Written by Lainey Ward
Indices are a standard tool for monitoring and reporting weather and climate extremes. An index applies a fixed formula to a long series of weather or climate data and reduces it to a single value that measures one property of an extreme, for example its frequency, intensity or duration. A widely used example is “consecutive dry days”, the longest unbroken run of days with under 1 mm of rain. Over the past thirty years the World Meteorological Organization has relied on a core set of 27 such indices, based on daily temperature and rainfall, and draws on them in its annual climate reporting. They have endured partly because they are simple to compute, transferable across very different climates, and reproducible.
Many high-impact events, however, emerge from interactions between several variables, none of which need be extreme on its own, and those interactions can amplify the impact or, at times, offset it. Events of this kind are called compound events, and a single-variable view tends to miss them and to understate their risk. Indices for compound events exist, but there is no widely agreed standard, unlike the original 27. My scientific mission set out to understand why the single-variable indices were so widely adopted, to survey what exists for compound events, and to see whether any multivariate indices share the same qualities.
In the first week I built the code that queried the literature databases for the systematic review. I set explicit criteria for including and excluding papers, and refined them against the literature to see what was feasible. I used agentic AI to apply these criteria across the literature, in line with current best practice for transparency.
In the second week I focused on analysing the papers that passed screening. I developed a set of categories to describe each index. Some covered simple metadata, such as an index’s name and the variables it combines. Others were more interpretive, such as the type of hazard it targets and how it is constructed. Robert Dunn at the UK Met Office was a great help in understanding the practical implications, including which observational data is available operationally and which compound indices could realistically be run. I then produced a condensed catalogue for the UK Met Office’s effort on compound indices for climate monitoring.
The review covered over a thousand papers and several hundred compound indices, far more than either of us had anticipated. By categorising them, some of the field’s conventions and gaps became clearer. I then asked which of these have the qualities that made the original 27 succeed. Only a few compound events have a single, widely used index. For others, such as heat and drought, several competing indices exist for the same event, while some have almost none.
The review also confirmed something I had suspected from my own PhD work, that compound indices are rarely studied with weather forecasts. My PhD addresses this through the detection and prediction of drought-to-flood events, one of the lesser-covered compound events in this review.
The catalogue feeds a wider UK Met Office and WMO effort to scope candidate compound indices as it refreshes the original 27 single-variable set, and to bring such indices into the annual State of the Global Climate report.
I had a brilliant experience and am grateful to Chris White for hosting me, to Robert Dunn at the Met Office for his involvement, and to the ANTICIPATE COST Action for its support. I highly recommend a Short-Term Scientific Mission to other early-career scientists like me!