Global variability of simulated and observed vegetation growing season

Vegetation phenology and its variability have substantial influence on land-atmosphereinteraction, and changes in growing season length are additional indicators of climatechange impacts on ecosystems. For these reasons, global land surface models are routinelyevaluated in order to assess their ability to reproduce the observed phenological variabil-ity. In this work, we present a new approach that integrates a wider spectrum of growingseason modes, in order to better describe the observed variability in vegetation growingseason onset and offset, as well as assess the ability of state-of-the-art land surface mod-els to capture this variability at the global scale. The method is applied to the CommunityLand Model version 4.5 (CLM4.5) simulations and LAI3g satellite-observation. The com-parison between data and model outputs shows that CLM4.5 is capable of reproducingthe growing season features in the Northern Hemisphere mid- and high-latitudes, but alsodisplays its limitations in areas where water availability acts as the main driver of veg-etation phenological activity. Besides, the new approach allows evaluating land surfacemodels in capturing multi-growing-season phenology. In this regard, CLM4.5 proves itsability in reproducing the two-growing-season cycles in the Horn of Africa. In general,the new methodology expands the area of analysis from northern mid- and high-latitudesto the global continental areas, and allows to assess the vegetation response to the ongoingclimate change in a larger variety of ecosystems, ranging from semi-arid regions to rainforests, passing through temperate deciduous and boreal evergreen forests.