by Durga Velayudham
How much certainty does science owe the public when lives are already lost? When models disagree with observations, who carries the burden of doubt? And what happens when climate change appears to be everywhere in the data but nowhere definitive enough for attribution frameworks to satisfy their own standards?
These questions sit at the uneasy heart of the World Weather Attribution scientific report, published last week, titled Increasing heavy rainfall and extreme flood heights in a warming climate threaten densely populated regions across Sri Lanka and the Malacca Strait, authored by Sarah Kew and an international team of climate scientists working at the intersection of meteorology, hydrology, and disaster risk analysis. The authors come from institutions embedded in the applied science of climate extremes, not advocacy, and that professional posture shapes the document’s restrained but deeply unsettling conclusions. This is not a report that shouts. It hesitates, qualifies, and caveats its way through some of the most devastating floods South and Southeast Asia have seen in decades. Yet precisely in those hesitations lie its most controversial implications.
The report opens from a position that is both factual and quietly accusatory: “Globally, there has been an increase in extreme precipitation events because of increased global temperatures.” That sentence, almost banal in its familiarity, sets the tone for what follows. The floods that struck Sri Lanka and the Malacca Strait region in late 2025 were not random aberrations. They occurred in landscapes already shaped for disaster: steep central highlands draining into low-lying floodplains in Sri Lanka, and an archipelago of volcanic islands, deltas, and shallow drainage basins in Indonesia. As the authors note, “intense rainfall quickly channelled runoff from the hills into densely populated floodplains,” while in Indonesia “heavy rainfall easily triggers both flash floods and landslides.” Geography, in other words, was always complicit.
But geography alone does not explain scale. Cyclonic Storm Ditwah over Sri Lanka and Cyclonic Storm Senyar over the Malacca Strait produced rainfall events that the authors classify as rare even by historical standards. Ditwah’s rainfall corresponds to “roughly a 1-in-30-year event in today’s climate,” while Senyar’s aligns with “roughly a 1-in-70-year event.” These return periods, already alarming, mask deeper uncertainty. The report repeatedly stresses that “local return periods can be much higher,” a reminder that averages flatten extremes in precisely the places where people live.
The most arresting claims in the report emerge when observational data are compared against theoretical expectations. Observations show that heavy five-day rainfall events over Sri Lanka are now “about 28% to 160% more intense due to the warming to date.” Over the Malacca Strait, increases range from “about 9% to 50%.” These figures are not marginal deviations. They exceed what basic thermodynamics would predict. The authors themselves underline the anomaly: the observed increases are “between 3 and 7 times more than the 10% increase due to 1.3°C of warming until now expected from the Clausius-Clapeyron relationship.”
This is where the report becomes genuinely controversial. Climate models, the very tools used to attribute events to anthropogenic warming, do not reproduce these trends. In blunt terms, “the large trend that is visible in the observations in both study regions is not captured in the models.” The language is unusually direct for a scientific report. Models “do not show consistent results regarding the direction of change,” and there are “probably aspects of the atmospheric circulation that are systematically misrepresented by the models.” The admission is striking not because scientists are unaware of model limitations, but because attribution science relies on those models to speak with authority.
The consequence of this mismatch is profound. Despite strong observational evidence that rainfall extremes are intensifying, the authors conclude that they “cannot draw an overarching attribution conclusion that quantifies the influence of climate change on these events.” This is not denial. It is epistemic restraint. And yet the report immediately follows with a sentence that undercuts any comfortable ambiguity: “Nonetheless, we conclude that climate change is at least one driver of the observed trend given that the other major known drivers of heavy rainfall are accounted for and cannot explain the trend.” Climate change is both unquantifiable and unavoidable.
Natural variability is examined carefully, perhaps defensively. ENSO and the Indian Ocean Dipole are acknowledged contributors. In the Malacca Strait, La Niña and a negative IOD “contributed about 5% to 13% to observed rainfall.” But these influences are explicitly insufficient. They explain year-to-year variability, not long-term escalation. The report emphasizes that even when ENSO and IOD are included as covariates, “they do not discernibly affect the long-term trend driven by GMST.” Warming remains the background condition against which everything else plays out.
