Stochastic system dynamics modelling for climate change water scarcity assessment of a reservoir in the Italian Alps

Abstract

Abstract. Water management in mountain regions is facing multiple pressures due to climate change and anthropogenic activities. This is particularly relevant for mountain areas where water abundance in the past allowed for many anthropogenic activities, exposing them to future water scarcity. Here stochastic system dynamics modelling (SDM) was implemented to explore water scarcity conditions affecting the stored water and turbined outflows in the Santa Giustina (S. Giustina) reservoir (Autonomous Province of Trento, Italy). The analysis relies on a model chain integrating outputs from climate change simulations into a hydrological model, the output of which was used to test and select statistical models in an SDM for replicating turbined water and stored volume within the S. Giustina dam reservoir. The study aims at simulating future conditions of the S. Giustina reservoir in terms of outflow and volume as well as implementing a set of metrics to analyse volume extreme conditions. Average results on 30-year slices of simulations show that even under the short-term RCP4.5 scenario (2021–2050) future reductions for stored volume and turbined outflow are expected to be severe compared to the 14-year baseline (1999–2004 and 2009–2016; −24.9 % of turbined outflow and −19.9 % of stored volume). Similar reductions are expected also for the long-term RCP8.5 scenario (2041–2070; −26.2 % of turbined outflow and −20.8 % of stored volume), mainly driven by the projected precipitations having a similar but lower trend especially in the last part of the 2041–2070 period. At a monthly level, stored volume and turbined outflow are expected to increase for December to March (outflow only), January to April (volume only) depending on scenarios and up to +32.5 % of stored volume in March for RCP8.5 for 2021–2050. Reductions are persistently occurring for the rest of the year from April to November for turbined outflows (down to −56.3 % in August) and from May to December for stored volume (down to −44.1 % in June). Metrics of frequency, duration and severity of future stored volume values suggest a general increase in terms of low volume below the 10th and 20th percentiles and a decrease of high-volume conditions above the 80th and 90th percentiles. These results point at higher percentage increases in frequency and severity for values below the 10th percentile, while volume values below the 20th percentile are expected to last longer. Above the 90th percentile, values are expected to be less frequent than baseline conditions, while showing smaller severity reductions compared to values above the 80th percentile. These results call for the adoption of adaptation strategies focusing on water demand reductions. Months of expected increases in water availability should be considered periods for water accumulation while preparing for potential persistent reductions of stored water and turbined outflows. This study provides results and methodological insights that can be used for future SDM upscaling to integrate different strategic mountain socio-economic sectors (e.g. hydropower, agriculture and tourism) and prepare for potential multi-risk conditions.

Future droughts in northern Italy: high-resolution projections using EURO-CORDEX and MED-CORDEX ensembles

“…Future droughts and water scarcity were investigated for the European Alps, using RCM and GCM simulations (e.g. Terzi et al 2021;Brunner et al 2019). Owing to the significant precipitation decrease, future drought events for southern Italy were explored (Critto et al 2016).…”

Development of black ice prediction model using GIS-based multi-sensor model validation

“…In the model used in this study, the relationship between road moisture and freezing corresponded to a balanced loop. In the final stage, freezing and black ice have a positive relationship, and thawing and black ice have a negative relationship (Sterman, 2000;Terzi et al, 2021). In other words, freezing and thawing per hour are determined by several factors, and the amount of black ice generated per hour is predicted accordingly.…”

“…Black ice modelling for GIS simulation was performed based on system dynamics, a method of simulating dynamic phenomena by connecting the causal relationships of factors in the form of a block diagram (Kim, 2021;Forrester, 1994). Road freezing is difficult to predict with human intuition, as it occurs due to the time series relationship of factors; thus, system dynamics are more suitable for modelling this phenomenon (Terzi et al, 2021). The GIS is a tool for simulating system dynamics and conducting research on various disasters.…”

Future droughts in northern Italy: high resolution projections using EURO-CORDEX and MED-CORDEX ensembles

“…Future droughts and water scarcity were investigated for the European Alps, using RCM and GCM simulations (e.g. Terzi et al 2021;Brunnet et al 2019). Owing to the signi cant precipitation decrease, future drought events for southern Italy were explored (Critto et al 2016).…”