Rainfall–runoff modeling of recent hydroclimatic change in a subtropical lake catchment: Laguna Mar Chiquita, Argentina

TROIN, M.; VALLET-COULOMB C; PIOVANO, E; SYLVESTRE, F

JOURNAL OF HYDROLOGY

ELSEVIER SCIENCE BV

Lugar: Amsterdam; Año: 2012 p. 379 - 391

0022-1694

The 1970s abrupt lake level rise of Laguna Mar Chiquita in central Argentina was shown to be driven by an increase in the Rio Sali-Dulce discharge outflowing from the northern part of the lake catchment. This regional hydrological change was consistent with the 20th century hydroclimatic trends observed in southeastern South America. However, little is known about the impacts of climate or land cover changes on this regional hydrological change causing the sharp lake level rise. To address this question, the present study aims to provide an integrated basin-lake model. We used the physically-based SWAT model in order to simulate streamflow in the Sali-Dulce Basin. The ability of SWAT to simulate non-stationary hydrological conditions was evaluated by a cross-calibration exercise. Based on observed daily meteorological data over 1973–2004, two successive 9-year periods referred to as wet (P1976–1985 = 1205 mm/yr) and dry (P1986–1995 = 796 mm/yr) periods were selected. The calibration yielded similar Nash–Sutcliffe efficiencies (NSE) at the monthly time scale for both periods (NSEwet = 0.86; NSEdry = 0.90) supporting the model’s ability to adapt its structure to changing climatic situations. The simulation was extended in scarce data conditions over 1931–1972 and the simulation of monthly discharge values was acceptable (NSE = 0.71). When precipitation in the model was increased until it reach the change observed in the 1970s (DP=P ¼ 22%), the resulting increase in streamflow was found to closely match the 1970s hydrological change (DQ=Q ¼ 45%). Sensitivity analyses revealed that the land cover changes had a minor impact on the 1970s hydrological changes in the Sali-Dulce Basin. Integrating the SWAT simulations within the lake model over 1973–2004 provided lake level variations similar to those obtained using observed discharge values. Over the longer period, going back to 1931, the main features of lake levels were still adequately reproduced, which suggests that this basin-lake model is a promising approach for simulating long-term lake level fluctuations in response to climate.