A comprehensive evaluation of hydrological processes in a second-generation dynamic vegetation model

The global water and carbon cycles are greatly influenced by terrestrial vegetation, making trustworthy representations of dynamic biosphere-hydrosphere interactions a crucial component of both ecosystem and climate models. This paper comprehensively evaluates the hydrological performance of a leading dynamic global vegetation model Lund-Potsdam-Jena General Ecosystem Simulator (LPJ-GUESS), using a broad range of the latest available global observation-based gridded datasets that cover the main components of the hydrological cycle. Overall, we find that the hydrological components modelled by LPJ-GUESS agree well with global gridded datasets of runoff, evapotranspiration and surface soil moisture, though there are discrepancies in some regions and periods. Furthermore, LPJ-GUESS accurately captures both inter- and intra-annual variations of runoff in most regions and catchment areas, including the Danube, Murray, Yangtze, Yenisei and Nile basins. Total evapotranspiration modelled by LPJ-GUESS agrees closely with the evapotranspiration estimates of the Global Land Evaporation Amsterdam Model and PML-V2 datasets, but with some disagreement in the individual components, especially for evaporation. The surface soil moisture simulated by LPJ-GUESS aligns with ESA-CCI (v5.3) surface soil moisture datasets in most regions, with greatest discrepancies in subarctic areas. We attribute these discrepancies to two main sources: (1) absent or poor representation of processes such as river routing, storage and supply of water bodies, and cropland irrigation; and (2) uncertainties in both reference datasets and input to the model, including precipitation, soil texture, and land use. The latest ecosystem model Lund-Potsdam-Jena General Ecosystem Simulator (LPJ-GUESS) can reproduce observed patterns of runoff, evapotranspiration, and soil surface moisture across a range of temporal and spatial scales. Evaluation of hydrological processes of LPJ-GUESS is critical for identifying important processes that should be developed to further and provides valuable insights for the improvement of this and other ecosystem models. image