Estimation of Wood Fibre Length Distributions from Censored Data through an EM Algorithm
Abstract
Abstract. An expectation maximization (EM) algorithm is proposed to find fibre length distributions in standing trees. The available data come from cylindric wood samples (increment cores). The sample contains uncut fibres as well as fibres cut once or twice. The sample contains not only fibres, but also other cells, the so‐called ‘fines’. The lengths are measured by an automatic fibre‐analyser, which is not able to distinguish fines from fibres and cannot tell if a cell has been cut. The data thus come from a censored version of a mixture of the fine and fibre length distributions in the tree. The parameters of the length distributions are estimated by a stochastic version of the EM algorithm, and an estimate of the corresponding covariance matrix is derived. The method is applied to data from northern Sweden. A simulation study is also presented. The method works well for sample sizes commonly obtained from increment cores.
Citing Literature
Number of times cited according to CrossRef: 8
- Zhi-Qiang Chen, Bo Karlsson, Tommy Mörling, Lars Olsson, Ewa J. Mellerowicz, Harry X. Wu, Sven-Olof Lundqvist, María Rosario García Gil, Genetic analysis of fiber dimensions and their correlation with stem diameter and solid-wood properties in Norway spruce, Tree Genetics & Genomes, 10.1007/s11295-016-1065-0, 12, 6, (2016).
- Zhi-Qiang Chen, Konrad Abramowicz, Rafal Raczkowski, Stefana Ganea, Harry X. Wu, Sven-Olof Lundqvist, Tommy Mörling, Sara Sjöstedt de Luna, María Rosario García Gil, Ewa J. Mellerowicz, Method for accurate fiber length determination from increment cores for large-scale population analyses in Norway spruce, Holzforschung, 10.1515/hf-2015-0138, 70, 9, (829-838), (2016).
- Malte Kuhlmann, Claudia Redenbach, Estimation of Fibre Length Distributions from Fibre Endpoints, Scandinavian Journal of Statistics, 10.1111/sjos.12148, 42, 4, (1010-1022), (2015).
- Arttu Miettinen, Cris L. Luengo Hendriks, Gary Chinga-Carrasco, E. Kristofer Gamstedt, Markku Kataja, A non-destructive X-ray microtomography approach for measuring fibre length in short-fibre composites, Composites Science and Technology, 10.1016/j.compscitech.2012.08.008, 72, 15, (1901-1908), (2012).
- L. Sedda, P. M. Atkinson, E. Barca, G. Passarella, Imputing censored data with desirable spatial covariance function properties using simulated annealing, Journal of Geographical Systems, 10.1007/s10109-010-0145-1, 14, 3, (265-282), (2010).
- Ingrid Svensson, Sara Sjöstedt-de Luna, Asymptotic properties of a stochastic EM algorithm for mixtures with censored data, Journal of Statistical Planning and Inference, 10.1016/j.jspi.2009.06.014, 140, 1, (111-127), (2010).
- Ingrid Svensson, Sara Sjöstedt-de Luna, Tommy Mörling, Anders Fries, Tore Ericsson, Adjusting for fibre length-biased sampling probability using increment cores from standing trees, Holzforschung, 10.1515/HF.2007.016, 61, 1, (2007).
- Quang V. Cao, Qinglin Wu, Characterizing wood fiber and particle length with a mixture distribution and a segmented distribution, Holzforschung, 10.1515/HF.2007.023, 61, 2, (2007).
