ABSTRACT
The Bermejo River, located in northern Argentina, has a flow regime controlled by precipitation. In an area characterized by its risk of flooding and land-sliding during the summer, seasonal precipitation forecast becomes a valuable tool for risk assessment and better management of hydric resources. This study focuses on identifying remote forcings of precipitation variability for the upper sub-basin of the Bermejo River Basin, and developing multiple linear regression models of areal spring precipitation (September to November), the beginning of the rainy season, considering predictors monitored on the preceding August. Positive rainfall anomalies in spring relate to higher monthly and maximum daily streamflow in the upper and lower sub-basins. Two forecast models arose as the ones with best performance when using leave-one-out-cross-validation. Predictors involved in these models (four and three predictors, respectively) emphasize the influence of the circulation in middle-low levels over the Pacific Ocean, as well as of the sea surface temperature in the El Niño region and the low-level meridional wind in tropical South America. The two models share similar performance metrics, although the model with less predictors has a better skill for the detection of normal and above-normal rainfall seasons.
Acknowledgements
Rainfall and streamflow data were provided by the National Weather Service and the Secretariat of Infrastructure and Water Policy of Argentina. Large scale variables were obtained from the NCEP-NCAR reanalysis. This research was supported by 2020–2022 UBACyT 20020190100090BA and UBACYT 2017–2019 20020160100009ba projects.
Disclosure statement
No potential conflict of interest was reported by the authors .