 2018

  5. [5]

    O., Weiskittel, A., and Cook, B

    Babcock, C., Matney, J., Finley, A. O., Weiskittel, A., and Cook, B. D. (2013). Multivariate spatial regression models for predicting individual tree structure variables using lidar data. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing , 6(1):6--14

    work page 2013

  6. [6]

    Banerjee, S., Carlin, B., and Gelfand, A. (2014). Hierarchical Modeling and Analysis for Spatial Data, Second Edition . Chapman & Hall/CRC Monographs on Statistics & Applied Probability. Taylor & Francis

    work page 2014

  7. [7]

    T., Cohen, W

    Berterretche, M., Hudak, A. T., Cohen, W. B., Maiersperger, T. K., Gower, S. T., and Dungan, J. (2005). Comparison of regression and geostatistical methods for mapping leaf area index (lai) with landsat etm+ data over a boreal forest. Remote Sensing of Environment , 96(1):49--61

    work page 2005

  8. [8]

    and Astrup, R

    Breidenbach, J. and Astrup, R. (2012). Small area estimation of forest attributes in the norwegian national forest inventory. Eur J Forest Res , 131:1255--1267

    work page 2012

Show all 71 references
  1. [9]

    Buddenbaum, H., Schlerf, M., and Hill, J. (2005). Classification of coniferous tree species and age classes using hyperspectral data and geostatistical methods. International Journal of Remote Sensing , 26(24):5453--5465

    2005

  2. [10]

    W., Green, P

    Coulston, J. W., Green, P. C., Radtke, P. J., Prisley, S. P., Brooks, E. B., Thomas, V. A., Wynne, R. H., and Burkhart, H. E. (2021). Enhancing the precision of broad-scale forestland removals estimates with small area estimation techniques. Forestry: An International Journal ...

    2021

  3. [11]

    Cressie, N. A. C. (1993). Statistics for spatial data . John Wiley & Sons, Inc

    1993

  4. [12]

    Curtis, R. O. and Marshall, D. D. (2000). Technical Note: Why Quadratic Mean Diameter? Western Journal of Applied Forestry , 15(3):137--139

    2000

  5. [13]

    T., Radtke, P

    Dettmann, G. T., Radtke, P. J., Coulston, J. W., Green, P. C., Wilson, B. T., and Moisen, G. G. (2022). Review and synthesis of estimation strategies to meet small area needs in forest inventory. Front. For. Glob. Change , 5:813569

    2022

  6. [14]

    Eliasson, L. (1999). Simulation of thinning with a single-grip harvester. Forest Science , 45(1):26--34

    1999

  7. [15]

    O., Andersen, H.-E., Babcock, C., Cook, B

    Finley, A. O., Andersen, H.-E., Babcock, C., Cook, B. D., Morton, D. C., and Banerjee, S. (2024). Models to support forest inventory and small area estimation using sparsely sampled lidar: A case study involving g-liht lidar in tanana, alaska. Journal of Agricultural, Biologic...

    2024

  8. [16]

    O., Banerjee, S., and Carlin, B

    Finley, A. O., Banerjee, S., and Carlin, B. P. (2007). spbayes: an r package for univariate and multivariate hierarchical point-referenced spatial models. Journal of statistical software , 19(4):1--24

    2007

  9. [17]

    O., Banerjee, S., and Cook, B

    Finley, A. O., Banerjee, S., and Cook, B. D. (2014). Bayesian hierarchical models for spatially misaligned data in r. Methods in Ecology and Evolution , 5(6):514--523

    2014

  10. [18]

    O., Banerjee, S., Cook, B

    Finley, A. O., Banerjee, S., Cook, B. D., and Bradford, J. B. (2013). Hierarchical bayesian spatial models for predicting multiple forest variables using waveform lidar, hyperspectral imagery, and large inventory datasets. International Journal of Applied Earth Observation and...

