The forecast quality of multi-model seasonal-to-decadal climate predictions, as measured by metrics of, among others, accuracy and reliability, has been traditionally estimated considering time-average products and products for event thresholds that do not target the occurrence of unusual events of either monthly or seasonal duration. However, there is an increasing interest in some user communities for products that represent extreme and unusual events. This presentation will discuss the differences in forecast quality between traditional forecast products, like mean seasonal temperature, and products for intraseasonal extremes (e.g., those measured with the 95th percentile of high-frequency temperature over periods like a month or a season) and monthly and seasonal unusual events (such as the frequency of exceeding the 90th percentile of the daily climatological distribution of temperature at a given time of the year). The results will be discussed in the context of their implications to address a number of user requirements from different sectors. The relevance of the forecast quality estimated from hindcast sets and the role of the observational uncertainty will be discussed when delivering forecast products in a climate service context. The implications of this work for the standardisation of climate services based on climate predictions will also be discussed.

How to cite: Doblas-Reyes, F. J., Agudetse, V., Delgado-Torres, C., Donat, M. G., González-Reviriego, N., De Luca, P., Milders, N., G. Muñoz, A., Palma, L., Pérez-Zanón, N., Ramon, J., Solaraju-Murali, B., Soret, A., and Torralba, V.: Forecast quality of climate extreme predictions and its relevance for climate services, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-11143, https://doi.org/10.5194/egusphere-egu23-11143, 2023.

Skilful predictions of near-term climate extremes are key to a resilient society. However, standard methods of analysing seasonal forecasts are not optimised to identify the rarer and most impactful extremes. For example, standard tercile probability maps, used in real-time regional climate outlooks, failed to convey the extreme magnitude of summer 2022 Pakistan rainfall that was widely predicted by seasonal forecasts. We argue that in this case, a strong summer La Niña provided a window of opportunity to issue a much more confident forecast for extreme rainfall than average skill estimates would suggest. We explore ways of building forecast confidence via physical understanding of dynamical mechanisms, perturbation experiments to isolate drivers, and simple empirical relationships. We highlight the need for more detailed routine monitoring of forecasts, with improved tools, to identify regional climate extremes and hence exploit windows of opportunity to issue trustworthy and actionable early warnings.

How to cite: Dunstone, N., Smith, D., Hardiman, S., Ineson, S., Jain, S., Martin, G., and Scaife, A.: Windows of opportunity for predicting seasonal climate extremes: Pakistan floods of 2022, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-7799, https://doi.org/10.5194/egusphere-egu23-7799, 2023.

It is known from previous studies that the winter windstorm season is significantly predictable on a seasonal timescale, especially over the British Isles and southern Scandinavia. Winter windstorms are one of the most damaging extreme events for the European continent. Hence, it is important to know that this skill exists as well as to understand how the forecast model reaches this performance to increase the usability of such forecasts.

Here, we link these extreme events to the three most dominant large-scale weather patterns over Europe. A combination of the three leading patterns explains up to 80% of the variability in windstorm frequency and ~60% of storm intensity. A statistical multi-linear model based on these patterns shows similar areas of skill but with lower skill over Europe.

This new investigation uses multiple dynamical atmospheric factors known to be related to windstorms, cyclones, their intensification and genesis. Among the factors examined are jet stream strength and location, Rossby wave source, Eady growth rate and potential vorticity. To understand the influence of these factors on windstorm forecast skill, we apply a three step conceptual approach: first to understand the link between windstorms in observations and hindcasts. Second, we analyse the forecast skill of the factors themselves. In the last step we diagnose significant changes in forecast skill of the dynamical factors between well and poorly predicted windstorm years.

Factors like MSLP, tropical Atlantic rainfall, jet location, PV in 350K, or Eady Growth Rate all show significant results in individual steps but none of the dynamical factors show significant results in all 3 steps. This could mean that an improved representation of factors and their link to windstorms could improve windstorm seasonal forecast skill.

