Performance of dynamic vegetation in land-surface models - a multi-site comparison

Land-surface models aim to represent exchange processes between soil and atmosphere via the surface by coupling hydrological and carbon fluxes. Vegetation directly links between hydrological and carbon cycle and, thus essentially, is included in land-surface models. But its dynamics are challenging to capture in models, which is not least because of difficulties in data acquisition. Nevertheless, some land-surface models are available that come with modules for dynamic vegetation. Here, we conducted a model-data comparison to evaluate the representation of dynamic vegetation and related surface fluxes of the two models ECLand and Noah-MP by using the FLUXNET 2015 dataset, data from the TERENO site “Hohes Holz” and a MODIS leaf area product. With the current implementation, using dynamic vegetation modules did not enhance representativeness of the vegetation in ECLand. Dynamic vegetation in Noah-MP improved vegetation representations at least for some sites. For the exchange fluxes, using prescribed leaf-area climatology from remote sensing products appeared with the best model performance. The representation of hydrological fluxes and soil moisture remained almost untouched for both models. Additionally, the performance of the models in vegetation- and hydrology-related variables did not depend on each other. The current implemented modules for dynamic vegetation in these two models yielded no better model performance compared to runs with prescribed leaf-area climatology. Hence, they provide an example that additional model complexity does not lead to improved model performance.