Assessing the impacts of land-use change on the hydrology of the tropical Andes

Abstract

Land-use and land-cover change (LUCC) has been identified as a major driver of change to the hydrological cycle. However, it is still a scientific challenge to quantify these effects. Land surface models are increasingly being used for such hydrological assessment because of their state-of-the-art representation of physical processes and versatility. A physically-based model has the advantage to map the modeller’s knowledge about the hydrological impacts of land-use and land-cover change into physically meaningful parameters. This PhD thesis explores the use of a land surface model (Joint UK Land-Environment Simulator, JULES) in combination with high temporal resolution in-situ data on streamflow, precipitation, and several weather variables, collected by a grassroots hydrological monitoring initiative (called iMHEA) in the tropical Andes. I find that the in-situ data can improve the hydrological simulation substantially, mainly by reducing uncertainty inherent in using large-scale precipitation data. The commonly used soil parameters based on pedotransfer functions lead to an underestimation of the flow. Therefore, I modified the soil parameterisation with experimental data for a more accurate representation of subsurface flow generation. Subsequently, I assessed the potential impacts of watershed interventions (grazing, afforestation, cultivation) using the calibrated soil parameters. A reduction in water yield and water regulation ability under these land use scenarios was identified, which is in line with observed impacts and relevant for water resources managers. In a next step, I implemented an open source land use change model, the lulcc R package, to analyse the regional land cover changes in the Andean region, and to generate predictive land use maps that can be used to drive the JULES model. For this purpose, the JULES model has been implemented at a regional scale using multiple sources of global data. The use of the JULES model allows the effects of LUCC to be assessed using knowledge about physical processes. My results show a further 3.7% of deforestation occurring in the region, which changes the flow by ±17% consequently.