How to cite: Degenhardt, L., Leckebusch, G. C., and Scaife, A. A.: Assessing the influence of dynamical factors on seasonal skill of severe winter windstorm predictions, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-2808, https://doi.org/10.5194/egusphere-egu23-2808, 2023.

Understanding large-scale drought patterns and the mechanisms producing extreme drought events is vital to understanding future drought risks. Here we investigate the influence of teleconnections originating in the Pacific and Atlantic Oceans on regional drought variability in the United States. For this work, the Standardized VPD Drought Index (SVDI) is calculated with Vapor Pressure Deficit (VPD), a method of drought detection based on air temperature and relative humidity, rather than precipitation deficit. This work focuses on summer months between 1980 – 2021, and SVDI was calculated with NASAs North American Land Data Assimilation System (NLDAS) data. Initially, the spatial drought characteristics were extracted from SVDI with EOF analysis. To identify extreme events, a k-means clustering technique was applied to primary principal components, multivariate ENSO index (MEI), and northern hemisphere sea surface temperature anomalies (SSTA). Additionally, to identify mechanisms driving drought variability, statistics from individual clusters (i.e. drought events) are retained for analyses using atmospheric variables (e.g. wind, HGT, and MSLP). Results show large-scale drought in the Central and Southern U.S. stem from mechanisms originating in the northern Pacific Ocean (e.g. PDO and SSTA trends) and northern Atlantic Ocean (e.g. NAO), modulating the variability in the onshore flow along the Gulf of Mexico. However, mechanisms influencing summer drought patterns in the southwestern U.S., especially the recent long-term drought patterns, originate in the equatorial Pacific Ocean and are driven by ENSO related processes.

How to cite: Gamelin, B., Bessac, J., Rao, V., and Altinakar, M.: ENSO and extra-tropical ocean variability drives summer drought extremes in the United States, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-2904, https://doi.org/10.5194/egusphere-egu23-2904, 2023.

The temperature observed in Taiwan has been on the rise for the past 100 years. At the same time, the number of summer days in the 2020s has increased significantly compared with that in the 1960s, while the frequency and intensity of heatwave events are also increasing. Extremely high-temperature (EHT) and heat wave events will cause huge effects on human activities, therefore, pre-warning of EHT and heat wave events is essential.

This research investigates the association between the Taiwan summer temperature and the Pacific Meridional Mode (the PMM), an anomalous north-south sea surface temperature gradient over the northeastern subtropical Pacific. Because of the PMM's 5-6 years cycle, it is a good predictability source on a decadal timescale. It was found that when the PMM was in a positive (negative) state, the summer temperature in Taiwan significantly increased (decreased) on a decadal timescale. The zonal circulation in the sub-tropical north Pacific and the subsidence in Taiwan were considered to be the physical mechanism in the linkage.

Besides, the calendar day 90th percentile of max temperature (CTX) and heatwave magnitude scale (HWMS) were used in this research to trace the extremely high-temperature and heat wave events. The research results indicate that during the PMM-positive state, the frequency and duration of EHT and heat wave events become higher and longer than they are during the negative state.

This linkage found in research could help to improve the decadal prediction of summer temperature as well as EHT and heat wave events in Taiwan and provide climate information for the decision-makers to formulate adaptation strategies.

How to cite: Tsai, C.-T., Wang, Y.-C., and Tseng, W.-L.: Linkage between the summer hot extremes in Taiwan and Pacific Meridional Mode, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-10321, https://doi.org/10.5194/egusphere-egu23-10321, 2023.

