When don't we need a new extreme event attribution study?

When don't we need a new extreme event attribution study?

Abstract

The influence of anthropogenic climate change on the frequency and intensity of extreme weather events is becoming increasingly well understood. Extreme event attribution studies now exist for hundreds of events over the past few decades. However, there remain large heterogeneities in the number of attribution studies across the world and for different hazards, as well as limited capacity to conduct new studies. In this study, we suggest that there is more information to draw from past attribution studies about recent events. This is because, even though anthropogenic forcing continues to increase, many new events share meteorological characteristics with previously attributed ones. Here, we explore the possibility of using related studies and other lines of existing evidence such as projections and trend analysis to create rapid, low-resource attribution statements. To do this, we discuss the potential use cases for attribution results, including raising awareness of climate risks, preparing adaptation measures and attributing climate loss and damage. Then we discuss the considerations necessary to fulfil these uses in three case studies, including a heatwave in the UK, a tropical storm in the Caribbean and a drought in East Africa. To conclude, we highlight the regions and hazards for which information can be drawn without new quantitative analysis, and those in which it remains urgent. This could aid prioritisation of limited resources for research into less well understood regions and hazards.

One Introduction

Anthropogenic interference in the climate system is increasingly manifesting as altered intensities and likelihoods of extreme weather events. The science of extreme event attribution is the toolbox used to understand and quantify these changes in intensity and likelihood of individual extreme events attributable to human-caused climate change. For any given observed event, multiple possible event definitions exist, each of which relates to slightly or significantly different aspects of the extreme; for example, a tropical cyclone could be characterised by wind speed or associated heavy rainfall. Estimates of the influence of anthropogenic climate change also differ depending on the event definition, including spatiotemporal factors, as illustrated in recent heatwave attribution and multiple studies of rainfall from Hurricane Harvey. Similarly, the framing of the attribution study and methodological decisions also play an important role in the overall estimate of the role of climate change.

After more than a decade of development of this science, consensus on various best practices are beginning to emerge. For example, using multiple definitions for a given event informs understanding of the dependence of the result on this choice, as well as providing utility to a wide range of prospective stakeholders. More broadly, the conditions required for ranking the quality of evidence provided by studies are now well established and standardised protocols are being laid out by the community. In addition, combining quantitative approaches referenced above and narrative or storyline approaches can address different stakeholder needs.

As the science is further refined, its reliability and credibility grow. Moreover, our ability to attribute different extreme meteorological events increases and the cumulative number of attributed events increases every year. However, there is limited capacity for producing attribution studies. Relatively few resources are available and attribution studies are still produced on an ad hoc basis, with an overwhelming focus on the global north. This is for a variety of reasons including historical data availability, climate model performance and the prevalence of local experts. Critically, attribution is not yet undertaken systematically in a single region of the world. This not only prevents the provision of valuable information in the wake of weather-related disasters but also hinders a comprehensive overview and understanding of climate change impacts, and thus loss and damage.

By their nature, event attribution studies are tailored to specific events. However, over the course of several years in a given location, multiple events can occur that share common characteristics. For many meteorological hazards, certain synoptic circumstances are effectively a necessary precondition for occurrence. Thus, we posit that attribution studies for specific events contain far more information, about other hazards of the same type, than is currently utilised.

Even in an ideal world with sufficient capacity to undertake many studies, it may be both inefficient and unnecessary to study afresh every event that occurs. The untapped information in past studies can be combined with improved recording of hazards and impacts, and ever-increasing knowledge of trends in extremes and large-scale detection and attribution studies. This has the potential to facilitate attribution scientists making expert judgements for very rapid, low-resource attribution statements.

Important considerations when deriving statements for new events based on past attribution studies include both how comparable different events are, and the intended use of the attribution results. First, not all events in the same region of the same type will have a comparable influence from human-caused climate change, at least not on a quantitative basis. For instance, extreme seasonal rainfall events in the UK have elicited a variety of statements: the wet summer of twenty twelve was not influenced significantly.

the wet winter of twenty thirteen/fourteen had a possible link but was ultimately inconclusive; autumn two thousand rainfall-driven floods in England and Wales and winter twenty fourteen floods in southern England were made more likely. This arises partly due to the seasonality itself, which plays a significant role in the link between many extremes and climate change, and more broadly the choice of event definition in each case.

Second, the uses for attribution results are becoming increasingly diverse. When an extreme event occurs, a variety of stakeholders raise questions that are answered using attribution science. For example, 'what was the role of climate change in this event?', and 'what are the implications for disaster risk reduction or for loss and damage from climate change?'.

In this paper, we examine when and under what circumstances some common questions can be answered rapidly based on past attribution studies and other information sources, such as projections. In particular, when can we use attribution literature to provide information to interested stakeholders? And how can we do so in a transparent manner that facilitates reproducibility?

To explore this challenge, we assess the statements that can be made for two region-specific case studies of heatwaves in the UK and tropical storms in Puerto Rico. We then comment more generally on where and for which hazards new studies offer value, or if drawing insights from existing studies allows for sufficiently reliable conclusions on the role of climate change. The conclusions of this study provide an assessment of where the attribution community could prioritise limited resources to add the greatest value.