Water Availability Assessment in a Data-Scarce Region: Application to the District of Southern Italy

Hydrological modeling is an essential tool for understanding and describing hydrological processes, serving as a cornerstone in the quantification and management of water resources. The major challenge of hydrological modeling lies in model calibration, which becomes particularly demanding in large-scale applications and in data-scarce regions.

Data scarcity is a significant constraint in modeling, complicating the calibration process and reducing model accuracy. Generally, the availability of high-quality streamflow measurements is considered vital for the calibration and evaluation of hydrological models. However, in many scenarios data may be of low quality, incomplete, or entirely unavailable, as it happens in many areas of the National Territory, including regions in Southern Italy where the streamflow observations are limited, fragmented and discontinuous. Most hydrometric stations record only water levels, often without updated flow rating curves, making reliable hydrological model calibration a challenging task. 

In order to overcome such limitations, we compared three different setups to get the best parametrization during the model calibration. At first, we used the biggest hydrological basin (Volturno river catchment) of the entire district, as representative of the regional study area. The calibration of the model was done for the representative catchment, and the parameters were applied at the regional scale.   Then, we used reconstructed streamflow measurements derived from water balance of nine artificial reservoirs as a reference for a multiobjective calibration. At last, we used remote sensing data, such as soil moisture maps, as a reference for calibrating the model. Multi-objective functions, focusing on high-flows and low-flows aspects of the time series, were used in automatic optimization based on genetic algorithms to perform space-time operational testing of the large-scale model. The reference hydrological model used is the DREAM model (Distributed model for Runoff, Evapotranspiration, and Antecedent Soil Moisture simulation), applied to the vast area within the jurisdiction of the Southern Apennine District Basin Authority.

These calibration procedures have been compared exploiting available data. The study provides guidance in the use of limited data in order to identify the most suitable approach to build a reliable model calibration of the entire district and assess the impact of climate change on water resources in future climate scenarios. The encouraging performances of the regional model motivate the extension of the present approach to other data-scarce regions.