    2013

  11. [19]

    O., Banerjee, S., Ek, A

    Finley, A. O., Banerjee, S., Ek, A. R., and McRoberts, R. E. (2008). Bayesian multivariate process modeling for prediction of forest attributes. Journal of Agricultural, Biological, and Environmental Statistics , 13(1):60--83

    2008

  12. [20]

    O., Banerjee, S., and Gelfand, A

    Finley, A. O., Banerjee, S., and Gelfand, A. E. (2015). spBayes for large univariate and multivariate point-referenced spatio-temporal data models. Journal of Statistical Software , 63(13):1--28

    2015

  13. [21]

    E., Schmidt, A

    Gelfand, A. E., Schmidt, A. M., Banerjee, S., and Sirmans, C. (2004). Nonstationary multivariate process modeling through spatially varying coregionalization. Test , 13(2):263--312

    2004

  14. [22]

    Gelman, A., Goodrich, B., Gabry, J., and and, A. V. (2019). R-squared for bayesian regression models. The American Statistician , 73(3):307--309

    2019

  15. [23]

    E., Monleon, V

    Goerndt, M. E., Monleon, V. J., and Temesgen, H. (2011). A comparison of small-area estimation techniques to estimate selected stand attributes using lidar-derived auxiliary variables. Canadian Journal of Forest Research , 41(6):1189--1201

    2011

  16. [24]

    E., Monleon, V

    Goerndt, M. E., Monleon, V. J., and Temesgen, H. (2012). Small-area estimation of county-level forest attributes using ground data and remote sensed auxiliary information. Forest Science , 59(5):536--548

    2012

  17. [25]

    Gollob, C., Ritter, T., and Nothdurft, A. (2020). Forest inventory with long range and high-speed personal laser scanning (pls) and simultaneous localization and mapping (slam) technology. Remote Sensing , 12(9)

    2020

  18. [26]

    Gollob, C., Ritter, T., Wassermann, C., and Nothdurft, A. (2019). Influence of scanner position and plot size on the accuracy of tree detection and diameter estimation using terrestrial laser scanning on forest inventory plots. Remote Sensing , 11(13)

    2019

  19. [27]

    C., Burkhart, H

    Green, P. C., Burkhart, H. E., Coulston, J. W., and Radtke, P. J. (2019). A novel application of small area estimation in loblolly pine forest inventory. Forestry: An International Journal of Forest Research , 93(3):444--457

    2019

  20. [28]

    C., Hogg, D

    Green, P. C., Hogg, D. W., Watson, B., and Burkhart, H. E. (2022). Small area estimation in diverse timber types using multiple sources of auxiliary data. Journal of Forestry , 120(6):646--659

    2022

  21. [29]

    M., Schliep, E

    Hanks, E. M., Schliep, E. M., Hooten, M. B., and Hoeting, J. A. (2015). Restricted spatial regression in practice: geostatistical models, confounding, and robustness under model misspecification. Environmetrics , 26(4):243--254

    2015

  22. [30]

    and Heikkinen, J

    Katila, M. and Heikkinen, J. (2020). Reducing error in small-area estimates of multi-source forest inventory by multi-temporal data fusion. Forestry: An International Journal of Forest Research , 93(3):471--480

    2020

  23. [31]

    Kojola, S., Penttil \"a , T., and Laiho, R. (2005). First commercial thinnings in peatland pine stands: Effect of timing on fellings and removals. Baltic Forestry , 11:51--58

    2005

  24. [32]

    and Temesgen, H

    LeMay, V. and Temesgen, H. (2005). Comparison of Nearest Neighbor Methods for Estimating Basal Area and Stems per Hectare Using Aerial Auxiliary Variables . Forest Science , 51(2):109--119

    2005

  25. [33]

    Louis, L., Kizha, A., Daigneault, A., Han, H.-S., and Weiskittel, A. (2022). Factors affecting operational cost and productivity of ground-based timber harvesting machines: a meta-analysis. Curr Forestry Rep , 8:38--54

    2022

  26. [34]

    Magnussen, S. (2013). An assessment of three variance estimators for the k-nearest neighbour technique. Silva Fennica , 47(1):925

    2013

  27. [35]

    Maltamo, M., Malinen, J., Packa \'e n, P., Suvanto, A., and Kangas, J. (2006). Nonparametric estimation of stem volume using airborne laser scanning, aerial photography, and stand-register data. Canadian Journal of Forest Research , 36(2):426--436

    2006

  28. [36]

    Mandallaz, D. (2000). Estimation of the spatial covariance in universal kriging: Application to forest inventory. Environmental and Ecological Statistics , 7(3):263--284

    2000

  29. [37]

    Mandallaz, D. (2014). A three-phase sampling extension of the generalized regression estimator with partially exhaustive information. Canadian Journal of Forest Research , 44(4):383 – 388

    2014

  30. [38]

    Mandallaz, D., Breschan, J., and Hill, A. (2013). New regression estimators in forest inventories with two-phase sampling and partially exhaustive information: A design-based monte carlo approach with applications to small-area estimation. Canadian Journal of Forest Research ,...

