A statistical approach to estimate global heatwave risk

Abstract

<p>Following the conceptual risk framing of the IPCC that defines risk as a function of hazard, exposure and vulnerability, we estimate global heatwave risk by using a statistical approach that combines the distribution of indicators for heatwave magnitude, population exposure and human development. We fit a general extreme value distribution (GEV) to the maxima of the heatwave magnitude index in a block of 10-years to estimate extreme heatwave conditions with a 500yr return period under current climate (HW500yr). To consider the impact of changes in heatwaves in populated regions of the world, we use a set of global, spatially explicit population projections that are consistent with the new Shared Socioeconomic Pathways (SSPs). As a proxy for vulnerability we use the Human Development Index (HDI) based on the geometric average of three dimensions: health, education and standard of living. We derive an illustrative heatwave risk indicator (expressed in %) for each location (i.e. grid box) of the globe as the product of the probability of occurrence of HW500Y multiplied by normalized population density and 1-HDI values with all components of the product being normalized. Using this illustrative heatwave risk indicator on a global scale, we project heatwave risk for global warming of 1.5 and 2 °C, in accordance with the Paris agreement, for two future pathways of societal development representing low and high vulnerability conditions (SSP1 and SSP4, respectively). This method demonstrates how including a measure of vulnerability could produce a distribution of risk that is different from the distribution of hazard or exposure under different scenarios. Our results show that the heatwave risk for the low and very high development countries would be significantly reduced if global warming is stabilized below 1.5 °C, and in the presence of rapid social development. The latter is most important for low development countries for decreasing their vulnerability towards heatwaves or other hazards being amplified by climate change. The results illustrate how hazard-specific policies could be better informed by analyses that account for vulnerabilities to the respective hazard. However, we also discuss several caveats associated with using a normalized risk indicator on a global scale with implications for the interpretation of risk on a local scale, including the need for better indicators to describe vulnerability to specific or multiple hazards.</p>