It has been demonstrated that modes of climate variability influence ocean wave and wind climate variability at inter-annual and multi-decadal scales (Odériz et al., 2021), trigger local impacts modifying coastal erosion patterns (Barnard et al., 2015) and disturbing seasonal coastal risk (Wahl & Plant, 2015). Besides, extreme events and modes of variability that have occurred simultaneously have above-normal struck coastlines around the world (Barnard et al., 2017) and have evidenced that omitting them in hazard forecasting can lead to underestimating coastal impacts. Moreover, in long-term analysis, the internal natural variability that modes of variability cause on wave climate can mask global warming trends in areas with vast natural fluctuations, such as the Southern Ocean (Odériz et al., 2021).
The complexity of spatiotemporal scales, in addition to a misconception of teleconnections, have led modes of variability to not integrate into coastal management and underestimate their impact on physical and biological hazards. This study identifies energetic and calm teleconnections induced by the leading polar (Arctic Oscillation-AO and Antarctic Oscillation-AAO) and tropical (El Niño Southern Oscillation-ENSO) modes of variability on the world’s coasts. Teleconnections are comprehensively characterized by (1) sign, (2) duration, (3) amplitude, and (4) spatial patterns. Global spatial-temporal fluctuations are analysed by season, parameter (near-surface wind velocity, total-wave power, swell-wave power, and wind sea-wave power), and planetary systems (winds and wave climates).
As an example of the results, we found that wind velocity increases up to ~+1m/s around Tuvalu Island, induced by La Niña (the negative phase of ENSO); in Chile induced by the positive phase of AO; while in Guinea, Indonesia, and Papua New Guinea this increase is triggered by El Niño (the positive phase of ENSO). In addition, the wave power of westerly swells increases up to ~+10 kW/m over an average season, induced by the positive phase of AO in Ireland, Norway, and the UK; in the USA induced by El Niño; and in Australia, New Zealand, and Chile influenced by the positive phase of AAO. This framework can serve as a source of predictability and provide a basis for a proactive response to coastal impacts in anomalous seasons and be transferred to financial risk and insurance instruments.

How to cite: Odériz, I., Losada, I. J., Silva, R., and Mori, N.: Inter-annual and multi-decadal climate variability in hazard forecasting can exacerbate coastal impacts, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-12718, https://doi.org/10.5194/egusphere-egu23-12718, 2023.

Fuelled by our changing climate, the summer of 2022 was one of the warmest on record, with numerous heatwaves and other weather extremes occurring around the globe. However, there is limited quantitative evidence on the contribution of human-induced warming to the weather-related health impacts observed in recent extreme weather events. We present a health attribution analysis of heat-related mortality attributable to human-induced climate change in the past summer of 2022 in Switzerland. We combined state-of-the-art methods in climate science and epidemiology with high-resolution mortality and temperature data to estimate the number and fraction of all-cause deaths that could be attributed to heat between June and August 2022. We, thus, estimated that 623 [95% CI: 151 - 1,068] all-cause deaths can be attributed to heat between June and August 2022, representing 3.5% [95% CI: 0.9-6.1] of total all-cause mortality during the same period. In a second step, we modelled counterfactual daily temperatures representing summer 2022 in absence of anthropogenic climate change. Specifically, four counterfactual daily mean temperature series were derived by subtracting the anthropogenic signal from the observed series which ranged between 1.19 and 2.75 ºC. Then, we quantified the hypothetical heat mortality burden in absence of climate change, and finally, estimate the contribution of climate change by subtracting it from the observed heat mortality impacts. We estimated that, in absence of an anthropogenic signal, the heat-related burden would have amounted to 1.4% [95% CI: -0.2 - 3.4] of all-cause mortality, corresponding to 253 deaths [95% CI: -27;594]. Thus, 2.1% [95% CI: 0.8 - 3.7] of the all-cause mortality in summer 2022 would have been avoided in absence of anthropogenic warming. This corresponds to 370 [95% CI: 133-644] deaths and 60% of the observed heat-mortality burden between June-August 2022. Females and the oldest age group were the most affected. Specifically, 60% of heat-related deaths attributed to climate change were in females (220 [69 - 393] vs. 150 [62 - 250] in males), and 90% in older adults (330 [129-565] vs. 39 [-5 - 84]). Our findings confirm that climate change is already affecting the health of the population in Switzerland by amplifying the heat-related mortality burden, with a stronger impact on females and older adults.

