Improving the representation of high-latitude vegetation distribution in dynamic global vegetation models

Abstract. Vegetation is an important component in global ecosystems, affecting the physical, hydrological and biogeochemical properties of land surface. Accordingly, way vegetation parameterized strongly influences predictions future climate by Earth system models. To capture spatial temporal changes cover its feedbacks to system, dynamic models (DGVMs) are included as components surface Variation predicted from DGVMs therefore has large impacts on modelled radiative non-radiative properties, especially over high-latitude regions. mostly evaluated remotely sensed products less often other or situ field observations. In this study, we evaluate performance three methods for representation present-day with respect prediction plant functional type (PFT) profiles – one based upon distribution (DMs), that uses a remote sensing (RS) dataset DGVM (CLM4.5BGCDV; Community Land Model 4.5 Bio-Geo-Chemical cycles Dynamical Vegetation). While predict PFT physiological ecological processes, DM relies statistical correlations between set predictors target, RS classification spectral reflectance patterns satellite images. obtained independently collected field-based Norway were used evaluation. We found RS-based matched reference best, closely followed DM, whereas deviated reference. overestimated area covered boreal needleleaf evergreen trees bare ground at expense broadleaf deciduous shrubs. Based environmental identified important, new variables (e.g. minimum temperature May, snow water equivalent October precipitation seasonality) selected threshold establishment these PFTs. performed series sensitivity experiments investigate if thresholds improve method. our results, suggest implementation novel PFT-specific (i.e. The results highlight potential using development broader Also, emphasize establishing DMs reliable method providing distributions evaluation alongside RS.