Work Package 4 (HydroSim): Water resources simulation

This WP consists on two main activities: data collection and processing, and hydrological modelling. In the first WP4 activity, remotely-sensed, gauge-based, in-situ and model-based hydro-meteorological and other ancillary data will be collected for all the pilot catchments during the last 30 years. In the second activity, these data will be used to implement both Liuxihe (Chen et al., 2011) and SWAT+ (Bieger et al., 2017) models to simulate hydrological processes at the catchment scale.

After data collection and processing, WP4 will develop a comprehensive catchment hydrological modelling framework in order to reliably simulate the main hydrological processes in the selected catchments, including precipitation, snowmelt, evapotranspiration, infiltration and surface runoff at hourly and daily time steps. A novel combination of two hydrological models (Liuxihe and SWAT+) will be used to estimate uncertainties in streamflow projections coming from model structure.

  • The Liuxihe model (Chen et al., 2011) is a physically-based and fully-distributed hydrological model developed by the Chinese team mainly for flood forecasting. To adequately represent the catchment conditions in the Chilean territory, two new components will be developed for the Liuxihe model. Firstly, a snowmelt component for simulating the snow accumulation and melting process in catchments with nival or pluvio-nival hydrological regime (Chilean Andean catchments, and a Chinese catchment). Secondly, a more detailed ET component will be developed to better represent actual ET and water infiltration into the soil (required for agricultural and forestry catchments). The Liuxihe upgraded model structure will be evaluated by comparing its simulated streamflow, actual ET and snow cover (where applicable) with those simulated with SWAT+, and with observed data.

  • The SWAT+ hydrological model (Bieger et al., 2017) is a newly restructured version of the well-known Soil and Water Assessment Tool (SWAT, Arnold et al., 1998; 2012), which has been developed to cope with ongoing and future challenges in water resources. Among the most important changes, those related to greater flexibility in defining aquifers, reservoirs, and channels would be particularly important for this project. In SWAT+ the user can define any number of aquifers, and a new option is available to use MODFLOW for simulating groundwater processes. Reservoirs no longer have to be placed on the main channel, and they can interact with any other spatial object (Bieger et al., 2017). The Chilean team has strong experience in using the SWAT model, and this project will give the opportunity to implement this new version of the model to both Chilean and Chinese catchments.

A sensitivity analysis of the two hydrological models will be carried out to identify the most sensitive model parameters, which will then be calibrated using a multi-objective PSO algorithm specially developed within this project. Then, changes in water resources in the selected pilot catchments, in particular changes in the frequency of floods and droughts, will be simulated both in the past 30 years and for the 21st century. Methodologies for executing these tasks are described below.

This newly proposed comprehensive hydrological simulation framework will be agreed upon to carry out the sensitivity analysis, calibration and verification of both Liuxihe and SWAT+ models in all the selected pilot catchments. The upgraded Liuxihe model will be implemented in the Tianhe No. 2 supercomputer from the Chinese team. This system will be used to carry out the simulation of hydrological processes for all the catchments in Chile and China. On the other hand, the SWAT+ hydrological model will be implemented in the cluster (project PII180008) currently available at the Universidad de La Frontera, and it will be used to carry out the simulation of hydrological processes in all the Chilean and Chinese pilot catchments.