How to cite: Vicedo Cabrera, A. M., De Schrijver, E., Ragettli, M. S., Shumacher, D., Fischer, E., and Seneviratne, S.: Heat mortality during summer 2022 in Switzerland attributable to human-induced climate change, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-5169, https://doi.org/10.5194/egusphere-egu23-5169, 2023.

Tropical cyclones (TCs) are among the most devastating natural hazards putting populations and assets at risk. This risk is expected to increase further in a warming climate and with socio-economic development. It is, therefore, of great importance and the aim of our study to assess the drivers and uncertainties of global TC risk in the future. We use a large set of synthetic TCs downscaled from various general circulation models (GCMs) and different warming scenarios of the CMIP6 generation to simulate TC activity at the middle and end of this century. In parallel, we derive economic growth factors from different Shared Socioeconomic Pathways (SSPs) to approximate socio-economic development. We combine these future representations of hazard and exposure data with vulnerability functions to estimate the TC risk increase in the future, using an open-source probabilistic impact model (CLIMADA). Furthermore, we perform a systematic uncertainty and sensitivity analysis to understand which of the model input factors contribute most to the uncertainty in the future TC risk increase. First, we find a non-linear effect between climate change and socio-economic development that drives the total future risk. Second, we show that the choice of GCM affects the output uncertainty most among all varied input factors. However, we note that exposure and vulnerability data are notoriously sparse and that advances in future TC risk assessment also depend on a better representation of these components. Ultimately, unraveling these unknowns of global TC risk in the future may help focus future research efforts and enables better-informed adaptation decisions and mitigation strategies.

How to cite: Meiler, S., Emanuel, K., and Bresch, D. N.: Unraveling the unknowns of global tropical cyclone risk in the future, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-1323, https://doi.org/10.5194/egusphere-egu23-1323, 2023.

Flood hazards from Tropical cyclones (TCs) are frequently the leading cause of mortality and damages. Current research indicates that TC rainfall will increase by 7 % per degree of ocean warming with a greater proportion of them being extreme . It is vital to understand how this increase in rainfall will translate to flood risk from tropical cyclones by first accurately modelling TC rainfall under present climatic conditions.

General circulation models struggle to reproduce TC rainfall fields so downscaling models are often used to generate more realistic TC rainfall data. Increasingly, rainfall downscaling studies have adopted deep learning techniques from the Computer Vision field to achieve comparable results to traditional methods at a fraction of the computational cost. Initially, convolutional neural networks (CNNs), specifically U-NETs, showed promise in precipitation downscaling. But more recently, the use of Generative models has been explored following the success of GANs in classical image super-resolution problems compared to CNNs. Generative approaches have shown potential at reproducing the fine spatial detail and stochastic nature of precipitation.

Here, we develop upon the WGAN and Variational Autoencoder GAN (VAEGAN) from Harris et al. (2022) and apply it to rainfall data from TCs to increase the resolution of rainfall measurements from 100 km resolution to 10 km resolution.

Overall, the Wasserstein GAN, performed better than other methods, the variational autoencoder GAN, U-Net and bilinear interpolation, across all diagnostics explored. We showed that for regular TCs the WGAN had the most realistic power spectra for all wave numbers, closely followed by the VAEGAN which only deviated for scales of around 5 pixels or fewer. The U-Net and Bilinear Interpolation methods both reproduced power spectra poorly compared to observations, with significant differences present from wave numbers greater than 3. We found that the WGAN had the lowest mean bias overall with errors around the core of the TC within 5 % error, while the VAEGAN had a dry bias of over 5 % outside of the inner core region. Both models had a low negative bias in standard deviation of between 0-5 %.

