Physical and statistical analysis of climate extremes in large datasets

Societal-relevant information about climate change and changes in extreme events, such as heatwaves and coastal floods, are crucial for mitigation and adaptation decision making. Results from the ClimateXL project show that we need large ensembles (20-50 ensemble members) from single climate models to simulate strong heatwaves with very low probabilities like the ones in Europe in 2003 or in Russia in 2010 and to see if even more extreme heatwaves could happen in today?s and future climate. Potentially, a better understanding of the relationship between heatwaves and atmospheric blocking can lead to improved preparedness for these events reducing some of their adverse effects on society. We also found that in large European cities, stabilizing climate warming at 1.5 °C rather than 2 °C would decrease extreme heat-related mortality by 15?22% per summer. High-population regions, such as Central Africa, India and Europe, could experience an additional 10-20 days of extreme heat on average every year under a scenario with 2 °C warming, as compared to 1.5 °C. In the recent years, Africa already experienced hotter, longer and more extent heat waves than in the last two decades of the 20th century. In the future, heat waves in Africa that are unusual under present climate conditions will occur on a regular basis by 2040 if we continue to emit greenhouse gases at current rate. In particular, the co-occurrence of consecutive hot and humid days during a heat wave can strongly affect human health. Considering the effect of humidity at 1.5° and 2° C global warming, highly populated regions, such as the Eastern US and China, could experience heat waves with magnitude greater than the one in Russia in 2010 (the most severe of the present era). The apparent temperature peak during such humid-heat waves can be greater than 55 °C, which is a very critical threshold at which humans are very likely to suffer from heat strokes. Humid-heat waves with these conditions were never exceeded in the present climate but are expected to occur every other year at 4° global warming. This calls for respective adaptation measures in some key regions of the world along with international climate change mitigation efforts. In another study, extreme temperature and precipitation were put in relationship with each country?s per capita CO2 emissions, income and vulnerability and we found that the countries most affected by large changes in extreme heat and precipitation are also characterized by small CO2 emissions, low-income and high vulnerability. Achieving the 1.5 °C goal therefore limits the unequal distributions of extreme weather events and impacts. In addition, strengthening the socio-economic development of the most vulnerable, i.e. least developed countries, will significantly reduce severe impacts of climate change, for instance due to heatwaves, in those countries. The conclusion is that there are clear scientific and ethical arguments for stabilizing warming at 1.5 °C rather than 2 °C in combination with accomplishing the sustainable development goals. Our work with robust decision approaches highlights the importance to understand and quantify uncertainties in hydrologic projections and how it can be coupled with concrete decision-problems framed by the needs of the end-users using statistical formulations. Different components of uncertainty have been visualized in the different examples and demonstrate the value of uncertainties. The work using Katwijk as an illustrative example, simulates flexible timing of investment decisions and uncertain climate-induced sea-level rise, detailing costs and benefits of different types of coastal flood protection measures. The work using Bergen and Esbjerg as examples, discusses how the optimal adaptation timing depends on the decision-makers loss function/risk aversion and illustrates that including uncertainty is vital, where uncertainty on the damage costs has the largest effect.

ClimateXL provided important insights for preparedness and adaptation planning to reduce severe impacts from climate extremes. Uncertainties in simulating extremes in models can be reduced or better quantified when using large ensembles and appropriate statistical methods. In a decision making context, we reveal that uncertainties related to damages and socio-economic consequences from the impacts of climate change, such as sea level rise and coastal flooding, are much higher than the uncertainties related to climate simulations. Overall, societal consequences due to climate change can be significantly reduced when limiting global warming to 1.5 degree. Hence, private and public sectors as well as policy makers should make substantial efforts to mitigate climate change. Our results highlight also the need for investing in adaptation as we are seeing an increasing frequency and intensity of climate extremes that affect various sectors already under current climate.

ClimateXL brings together national and international experts from climate sciences, statistics and economics to address some of the World Climate Research Program (WCRP) Grand Challenges on Extremes. Weather and climate extremes (e.g., heavy precipitation events and flooding) can cause severe damages to infrastructure. The occurrence, frequency, and intensity of such extremes are determined by a complex interplay of natural and anthropogenic factors. Climate extremes, in general, are projected to increase and intensify significantly due to anthropogenic climate change until the end of this century. In the near-term, internally generated climate variability will dominate the anthropogenic greenhouse gas forced changes in climate extremes. Adaptation planning to the impacts of climate extremes is therefore challenged to take into account both near- and long-term changes and associated uncertainties. ClimateXL will investigate uncertainties related to simulations of climate extremes in climate models and, subsequently, improve our understanding of present and future changes in climate extremes. Based on the analysis of large ensembles of state-of-the-art global climate models (GCMs), we will be able to investigate the robustness of the relationship between important large-scale atmospheric patterns that determine near- and long-term variability in climate extremes. This knowledge will be incorporated in the development of advanced model performance metrics within a multivariate context. The ability of GCMs to represent climate extremes will be assessed using observational and high-resolution reanalysis datasets. The project will apply and further develop extreme value theory and methods of uncertainty assessment, which are the key concepts to gain a better understanding of changes in climate extremes. This knowledge will further feed into cutting-edge approaches to decision making for adaptation planning to avoid severe impacts of climate extremes on infrastructure.