Ecohydrological Modelling and Performance Evaluation of a Large-scale Land Surface over the Central Andes

Alpine mountain ecosystems in the tropical Andes are critical water sources for both human societies and natural systems. These regions store and gradually release water from glaciers, regulate runoff patterns, support irrigated agriculture, facilitate hydropower generation, and sustain delicate ecosystems. The catchments in the tropical Andes are characterized by complex mountainous topography, diverse climates, and dynamic land-use changes. The rapid expansion of agriculture and urbanization has driven significant deforestation, followed by forest recovery efforts, which have substantially altered evapotranspiration patterns and runoff dynamics. In addition, climate change has accelerated glacier retreat and shifted precipitation patterns, causing profound impacts on hydrological processes and ecosystem dynamics. Despite these significant changes in blue, green, and white water fluxes in the region, severe data limitations impede the understanding of regional ecohydrological cycles under the changing environment as well as the development of high-resolution eco-hydrological models. 

Our study addresses these gaps by employing innovative computational modelling alongside in-situ and remote sensing observations. We conducted hyper-resolution simulations of coupled water, energy, and carbon dynamics in the tropical Andes using the physics-based T&C model, which we parameterized with data from multiple sources. This approach allowed us to analyse the fate of blue, green, and white water fluxes under climate change scenarios. To streamline regional studies in terms of scalability and applicability, we developed automated input data preparation and model parameter generalization algorithms integrating machine learning and remote sensing methods for the physics-based model. This not only allows flexibility when composing catchment plant functional types (PFTs) but significantly speeds up the model setup process. Validation of the algorithm is through plot-scale simulations using PLUMBER2 sites and spatial scale simulations using CARAVAN catchments before expanding the simulations to larger extents in the tropical Andes.  

As a proof of concept, we applied our methodology to the Vilcanota catchment in Peru, a catchment around 9000 square kilometres. This catchment presents complex land uses ranging from glacial coverage above treeline, diverse natural vegetations and intricate crop rotation systems and an elevational span of 4,000 meters.