When looking at the 100 most extreme samples, beyond the intensity of storms used in training the models, the WGAN is able to produce results of similar quality to those for TCs of intensities used in training, except for predictions having too low spread. This indicates that if the WGAN were trained on the full observational dataset, it could perform well for storms more intense than those previously observed, which is important for judging the model's robustness. Conversely, the power spectra of the VAEGAN became more unrealistic and predictions more artificial. There were some very large errors present in VAEGAN at the upper end of the extreme test set which demonstrates the importance of evaluating models on the most extreme, unseen, cases.  Overall, these results show that generative approaches have the potential to generate TC rainfall fields with a high degree of accuracy.

How to cite: Vosper, E., Watson, P., Harris, L., McRae, A., Santos-Rodriguez, R., Aitchison, L., and Mitchell, D.: Deep learning for downscaling tropical cyclone rainfall, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-8325, https://doi.org/10.5194/egusphere-egu23-8325, 2023.

To date most natural hazard risk assessments in Australasia do not incorporate long-term and/or prehistoric records of extreme events and coastal development continues to rely on short historical records as a reflection of the long-term behaviour of a hazard. In some locations such as southern China or the Philippines historical records may be appropriate as consistent records have been kept for several centuries or even a millennium. However, for much of the Asia-Pacific this is rare as the historical archives rarely stretch beyond World War II. Clearly such short records are inadequate for determining the natural variability of a hazard at multi-decadal timescales and for the extrapolation of extreme events.  While it is well known that the historical record is fragmentary, incomplete and limited in spatial balance, the historical record does provide a key link between instrumental datasets and the prehistoric record that allows for the detailed reconstruction of past events. Here, we compare known historical tropical cyclone events to recent ones in Western Australia (UC1921 and cyclone Herbie) and the central Philippines (Typhoon Haiyan and Ty1897) as examples of integrated studies. The two examples demonstrate the utility of the integrated approach and allow an examination of the similarities and differences between the events. Such efforts must become familiar to those outside of academia, as familiarity breeds awareness and it is through awareness and adoption that the true potential of integrating across disciplines will be recognized.

How to cite: Switzer, A. D., Christensen, J., and Soria, L.: Comparing modern and historical records of cyclone induced extreme sea level events : examples from Australia and the Philippines., EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-10428, https://doi.org/10.5194/egusphere-egu23-10428, 2023.

The temporal compounding of two contrasting extremes of the hydrological spectrum (droughts and floods) reflects a volatile hydrological cycle and makes water resources management more challenging. To this end, the present study examines extreme wet-dry events and their temporal compounding over the Upper Jhelum Basin (UJB) in the future climate context based on simulations of climate models from three modeling initiatives (CMIP6, CORDEX - WAS-44 domain and CORDEX-CORE - WAS-22 domain) under low, medium and high emission scenarios for two-time segments i.e., near future (2040-2059) and far future (2080-2099). Wet and dry events are characterized by using a multivariate drought index (namely the Standardized Precipitation Evapotranspiration Index, SPEI), which is derived from daily precipitation and maximum and minimum temperatures. The temporally compound event (hereafter referred as compound event-CE), which is defined as a successive transition from one powerful state to another, includes dry to wet (D-to-W) events and wet to dry (W-to-D) events in the adjacent month. Therefore, the minimum duration of a compound event is 2 months. A D-to-W compound event is defined as a dry spell (SPEI_i ≤−1) abruptly changing into a wet spell (SPEI_i+1 ≥1) in the next month. Conversely, a W-to-D compound event is defined as a wet spell (SPEI_i ≥1) abruptly changing into a dry spell (SPEI_i+1 ≤−1) in the next month. The statistical interdependency of temporally compound wet and dry events (CEs) and their statistical significance are investigated using event coincidence analysis (ECA). The significance test is performed based on the assumption of a Poisson process with the null hypothesis that the successive occurrence of wet and dry event is randomly distributed. Results show that the probability of D-to-W CEs is much higher than the W-to-D CEs under both retrospective (i.e., past) or prospective (i.e., future) climate contexts. Specifically, the probability of D-to-W events is high in the southwest of the basin (up to 80 %, statistically significant at 5% level) both in the historical and projections. In contrast, the W-to-D CEs are found to be statistically non-significant for a 95% confidence level (up to 40 %) with no clear pattern of occurrence. There are some differences depending on the climate model ensembles used. CORDEX models (WAS44 and WAS22) show decreasing probability of D-to-W CEs in the southwest part of the basin by the end of the century whereas the CMIP6 ensemble shows a negligible increase from near to far future especially under the highest emission scenario. Overall, the CMIP6 ensemble presents higher probability of CEs under all scenarios and time segments. Climate projections of this kind of extreme events, spanning different scenarios and all sources of uncertainty are essential to fully characterize their impacts on water-related sectors and to plan possible adaptation strategies, such as developing more efficient reservoir operation rules or agricultural planning.

