Assessing the Frequency, Duration and Intensity of Heatwaves from a Dynamically Downscaled Initial-Conditions Large Ensemble

One of the most direct impact of anthropogenic climate change is the increasing severity of heatwaves, which already act today as a major threat to human health and especially for the most vulnerable populations. For instance, during the well-known European heatwave of August 2003, about 15,000 excess deaths were attributed to the occurrence of this event in France (Fouillet et al. 2006). While extreme heatwaves are increasingly observed all around the world (e.g. Australia 2019, India and Pakistan 2019, Europe 2018, Russia 2010), it is worth noting that they also occurred in a more distant past (e.g. the "Dust Bowl" heatwave of 1936 in United States). Therefore, it is not always clear to which extent such extreme events can solely be attributed to anthropogenic greenhouse gases emissions. In fact, the climate system is affected by a relatively high level of internal variability owing to non-linear interactions between its internal components (atmosphere, ocean, land, cryosphere, and biosphere). This leads to a range of climate outcomes that always remain possible, irrespective of the rate of the greenhouse gases emissions or of the climate system's response to human-induced forcing. Therefore, natural climate variability may also lead to the occurrence of extreme events such as heatwaves.

The impacts of heatwaves on human health depend on several factors related to the vulnerability, exposure or adaptative capacity of our society. Naturally, these impacts also depend on various characteristics of the heatwave itself, notably its duration, but also on concurrent states of other variables. For instance, the human body cools itself by releasing latent heat through transpiration, so high relative humidity combined with high temperatures generally amplifies the impacts on human compared to high temperatures taken alone. But also, both the daily minimum and maximum surface air temperatures seem to have an impact on mortality during a heatwave event, as high temperature during the night prevents the body to recover from high temperatures that were experienced during the day. Therefore, different meteorological variables as well as the duration of heatwaves are playing a role in terms of impacts. High-impact heatwaves must therefore be treated as multivariate events, or compound events, which raise several issues regarding characterization of these events, as well as to their paucity in both observational records and climate model simulations.

In this study, we aim to disentangle the forced response (due to anthropogenic forcing) and the internal variability in terms of their relative contributions to the occurrence of heatwaves. This research is conducted using the new Canadian Regional Climate Model version 5 Large Ensemble (CRCM5-LE; Leduc et al. 2019), which was produced within the ClimEx project of the Bavaria-Québec international collaboration on climate change. This unprecedented dataset consists of a dynamically downscaled version of the Canadian Earth System Model version 2 (CanESM2) 50-member initial conditions ensemble toward a 12-km resolution grid mesh from 1950 to 2100 (following RCP8.5) over two domains covering northeastern North America and Europe respectively. This ensemble allows to obtain a large sample of high-resolution climate model simulations, which can be considered as independent realizations of the internal climate variability under the same anthropogenic greenhouse gases forcing.

Within this framework, heatwaves are studied as compound events at the local scale and are characterized according to their duration, intensity and frequency. By first defining a reference heatwave event based on observational records using daily minimum (Tmin) and maximum (Tmax) temperature thresholds, these characteristics are thus used to assess the occurrence of the event in the ensemble of simulations. The Montreal 2010 heatwave is taken here as the reference event and is characterized by values of Tmin=20°C and Tmax=33°C that where exceeded during 5 consecutive days. Correcting these thresholds values using the CRCM5 bias from observations, results from the large ensemble over Montreal show that such 5-day or longer events happen about once in 25 years in the present climate and becomes commonly found (about twice a year) by the end of the 21st century. Moreover, such events could last for an appreciable fraction of summers in the future, as some 30-day (or longer) events are detected in the ensemble from 2066 and especially after 2090. While it is worth recalling that the CRCM5-LE simulations were forced using the high-end RCP8.5 emission scenario, the contribution from internal variability is important. The use of a large initial-conditions ensemble allows a detailed characterization of long-duration heatwaves, which would not be possible within the classical single-realization framework.

References:

Fouillet, A. , G. Rey, F. Laurent, G. Pavillon, S. Bellec, C. Guihenneuc-Jouyaux, J. Clavel, E. Jougla and D. Hémon (2006) Excess Mortality Related To the August 2003 Heat Wave in France, International Archives of Occupational and Environmental Health, 80(1), 16–24.

Leduc M., A. Mailhot, A. Frigon, J.-L. Martel, R. Ludwig, G.B. Brietzke, M. Giguère, F. Brissette, R. Turcotte, M. Braun et J. Scinocca (2019) ClimEx project: A 50-member ensemble of climate change projections at 12-km resolution over Europe and northeastern North America using the fifth-generation Canadian Regional Climate Model (CRCM5). J. Appl. Meteor. Climatol., 58, 663–693.