Fitting diameter distribution models to data from forest inventories with concentric plot design

Aim: Several national forest inventories use a complex plot design based on multiple concentric subplots where smaller diameter trees are inventoried when lying in the smaller-radius subplots and ignored otherwise. Data from these plots are truncated with threshold (truncation) diameters varying according to the distance from the plot centre. In this paper we designed a maximum likelihood method to fit the Weibull diameter distribution to data from concentric plots.Material and methods: Our method (M1) was based on multiple truncated probability density functions to build the likelihood. In addition, we used an alternative method (M2) presented recently. We used methods M1 and M2 as well as two other reference methods to estimate the Weibull parameters in 40000 simulated plots. The spatial tree pattern of the simulated plots was generated using four models of spatial point patterns. Two error indices were used to assess the relative performance of M1 and M2 in estimating relevant stand-level variables. In addition, we estimated the Quadratic Mean plot Diameter (QMD) using Expansion Factors (EFs).Main results: Methods M1 and M2 produced comparable estimation errors in random and cluster tree spatial patterns. Method M2 produced biased parameter estimates in plots with inhomogeneous Poisson patterns. Estimation of QMD using EFs produced biased results in plots within inhomogeneous intensity Poisson patterns.Research highlights:We designed a new method to fit the Weibull distribution to forest inventory data from concentric plots that achieves high accuracy and precision in parameter estimates regardless of the within-plot spatial tree pattern.

Spatial distribution of trees on light taiga plots before selective thinning

Since 2009 the School of Agroecology and Business, Institute of Plant and Agricultural Sciences of the Mongolian University of Life Sciences in Darkhan has established research plots in two research areas in the Selenge aimag. The establishment was conducted in close cooperation with development organisations (FAO, GIZ) and the University of Goettingen. The purpose of the research initiative is to combine capacity development and monitoring of forest structure in the mountain forest steppe zone and taiga zone. Here we report results on the horizontal spatial structure of forest stands. We analysed the spatial distribution of trees on birch and larch plots of the research area «Altansumber» before a selective thinning took place on some plots in 2009. The research area is situated in the mountain forest steppe zone. The forests belong to the light taiga. The selected stands approach a chronosequence. The results showed that the tree distributions were mainly irregular («clumped»).Random spatial tree distribution occurred especially in the medium-aged birch stand. We found no indication of regular tree distributions in any of the plots. We assume that the disturbance regime and successional processes are the driving factors leading to the specific tree distribution pattern on the plots. Due to different regeneration strategies and life span of the dominating species the birch stands and the larch stands seem to differ slightly concerning the chronological occurrence of clumped and random spatial tree distribution. We finally conclude that a better control of the disturbance regime would not only support an undisturbed forest succession to riper forest stands but also result in less forest stands with irregular spatial distribution. This may also have implications on forest productivity.Mongolian Journal of Agricultural Sciences Vol.15(2) 2015; 91-99

Reconstructing spatial tree point patterns from nearest neighbour summary statistics measured in small subwindows

Spatial tree data are required for the development of spatially explicit models and for the estimation of summary statistics such as Ripley’s K function. Such data are rare and expensive to gather. This paper presents an efficient method of synthesizing spatial tree point patterns from nearest neighbour summary statistics (NNSS) sampled in small circular subwindows, which uses a stochastic optimization technique based on simulated annealing and conditional simulation. This nonparametric method was tested by comparing tree point patterns, reconstructed from sample data, with the original woodland patterns of three structurally different tree populations. Analysis and validation show that complex spatial woodland structures, including long-range tree interactions, can be successfully reconstructed from NNSS despite the limited range of the subwindows and statistics. The influence of the NNSS varies depending on the woodland under study. In some cases, the sampling results can be improved by reconstruction. Furthermore, it is clearly shown that it is possible to estimate second-order characteristics such as Ripley’s K function from small circular subwindows through the reconstruction technique. The results offer new opportunities for adding value to woodland surveys by making raw data available for further work such as growth projections, visualization, and modelling.