How to cite: Ansari, R., Casanueva, A., and Grossi, G.: Changes in the Probability of temporally compound wet and dry events in a warmer world, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-6223, https://doi.org/10.5194/egusphere-egu23-6223, 2023.

Communicating the uncertainty of natural climate variability to the public and researchers from other fields remains challenging. In this context, the concept of time of emergence (ToE) i.e., the year or decade when the climate signal emerges from the natural climate variability, has been well established over the past years. In addition, global warming levels (GWLs) are used more and more frequently to define the future projection horizon. However, only a few studies combined these two approaches. In this study, we utilize multiple initial condition large ensembles from CMIP6, to more robustly sample extreme events and account for natural climate variability, to estimate the global warming level of emergence (GWLoE) of various ETCCDI indicators. These indicators were selected to represent both precipitation and temperature extremes. Further, we analyze the impact of incremental temperature changes on the emergence of these indicators. Additionally, the GWLs are analyzed in relation to changes in the probability risk ratio to highlight that every degree of additional warming counts. Different scenarios for population changes are applied to estimate the population affected by the emergence of indicators as well as for a doubling in probability risk ratio. The combined GWLoE of all large ensembles highlights considerable regional differences among the individual ensembles. Similarly, regional differences arise for the GWL related to a doubling in probability risk ratio. The changes in population affected by these changes in risk ratio highlight the need to limit global warming as much as possible.

How to cite: Wood, R. R., Gampe, D., Böhnisch, A., Mittermeier, M., and Schwingshackl, C.: Global Warming Level of Emergence and related population exposure to temperature and precipitation extremes, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-15385, https://doi.org/10.5194/egusphere-egu23-15385, 2023.

Anthropogenic climate change leads to more extreme heat in most regions around the world, in greater magnitude and longer durations. With a global perspective in mind, we explore equity impacts of physiologically deadly heat in cities. In a warming climate, heat extremes reaching a physiological threshold is a phenomena predominantly affecting regions closer to the equator, in contrast to heatwaves defined as a deviation from the mean which appear more frequently in all parts of the world. Many of the regions hit by deadly heat are also vulnerable from factors such as economic challenges, or other climate hazards posing fundamental threats to societal stability and livability.
We here quantitatively explore the intersection of socio-economic variables with previously described occurence of deadly heat in urban environments. These variables include current and future GDP estimates, Gini coefficients and population dynamics. We use metrics of inequality borrowed from economics. We intersect global deadly heat in future climate scenarios with urbanization dynamics. Sub-Saharan Africa stands out as a region where both trends are pronounced. We also demonstrate that countries with responsibility for high historic emissions are overall less affected by the extreme heat.
The insights provided by this research contribute to an improved general understanding of global inequality and climate change as a driver of these. They further contribute to the loss and damage discussion.

How to cite: Lohrey, S. and Creutzig, F.: Urban deadly heat and global inequality, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-12122, https://doi.org/10.5194/egusphere-egu23-12122, 2023.