Forecasting the Disaster Is Not Preparing for It

Extreme weather is often described as a forecasting problem because forecasting is the part that arrives with maps, colors, and a gratifying sense that somebody has pressed a button. But the harder problem is administrative: deciding beforehand which homes, schools, power systems, and communities get protected, restored, or left to improvise after the water rises or the heat lingers.

An MIT climate-extremes project brings that gap into focus. More than 40 researchers have produced 29 papers alongside tools and datasets meant to translate hazards—flooding, cyclones, humid heat waves, and more—into operational choices. Its planners have worked with communities in Boston and Broward County, Florida, including on web tools that can ask an unnervingly practical question: which schools are likely to stay driest under different flood scenarios?

That is resilience in its least cinematic form: not heroics after catastrophe, but boring decisions made before it. The project’s researchers note that municipalities commonly maintain emergency-response plans, while far fewer prepare for recovery before an event, including the question of housing. America is relatively good at declaring an emergency. It remains less practiced at arranging life after one.

Risk is not evenly legible. Kerry Emanuel, one of the project’s co-leads, says of hurricane risk modeling, “For hurricanes, we’re pretty much there.” The confidence has consequences: his methods are used by First Street to inform local governments, insurers, developers, and real-estate platforms about environmental risk at private properties across the United States.

But severe convective storms—the family containing thunderstorms and tornadoes—remain a more stubborn scientific and civic problem. Emanuel’s assessment is blunt: “we’re not close to being there” on reproducing their statistics from coarse weather data in the way researchers can for hurricanes. He says these storms have caused more deaths and damage than hurricanes over the past decade, though the article does not provide the underlying figures.

The policy implication is not that communities should wait for perfect models, an approach that would make Nero look proactive. It is that uncertainty itself should shape preparedness: sturdier warnings, safer public buildings, recovery plans that do not assume a neat disaster boundary, and aid systems that can function when the hazard has not behaved according to the brochure. A forecast is useful precisely when it changes a decision; accuracy that cannot be converted into shelter, power, or a place to live is merely very sophisticated weather trivia.

Heat, floods, and storms do not politely remain in the environmental column of a spreadsheet. They turn up in electricity demand, equipment placement, insurance pricing, public-health exposure, and the ability of households to endure a long, hot afternoon when the grid is strained. Climate resilience is therefore not one project but a coordination problem wearing several hard hats.

MIT researchers are examining how an electrical system already being remade by renewables and storage should be planned under climate-driven extremes. In one Texas model, new wind resources were sited along the Gulf Coast because their daily wind patterns could complement wind in northwest Texas and solar generation. The finding is a model result, not a construction announcement, but it captures the governing logic: diversity of supply and geography can be a form of protection.

The researchers are appropriately cautious about where this work stands. As Michael Howland puts it, “We want to continuously push on model realism and accuracy to eventually make it more of a practical and useful tool for grid planners.” Eventually is doing honest work there. The country does not need another dashboard that can diagnose vulnerability with exquisite precision while utilities, landlords, and local governments debate whose budget contains the verb. The useful climate tool is the one that gets converted into a decision before the forecast becomes an obituary for normal life.