Solute Transport Processes and Water Quality Modelling of a Large Tropical River: Comparison of Aggregated and Distributed Approaches

Data based mechanistic, DBM, environmental models able to accurately reproduce input-output behavior of large natural systems are still required in practical water quantity and quality real-time control applications, due to their low computational cost and high efficiency. DBM models are also efficient and sufficient accurate tools to be used as the basis of complex decision support systems for the integrated management of water resources at the catchment scale. In this work an efficient aggregated hydrologic (MDLC), solute transport (ADZ) and water quality (QUASAR) model of a large 800 kilometer stretch of the Magdalena tropical River was implemented, calibrated and compared with a distributed model (Hec-Ras vs. 5.0).

Both models are capable to represent physical and chemical processes with high similitude along the large modelled stretch of the Magdalena River, since both model parameters have physical correspondence and are related by a moment matching technique. The aggregated model has however a simplified structure and exhibits higher numerical efficiency whilst maintaining the precision of the distributed model.

This work shows that it is possible to rigorously calibrate the aggregated model using objective optimization methods based on Monte Carlo simulations or genetic algorithms, and use the calibrated parameters to obtain a high predictive distributed model (see Fig. 1). In the cases analyzed savings of computer time of about 1000% are obtained in the long river stretch. This difference is important for instance if flow prediction is incorporated within adaptive flood warning schemes or probabilistic decision support frameworks that undertake Monte Carlo based uncertainty estimation.