    2013

  31. [39]

    and Chiesi, M

    Maselli, F. and Chiesi, M. (2006). Evaluation of statistical methods to estimate forest volume in a mediterranean region. IEEE Transactions on Geoscience and Remote Sensing , 44(8):2239--2250

    2006

  32. [40]

    Matala, J., Kärkkäinen, L., Härkönen, K., Kellomäki, S., and Nuutinen, T. (2009). Carbon sequestration in the growing stock of trees in finland under different cutting and climate scenarios. Eur J Forest Res , 128:493--504

    2009

  33. [41]

    E., Magnussen, S., Tomppo, E

    McRoberts, R. E., Magnussen, S., Tomppo, E. O., and Chirici, G. (2011). Parametric, bootstrap, and jackknife variance estimators for the k-nearest neighbors technique with illustrations using forest inventory and satellite image data. Remote Sensing of Environment , 115(12):3165--3174

    2011

  34. [42]

    E., Næsset, E., Saatchi, S., and Quegan, S

    McRoberts, R. E., Næsset, E., Saatchi, S., and Quegan, S. (2022). Statistically rigorous, model-based inferences from maps. Remote Sensing of Environment , 279:113028

    2022

  35. [43]

    E., Tomppo, E

    McRoberts, R. E., Tomppo, E. O., Finley, A. O., and Heikkinen, J. (2007). Estimating areal means and variances of forest attributes using the k-nearest neighbors technique and satellite imagery. Remote Sensing of Environment , 111(4):466--480

    2007

  36. [44]

    and Stage, A

    Moeur, M. and Stage, A. R. (1995). Most Similar Neighbor: An Improved Sampling Inference Procedure for Natural Resource Planning . Forest Science , 41(2):337--359

    1995

  37. [45]

    Nothdurft, A., Sarkleti, V., Ofner-Graff, T., Tockner, A., Gollob, C., Ritter, T., Kra nitzer, R., Svazek, P., K \"u hmaier, M., Stampfer, K., and Finley, A. O. (2025). Data and code for small area estimation of growing stock timber volume, basal area, mean stem diameter, and ...

    2025

  38. [46]

    D., Larson, B

    Oliver, C. D., Larson, B. C., and Oliver, C. (1996). Forest stand dynamics , volume 520. Wiley New York

    1996

  39. [47]

    S., Tocci, D., Venanzi, R., Stefanoni, W., and Pari, L

    Picchio, R., Latterini, F., Mederski, P. S., Tocci, D., Venanzi, R., Stefanoni, W., and Pari, L. (2020). Applications of GIS -based software to improve the sustainability of a forwarding operation in central italy. Sustainability , 12(14):5716

    2020

  40. [48]

    u tz, J. (1965). Eine neue Methode der Formzahl- und Massenbestimmung stehender St \

    Pollansch \"u tz, J. (1965). Eine neue Methode der Formzahl- und Massenbestimmung stehender St \"a mme - Neue Form- bzw. Kubierungsfunkionen und ihre Anwendung - A new method for the estimation of stem forms . Technical Report 68, Forstliche Bundesversuchsanstalt Mariabrunn

    1965

  41. [49]

    Rao, J. N. and Molina, I. (2015). Small area estimation . John Wiley & Sons

    2015

  42. [50]

    Ritter, T., Gollob, C., and Nothdurft, A. (2020). Towards an optimization of sample plot size and scanner position layout for terrestrial laser scanning in multi-scan mode. Forests , 11(10)

    2020

  43. [51]

    Ritter, T., Schwarz, M., Tockner, A., Leisch, F., and Nothdurft, A. (2017). Automatic mapping of forest stands based on three-dimensional point clouds derived from terrestrial laser-scanning. Forests , 8(8)

    2017

  44. [52]

    Roberts, G. O. and Rosenthal, J. S. (2009). Examples of adaptive mcmc. Journal of Computational and Graphical Statistics , 18(2):349--367

    2009

  45. [53]

