Heatwaves (HWs) are high-impact phenomena stressing both societies and ecosystems. Their intensity and frequency are expected to increase in a warmer climate over many regions of the world. While these impacts can be wide-ranging, they are potentially influenced by local to regional features such as topography, land cover, and urbanization. Here, we leverage recent advances in the very high-resolution modelling required to elucidate the impacts of heatwaves at these fine scales. Further, we aim to understand how the new generation of km-scale regional climate models (RCMs) modulates the representation of heatwaves over a well-known climate change hot spot. We analyze an ensemble of 15 convection-permitting regional climate model (CPRCM, ~ 2–4 km grid spacing) simulations and their driving, convection-parameterized regional climate model (RCM, ~ 12–15 km grid spacing) simulations from the CORDEX Flagship Pilot Study on Convection. The focus is on the evaluation experiments (2000–2009) and three subdomains with a range of climatic characteristics. During HWs, and generally in the summer season, CPRCMs exhibit warmer and drier conditions than their driving RCMs. Higher maximum temperatures arise due to an altered heat flux partitioning, with daily peaks up to ~ 150 W/m² larger latent heat in RCMs compared to the CPRCMs. This is driven by a 5–25% lower soil moisture content in the CPRCMs, which is in turn related to longer dry spell length (up to double). It is challenging to ascertain whether these differences represent an improvement. However, a point-scale distribution-based maximum temperature evaluation, suggests that this CPRCMs warmer/drier tendency is likely more realistic compared to the RCMs, with ~ 70% of reference sites indicating an added value compared to the driving RCMs, increasing to 95% when only the distribution right tail is considered. Conversely, a CPRCMs slight detrimental effect is found according to the upscaled grid-to-grid approach over flat areas. Certainly, CPRCMs enhance dry conditions, with knock-on implications for summer season temperature overestimation. Whether this improved physical representation of HWs also has implications for future changes is under investigation.

The weather forecasts for precipitation have considerably improved in recent years thanks to the increase of computational power. This allows for the use of both a higher spatial resolution and the parameterization schemes specifically developed for representing sub-grid scale physical processes at high resolution. However, precipitation estimation is still affected by errors that can impact the response of hydrological models. To the aim of improving the hydrological forecast and the characterization of related uncertainties, a regional-scale meteorological–hydrological ensemble is presented. The uncertainties in the precipitation forecast and how they propagate in the hydrological model are also investigated. A meteorological–hydrological offline coupled ensemble is built to forecast events in a complex-orography terrain where catchments of different sizes are present. The Best Discharge-based Drainage (BDD; both deterministic and probabilistic) index, is defined with the aim of forecasting hydrological-stress conditions and related uncertainty. In this context, the meteorological–hydrological ensemble forecast is implemented and tested for a severe hydrological event which occurred over Central Italy on 15 November 2017, when a flood hit the Abruzzo region with precipitation reaching 200 mm (24 h)⁻¹ and producing damages with a high impact on social and economic activities. The newly developed meteorological–hydrological ensemble is compared with a high-resolution deterministic forecast and with the observations (rain gauges and radar data) over the same area. The receiver operating characteristic (ROC) statistical indicator shows how skilful the ensemble precipitation forecast is with respect to both rain-gauge- and radar-retrieved precipitation. Moreover, both the deterministic and probabilistic configurations of the BDD index are compared with the alert map issued by Civil Protection Department for the event showing a very good agreement. Finally, the meteorological–hydrological ensemble allows for an estimation of both the predictability of the event a few days in advance and the uncertainty of the flood. Although the modelling framework is implemented on the basins of the Abruzzo region, it is portable and applicable to other areas.

The response of Mediterranean small catchments hydrology to climate change is still relatively unexplored. Regional Climate Models (RCMs) are an established tool for evaluating the expected climate change impact on hydrology. Due to the relatively low resolution and systematic errors, RCM outputs are routinely and statistically post-processed before being used in impact studies. Nevertheless, these techniques can impact the original simulated trends and then impact model results. In this work, we characterize future changes of a small Apennines (Central Italy) catchment hydrology, according to two radiative forcing scenarios (Representative Concentration Pathways, RCPs, 4.5 and 8.5). We also investigate the impact of a widely used bias correction technique, the empirical Quantile Mapping (QM) on the original Climate Change Signal (CCS), and the subsequent alteration of the original Hydrological Change Signal (HCS). Original and bias-corrected simulations of five RCMs from Euro-CORDEX are used to drive the CETEMPS hydrological model CHyM. HCS is assessed by using monthly mean discharge and a hydrological-stress index. HCS shows a large spatial and seasonal variability where the summer results are affected by the largest decrease of mean discharge (down to −50%). QM produces a small alteration of the original CCS, which generates a generally wetter HCS, especially during the spring season.