Synergistic Calibration of a Hydrological Model Using Discharge and Remotely Sensed Soil Moisture in the Paraná River Basin

Abstract

Hydrological models are useful tools for water resources studies, yet their calibration is still a challenge, especially if aiming at improved estimates of multiple components of the water cycle. This has led the hydrologic community to look for ways to constrain models with multiple variables. Remote sensing estimates of soil moisture are very promising in this sense, especially in large areas for which field observations may be unevenly distributed. However, the use of such data to calibrate hydrological models in a synergistic way is still not well understood, especially in tropical humid areas such as those found in South America. Here, we perform multiple scenarios of multiobjective model optimization with in situ discharge and the SMOS L4 root zone soil moisture product for the Upper Paraná River Basin in South America (drainage area > 900,000 km²), for which discharge data for 136 river gauges are used. An additional scenario is used to compare the relative impacts of using all river gauges and a small subset containing nine gauges only. Across the basin, the joint calibration (CAL-DS) using discharge and soil moisture leads to improved precision and accuracy for both variables. The discharges estimated by CAL-DS (median KGE improvement for discharge was 0.14) are as accurate as those obtained with the calibration with discharge only (median equal to 0.14), while the CAL-DS soil moisture retrieval is practically as accurate (median KGE improvement for soil moisture was 0.11) as that estimated using the calibration with soil moisture only (median equal to 0.13). Nonetheless, the individual calibration with discharge rates is not able to retrieve satisfactory soil moisture estimates, and vice versa. These results show the complementarity between these two variables in the model calibration and highlight the benefits of considering multiple variables in the calibration framework. It is also shown that, by considering only nine gauges instead of 136 in the model optimization, the model is able to estimate reasonable discharge and soil moisture, although relatively less accurately and with less precision than for the entire dataset. In summary, this study shows that, for poorly gauged tropical basins, the joint calibration of SMOS soil moisture and a few in situ discharge gauges is capable of providing reasonable discharge and soil moisture estimates basin-wide and is more preferable than performing only a discharge-oriented optimization process.

Assimilation of SMAP Products for Improving Streamflow Simulations over Tropical Climate Region—Is Spatial Information More Important Than Temporal Information?

“…In the tropical climate, streamflow is often of great variation, due to the impacts of large-scale phenomena such as ENSO on the seasonal and year-to-year variation in soil moisture, which results from the high variability in rainfall [39]. Any technique such as DA that could enhance hydrological model performances in the tropical climate region is essential, but such studies have rarely been investigated [40], owing to the difficulty of accessing streamflow records over these regions [41].…”

Simulating Discharge in a Non-Dammed River of Southeastern South America Using SWAT Model

“…Since then, many studies have addressed the variability of the streamflow over the basin [2,3,[10][11][12][13]. Additionally, several hydrological models of the UPRB were presented along with results discussed based on the main sub-basins [14][15][16]. Despite progress in the application of modeling in large-scale basins, its use has not yet been incorporated into the articulation that integrates the main user sectors and resource management bodies.…”