The storms themselves defied historical intuition. Both Ditwah and Senyar formed unusually close to the equator, where cyclonic rotation is typically weak. Their existence required a convergence of atmospheric anomalies: “an exceptionally active Intertropical Convergence Zone,” enhanced by equatorial Kelvin waves, warm sea surface temperatures, and deep tropospheric moisture. Once formed, they moved slowly, looping and stalling in ways that maximized rainfall. The authors note that these slow-moving tracks “significantly amplified rainfall duration and flood potential.” What emerges is a portrait of new risk pathways, where low-latitude cyclones become agents of extreme flooding rather than wind destruction.
The human consequences are detailed with sobering precision. In Sri Lanka, Ditwah “became the deadliest weather-related disaster since the 2004 tsunami.” Over 1.4 million people were affected, more than 233,000 displaced, and entire districts remained submerged or isolated at the time of writing. The cyclone struck regions responsible for “more than 80% of the country’s GDP,” inflicting losses estimated between Rs. 210 billion and Rs. 320 billion. In Indonesia, Senyar delivered “over a month’s rainfall in a day” to parts of Sumatra, affecting more than 3.3 million people and causing hundreds of deaths. The report is careful not to sensationalize, but the accumulation of figures speaks louder than rhetoric.
What makes the analysis especially unsettling is its insistence that physical hazard alone does not determine disaster. “The devastation…is indeed not a result of only the physical hazard, but also the exposure of people and assets to that hazard, and their vulnerability.” Floodplains filled with informal settlements, deforestation that stripped watersheds of buffering capacity, wetlands converted into urban land, and insurance systems that barely function all compounded the impact. In Sri Lanka, forest cover declined from about 90% in 1900 to less than 20% by 2002. Urban areas around Colombo expanded by roughly 51% between 1997 and 2015, erasing marshlands that once absorbed floodwaters.
Institutional responses were not absent, but they were uneven. Both countries issued warnings ahead of the storms. Sri Lanka’s multi-hazard early warning system “contributed to sustained monitoring and regular public advisories.” Indonesia’s meteorological agency provided continuous updates. Yet the report documents how “communication barriers,” language gaps, and damaged telecommunications infrastructure meant that some communities “received warnings late or not at all.” In Indonesia, alerts “came through different agencies and channels,” creating confusion. Forecast uncertainty and limited impact-based models hindered anticipatory action. Preparedness existed, but it was not equal to the speed or scale of the hazard.
The authors are particularly careful to situate their conclusions within methodological limits. “This analysis is conducted in the immediate aftermath of the disasters,” they note, and many impacts “will only fully materialize months to years after the event.” This humility does not weaken the report; it strengthens it. The refusal to overclaim stands in contrast to the magnitude of what is already visible. Entire health systems disrupted, increased risks of waterborne disease, long-term livelihood losses for farming communities, and displacement patterns that echo previous disasters rather than resolving them.
Perhaps the most consequential implication of the report lies in its challenge to the current architecture of climate attribution. Observations show dramatic increases in extreme rainfall. Physical understanding supports the plausibility of such increases in a warmer atmosphere. Yet models, especially over small islands and complex coastal regions, fail to reproduce the trends. The result is a scientific limbo where impacts are undeniable but attribution remains cautious. The authors acknowledge this tension openly: “Because models do not capture the observed trends we cannot combine these into a synthesised statement.”
This gap matters. Attribution science increasingly informs litigation, adaptation funding, and international climate negotiations. If the places experiencing the most severe impacts are also the places where models perform worst, then the global system risks systematically under-recognizing loss and damage. The report does not make this political argument explicitly, but its data-driven restraint gestures toward it. Science, bound by its own standards, may be lagging the lived reality of climate risk.
In its conclusion, the report returns to balance, almost deliberately. The floods were shaped by “climate change, natural variability, rising sea surface temperatures and vulnerability.” No single driver is allowed to dominate. And yet the final synthesis is unmistakable: “Observations across both regions show clear upward trends in the heaviest 5-day rainfall totals consistent with a warming climate.” The phrase “consistent with” does a great deal of work here, bridging evidence and caution.
What the report ultimately offers is not certainty but clarity of risk. It shows that extreme rainfall is intensifying faster than expected, that existing models are insufficiently equipped to capture this acceleration, and that vulnerability turns meteorological events into humanitarian crises. In doing so, it forces a reckoning not just with climate change, but with how science communicates danger when proof lags impact. The floods of Sri Lanka and the Malacca Strait are not waiting for attribution consensus. They are already rewriting what extreme weather looks like in a warming world.