    Rose, C. E. and Lynch, T. B. (2001). Estimating parameters for tree basal area growth with a system of equations and seemingly unrelated regressions. Forest Ecology and Management , 148(1–3):51–61

    2001

  46. [54]

    and Gotway, C

    Schabenberger, O. and Gotway, C. (2004). Statistical Methods for Spatial Data Analysis . Chapman & Hall/CRC Texts in Statistical Science. Taylor & Francis

    2004

  47. [55]

    Skovsgaard, J. P. and Vanclay, J. K. (2007). Forest site productivity: a review of the evolution of dendrometric concepts for even-aged stands. Forestry: An International Journal of Forest Research , 81(1):13--31

    2007

  48. [56]

    Steinmann, K., Mandallaz, D., Ginzler, C., and Lanz, A. (2013). Small area estimations of proportion of forest and timber volume combining lidar data and stereo aerial images with terrestrial data. Scandinavian Journal of Forest Research , 28(4):373--385

    2013

  49. [57]

    Sterba, H. (1981). Natürlicher Bestockungsgrad und Reinekes SDI. 1. Bestandesdichte und Ertragsniveau . Centralbl. Gesamte Forstwes. , 98:101--116

    1981

  50. [58]

    Sterba, H. (1985). Das Ertragsniveau und der maximale stand-density-index nach Reineke . Centralbl. Ges. Forstwes. , 102:78--86

    1985

  51. [59]

    Sterba, H. (1987). Estimating potential density from thinning experiments and inventory data . For. Sci. , 33:1022--1034

    1987

  52. [60]

    A., Reckhow, K

    Stow, C. A., Reckhow, K. H., and Qian, S. S. (2006). A bayesian approach to retransformation bias in transformed regression. Ecology , 87(6):1472--1477

    2006

  53. [61]

    Tockner, A. (2025). treeX R package . https://github.com/anditockner/treeX

    2025

  54. [62]

    Tockner, A., Gollob, C., Kraßnitzer, R., Ritter, T., and Nothdurft, A. (2022). Automatic tree crown segmentation using dense forest point clouds from personal laser scanning (pls). International Journal of Applied Earth Observation and Geoinformation , 114:103025

    2022

  55. [63]

    and Halme, M

    Tomppo, E. and Halme, M. (2004). Using coarse scale forest variables as ancillary information and weighting of variables in k-nn estimation: a genetic algorithm approach. Remote Sensing of Environment , 92(1):1--20

    2004

  56. [64]

    Tuominen, S., Fish, S., and Poso, S. (2003). Combining remote sensing, data from earlier inventories, and geostatistical interpolation in multisource forest inventory. Canadian Journal of Forest Research , 33(4):624--634

    2003

  57. [65]

    J., and Guillera-Arroita, G

    Valavi, R., Elith, J., Lahoz-Monfort, J. J., and Guillera-Arroita, G. (2019). blockcv: An r package for generating spatially or environmentally separated folds for k-fold cross-validation of species distribution models. Methods in Ecology and Evolution , 10(2):225--232

    2019

  58. [66]

    Venables, W. N. and Ripley, B. D. (2002). Modern Applied Statistics with S . Springer, New York, fourth edition. ISBN 0-387-95457-0

    2002

  59. [67]

    Venanzi, R., Latterini, F., Civitarese, V., and Picchio, R. (2023). Recent applications of smart technologies for monitoring the sustainability of forest operations. Forests , 14(7):1503

    2023

  60. [68]

    Wackernagel, H. (2003). Multivariate Geostatistics: An Introduction with Applications . Springer Berlin Heidelberg

    2003

  61. [69]

    E., and Wulder, M

    Zhang, C., Franklin, S. E., and Wulder, M. A. (2004). Geostatistical and texture analysis of airborne-acquired images used in forest classification. International Journal of Remote Sensing , 25(4):859--865

    2004

  62. [70]

    O., Nothdurft, A., and Banerjee, S

    Zhang, L., Finley, A. O., Nothdurft, A., and Banerjee, S. (2024). Bayesian modeling of incompatible spatial data: A case study involving post-adrian storm forest damage assessment

    2024

  63. [71]

    Zimmerman, D. L. and Ver Hoef, J. M. (2022). On deconfounding spatial confounding in linear models. The American Statistician , 76(2):159--167

    2022

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.