Future flood frequency curve of the Arno River (Central Italy) by using bias-corrected convection-permitting model projections in a semi-distributed hydrological model

Understanding how climate change affects the frequency and magnitude of floods is essential for adaptation strategies. Usually, the impact of climate change on extreme weather and floods is assessed using the projections of the Regional Climate Model (RCM) in a hydrological model. Nevertheless, RCM projections are usually too coarse to describe convective storms, and they can underestimate the intensity of short-duration extreme precipitation that is usually responsible for flash flood events in small river basins. For this reason, Convection Permitting Models have been proven to perform better than RCM in describing sub-daily extreme precipitation.

The objective of the work is to assess the impact of climate change on the flood frequency of the Arno River basin using CPM projections to also better describe future flood hazards in its small tributaries. The hydrological model used in the analysis is the Soil and Water Assessment Tool Plus (SWAT+). The projections used as input of the model are VHR-PRO_IT, Very High-Resolution Projections over Italy (Raffa et al., 2023). The projections have a 2.2 km spatial resolution and 1h temporal resolution, and they cover 90 years from 1981 to 2070 in the emission scenarios RCP 4.5 and RCP 8.5. First, the model has been calibrated with 15 years of observed data. Then, the projections have been bias-corrected and used as input in the continuous hydrological model. Therefore, SWAT+ performs a continuous hydrological simulation for 90 years at the hourly timestep.

The bias correction of the precipitation projections has been done with a parametric approach (Mamalakis et al., 2017) to adjust the frequency distribution of precipitation events. The correction of temperature projections has been done with an easier approach based on the linear scaling of monthly average temperatures. The bias correction used ground observations of rain gauges and thermometers of the Hydrological Regional Service of the Tuscany Region.

The results are expressed with a delta-change approach to extract possible trends in the simulated discharges. The delta change expresses the ratio between the peak discharge associated with a given return period T in the future and the peak discharge for the same frequency in the historical period. The Generalized Extreme Value Distribution (GEV) is used to fit the cumulative distribution of the annual maximum series for three-time windows of 30 years: 1981-2010 (Historical), 2011-2040 (Near Future), and 2041-2070 (Far Future). The results show an increase in the flood hazard in the city of Florence in the RCP8.5, especially in the far future.  

ACKNOWLEDGEMENT: The research is carried out within the RETURN – multi-Risk sciEnce for resilienT comUnities undeR a changiNg climate Extended Partnership and received funding from the  European Union Next-GenerationEU (National Recovery and Resilience Plan – NRRP, Mission 4, Component 2, Investment 1.3 – D.D. 1243 2/8/2022, PE0000005).