Current issue: 53(2)

Under compilation: 53(3)

Impact factor 1.683
5-year impact factor 1.950
Silva Fennica 1926-1997
1990-1997
1980-1989
1970-1979
1960-1969
Acta Forestalia Fennica
1953-1968
1933-1952
1913-1932

Articles containing the keyword 'linear prediction'.

Category: Research article

article id 1057, category Research article
Jouni Siipilehto, Lauri Mehtätalo. (2013). Parameter recovery vs. parameter prediction for the Weibull distribution validated for Scots pine stands in Finland. Silva Fennica vol. 47 no. 4 article id 1057. https://doi.org/10.14214/sf.1057
Highlights: A parameter recovery method (PRM) was developed for forest stand inventories and compared with previously developed parameter prediction methods (PPM) in Finland; PRM for the 2-parameter Weibull function provided compatibility for the main stand characteristics: stem number, basal area and one of the four optional mean characteristics; PRM provided comparable and at its best, superior accuracy in volume characteristics compared with PPM.
The moment-based parameter recovery method (PRM) has not been applied in Finland since the 1930s, even after a continuation of forest stand structure modelling in the 1980s. This paper presents a general overview of PRM and some useful applications. Applied PRM provided compatibility for the included stand characteristics of stem number (N) and basal area (G) with either mean (D), basal area-weighted mean (DG), median (DM) or basal area-median (DGM) diameter at breast height (dbh). A two-parameter Weibull function was used to describe the dbh-frequency distribution of Scots pine stands in Finland. In the validation, PRM was compared with existing parameter prediction models (PPMs). In addition, existing models for stand characteristics were used for the prediction of unknown characteristics. Validation consisted of examining the performance of the predicted distributions with respect to variation in stand density and accuracy of the localised distributions, as well as accuracy in terms of bias and the RMSE in stand characteristics in the independent test data set. The validation data consisted of 467 randomly selected stands from the National Forest Inventory based plots. PRM demonstrated excellent accuracy if G and N were both known. At its best, PRM provided accuracy that was superior to any existing model in Finland – especially in young stands (mean height < 9 m), where the RMSE in total and pulp wood volumes, 3.6 and 5.7%, respectively, was reduced by one-half of the values obtained using the best performing existing PPM (8.7–11.3%). The unweighted Weibull distribution solved by PRM was found to be competitive with weighted existing PPMs for advanced stands. Therefore, using PRM, the need for a basal area weighted distribution proved unnecessary, contrary to common belief. Models for G and N were shown to be unreliable and need to be improved to obtain more reliable distributions using PRM.
  • Siipilehto, Finnish Forest Research Institute, Vantaa Research Unit, P.O. Box 18, FI-01301 Vantaa, Finland ORCID ID:E-mail: jouni.siipilehto@metla.fi (email)
  • Mehtätalo, University of Eastern Finland, School of Computing, P.O. Box 111, FI-80101 Joensuu, Finland ORCID ID:E-mail: lauri.mehtatalo@uef.fi
article id 99, category Research article
Jouni Siipilehto. (2011). Local prediction of stand structure using linear prediction theory in Scots pine-dominated stands in Finland. Silva Fennica vol. 45 no. 4 article id 99. https://doi.org/10.14214/sf.99
This study produced a family of models for eight standard stand characteristics, frequency and basal area-based diameter distributions, and a height curve for stands in Finland dominated by Scots pine (Pinus sylvestris L.). The data consisted of 752 National Forest Inventory-based sample plots, measured three times between 1976 and 2001. Of the data, 75% were randomly selected for modelling and 25% left out for model evaluation. Base prediction models were constructed as functions of stand age, location and site providing strongly average expectations. These expectations were then calibrated with the known stand variables using linear prediction theory when estimating the best linear unbiased predictor (BLUP). Three stand variables, typically assessed in Finnish forest management planning fieldwork, were quite effective for calibrating the expectation for the unknown variable. In the case of optional distributions, it was essential to choose the weighting of the diameter distribution model such that the available input variables and the model applied were based on the same scale (e.g. arithmetic stand variables for frequency distribution). Additional input variables generally improved the accuracy of the validated characteristics, but the improvements in the predicted distributions were most noteworthy when the arithmetic mean and basal area-weighted median were simultaneously included in the BLUP estimation. The BLUP method provided a flexible approach for characterising relationships among stand variables, alternative size distributions and the height–diameter curve. Models are intended for practical use in the MOTTI simulator.
  • Siipilehto, Finnish Forest Research Institute, Vantaa Research Unit, P.O. Box 18, FI-01301 Vantaa, Finland ORCID ID:E-mail: jouni.siipilehto@metla.fi (email)

Register
Click this link to register for Silva Fennica submission and tracking system.
Log in
If you are a registered user, log in to save your selected articles for later access.
Contents alert
Sign up to receive alerts of new content
Your selected articles

Committee on Publication Ethics A Trusted Community-Governed Archive