ecological modelling 216 ( 2 0 0 8 ) 102–106 available at www.sciencedirect.com journal homepage: www.elsevier.com/locate/ecolmodel Forest growth in the light of the thermodynamic theory of ecological systems Georgii A. Alexandrov * Office for Global Environmental Database, Centre for Global Environmental Research, National Institute for Environmental Studies, Onogawa 16–2, Tsukuba, Ibaraki 305-8506, Japan article info Article history: Published on line 15 April 2008 Keywords: Forest ecology Exergy Jørgensen–Svirezhev theory Dimensional analysis Similarity principle abstract The observed growth of a particular forest stand can be described by many models and explained by some of them. The forest growth models are also successfully applied for extrapolating the growth curve. However, the known models of forest growth are not “one- point” models. They are not designed to predict the future growth of a forest stand from its current state: the model parameters either are not directly measurable or cannot be mea- sured with relevant accuracy. This article is an attempt to use Jørgensen–Svirezhev theory as a new clue to the choice of variables that determines forest growth. The postulates of this theory combined with the pipe theory of tree growth lead to conclusion that biomass of a stand should be proportional to the four-fifths power of its age. Empirical validation, however, disclosed that calendar age is rather approximate measure of ecosystem ontogeny. Delayed development or intensive thinning of a forest stand at the early stages leads to reju- venation bias. Thus derived 4/5-law model approximates well-known Chapman–Richards model in the neighborhood of the inflection point, and is applicable to middle-aged forest stands. © 2008 Elsevier B.V. All rights reserved. “It can be said that the world around us is the complex of stable harmonic forms and short periods of transitional processes between them. But one may think otherwise: the world is continuous development, eternal instability, and the periods of stabilization are short stops on this way.”—Yuri Svirezhev. 1. Introduction The forest growth is normally modeled either with a logis- tic model (Fukuda et al., 2003; Kohlmaier et al., 1995; Mende and Albrecht, 2001; Neeff and dos Santos, 2005; Parresol, 1995; Zeide, 2004) or with mass-balance equations that form a process-based model (Bossel, 1996; Huth and Ditzer, 2000; Komarov et al., 2003a; Webb, 1991). Both kinds of models may describe well the observed growth and successfully extrap- ∗ Tel.: +81 298 50 2974/2960; fax: +81 298 50 2974/2960. E-mail address: g.alexandrov@nies.go.jp. olate the growth curve. However, they are not designed to predict the future growth of a forest stand from its current state. The parameters of logistic models are not directly mea- surable: they are estimated from the observed part of the growth curve. The parameters of process-based models are measurable, but the major purpose of measurements is to establish some empirical bounds for model calibration. More- over, some parameters are measurable only in situ—that is, the usage of a process-based model in macro-scale assessments raises the question of whether its settings are valid. This article is an attempt to design a model for predicting the future growth of a forest from its current state as observed remotely. It starts from belief that current state of a forest con- tains information, which is essential for predicting its future state, and that the problem is to find the ‘governing variables’ summarizing this information in remotely measurable form. 0304-3800/$ – see front matter © 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.ecolmodel.2007.11.022