Improving the dynamics of Northern Hemisphere high-latitude vegetation in the ORCHIDEE ecosystem model

摘要

Processes that describe the distribution of vegetation and ecosystemsuccession after disturbance are an important component of dynamic globalvegetation models (DGVMs). The vegetation dynamics module (ORC-VD) withinthe process-based ecosystem model ORCHIDEE (Organizing Carbon and Hydrologyin Dynamic Ecosystems) has not been updated and evaluated since many yearsand is known to produce unrealistic results. This study presents a newparameterization of ORC-VD for mid- to high-latitude regions in the NorthernHemisphere, including processes that influence the existence, mortality andcompetition between tree functional types. A new set of metrics is alsoproposed to quantify the performance of ORC-VD, using up to five differentdata sets of satellite land cover, forest biomass from remote sensing andinventories, a data-driven estimate of gross primary productivity (GPP) andtwo gridded data sets of soil organic carbon content. The scoring of ORC-VDderived from these metrics integrates uncertainties in the observationaldata sets. This multi-data set evaluation framework is a generic method thatcould be applied to the evaluation of other DGVM models. The results of theoriginal ORC-VD published in 2005 for mid- to high-latitudes and of the newparameterization are evaluated against the above-described data sets.Significant improvements were found in the modeling of the distribution oftree functional types north of 40° N. Three additional sensitivityruns were carried out to separate the impact of different processes ordrivers on simulated vegetation distribution, including soil freezing whichlimits net primary production through soil moisture availability in the rootzone, elevated CO₂ concentration since 1850, and the effects offrequency and severity of extreme cold events during the spin-up phase ofthe model.