A deadly heatwave over Uttar Pradesh recently claimed as many as 100 lives. There were subsequently several reports saying that according to a model called the “Climate Shift Index” (CSI), developed by a reputed U.S. nonprofit called Climate Central, this heatwave was made twice as likely by climate change.
What is the scientific confidence level in such local attribution of individual weather events to climate change? And what are the consequences of such claims?
How can climate change influence weather?
An exercise to determine climate change’s influence on a weather event involves two exercises: detection and attribution. A heatwave is defined based on the normal temperature of a region; ‘normal’ in turn is defined based on long-term historical data. The temperatures in Uttar Pradesh during the June 14-16 period met the definition of a heatwave. Put another way, a heatwave was definitely detected. Next, in terms of attribution, the CSI implies that the heatwave was made twice as likely due to global warming.
Several scientific and socioeconomic questions arise with such proclamations from trusted climate organisations.
Equally importantly, the experts who developed methods to rapidly compute the extent to which a weather event can be attributed to climate change have set out caveats and shortcomings – and these tend to get lost when the impact of climate change on a particular event is reported to the general public in a context-agnostic manner.
What are attribution models?
Scientifically speaking, an attribution exercise compares real conditions that have occurred with a so-called counterfactual world – a hypothetical world where climate change has not occurred.
Scientists create counterfactual worlds for these weather events using historical weather data and model simulations. The observations are constrained by limitations and the models are never accurate. Setting them aside, we must also take a fuller view of attributions and their associated claims.
According to Climate Central, its CSI is “grounded” in work described in a paper published in June 2022.
How accurate are the models?
Experts developed rapid attribution methods to help with policies, climate adaptation, and for health-impact studies. On the other hand, the outcomes of heatwaves and such extreme weather events are related to the vulnerability of the population exposed to the hazard, which attributions must account for – but they don’t.
Attributions also don’t account for the confluence of multiple natural weather conditions as well as human decisions that led to a heatwave being so deadly. (The most dire consequence of natural hazards often tends to be the product of too little attention being paid to early warnings that may already have been issued.)
Our historic analysis of temperatures allows us to say, with high confidence, that in the last few decades, heatwaves have been getting worse over many parts of India even as other parts of the country appear to be cooling. On the other hand, our confidence in the changes in extreme rainfall events is not as high. This is partly due to the smaller spatial scales at which rainfall events happen and their ability to change at shorter timescales.
Some of the low-confidence in historic changes is also related to a lack of reliable data with sufficient spatial and temporal coverage, even though India has some of the best rainfall data among the world’s countries. Poor data coverage in turn affects the counterfactual world built by combining the sparse data and imperfect models. Ultimately, This is how the inferred impact of climate change on a particular weather event can be erroneous.
In fact, we must accept that there is really no way to scientifically ensure the accuracy of such attributions.
What is natural variability?
In this context, we need to ask some key science questions. A rather unique set-up of events – including warming of the northern Indian Ocean from January onwards and the cyclones and typhoons during May and June – have disrupted the normal march of the southwest monsoon this year. Also playing out in the background is the world’s transition from a La Niña winter in 2022-2023 to the emerging El Niño summer of 2023.
These events also underscore the fact that natural variability – i.e. natural variations in the climate – always adds to or subtracts from the effects of climate change at the local level. For example, South India can have its hottest summer and in the same summer Chennai can have its coolest day in June.
Climate change also affects the natural variability itself. The number and intensities of tropical cyclones as well as the El Niños and the La Niñas are also likely being affected by climate change. But the models do not agree on some of these estimates; the models used for attributions don’t even simulate cyclones!
What does this mean for the Uttar Pradesh heatwave?
The attribution approach that the CSI has taken does not consider such local weather systems. Studies have found that even irrigation can affect heatwaves, but neither the attribution data nor the models in the Uttar Pradesh case represent such effects.
This brings us to the socioeconomic and sociopolitical implications of claims that climate change made the heatwave X-times more likely. What is the longer-term context? Should farmers worry about what it means for the rest of the agricultural season? Should people start moving? Should businesses and investors begin to reconsider their plans in the State?
It is naïve to assume that limited indices – which have their purpose in a specific context, in a supplementary capacity – will only impel climate adaptation, to deal with heatwaves, and not have other off-target consequences.
So, we desperately need a 360-degree view of such claims, especially in light of their potential deficiencies. Event-by-event attribution on a daily timescale is neither possible with sufficient accuracy nor is it practically valuable. It can also divert resources away from other, more worthy efforts, such as improving early-warning systems.
Raghu Murtugudde is a visiting professor at IIT Bombay and an emeritus professor at the University of Maryland.
