In a scenario of climate changes forecasting the availability of water from springs is desirable to prevent shortage in water supply to the connected utilities like aqueducts and irrigation systems. That is, it would be useful to know the “signature” response of the aquifer.
This work focuses on the application of the ensemble smoother with multiple data assimilation (ES-MDA) to build a hydrological model for flow rate prediction in a karst aquifer using the well-known instantaneous unit hydrograph method and the rational formula. The aquifer studied is the Bossea-Artesinera karst aquifer, located in northwest Italy in the Maritime Alps with an infiltration basin of around 6 km2. The input data are observed discharge flow rates, daily precipitations, and temperatures. Some of the challenges that have been addressed in this research are the modelling of infiltration as a time-varying infiltration coefficient, the classification of precipitation into snowfall and rainfall, and the transformation of snow into water equivalent infiltration. Others parameters of the hydrologic model such as base flow, infiltration coefficient, or snow melting contribution were estimated.
As first step, individual events were examined and the ES-MDA produced excellent results with values of Nash-Sutcliffe efficiency never below 0.9. Then on the basis of 27 such events, two average models were identified: one to be applied in spring months when the impact of snow melting is important and the other during autumn months. These average models used for predictions did not work as remarkably as the individually fitted models: fall model performances were better that those of the spring model suggesting that the snow equivalent calculation has to be improved, being the main difference between the two seasons. It is difficult to conclude whether a signature exist for Bossea, still it is clear that the ES-MDA is a powerful tool to investigate single event and give insights on the aquifer dynamics.