Max-Planck-Institut für Biogeochemie, Hans-Knöll-Strasse 10, 07745 Jena, Germany.
Philos Trans R Soc Lond B Biol Sci. 2010 May 12;365(1545):1449-55. doi: 10.1098/rstb.2009.0309.
Feedbacks between water use, biomass and infiltration capacity in semiarid ecosystems have been shown to lead to the spontaneous formation of vegetation patterns in a simple model. The formation of patterns permits the maintenance of larger overall biomass at low rainfall rates compared with homogeneous vegetation. This results in a bias of models run at larger scales neglecting subgrid-scale variability. In the present study, we investigate the question whether subgrid-scale heterogeneity can be parameterized as the outcome of optimal partitioning between bare soil and vegetated area. We find that a two-box model reproduces the time-averaged biomass of the patterns emerging in a 100 x 100 grid model if the vegetated fraction is optimized for maximum entropy production (MEP). This suggests that the proposed optimality-based representation of subgrid-scale heterogeneity may be generally applicable to different systems and at different scales. The implications for our understanding of self-organized behaviour and its modelling are discussed.
水利用、生物量和半干旱生态系统入渗能力之间的反馈作用已被证明会导致在简单模型中自发形成植被模式。与均匀植被相比，模式的形成允许在较低的降雨量下维持更大的整体生物量。这导致在较大尺度上运行的模型忽略了子网格尺度的可变性。在本研究中，我们探讨了子网格尺度的非均质性是否可以作为裸土和植被区之间最优划分的结果进行参数化的问题。我们发现，如果将植被部分优化为最大熵产生（MEP），则两箱模型可以再现出在 100x100 网格模型中出现的模式的时间平均生物量。这表明，所提出的基于最优性的子网格尺度非均质性表示可能普遍适用于不同的系统和不同的尺度。讨论了这对子网格尺度非均质性的自组织行为及其建模的理解的影响。
