The possibility of spending big money on the wrong things

Why it matters that US precipitation data is out of date

Last night, President Biden happily reminded the crowd that his administration is spending a lot on infrastructure:

And thanks to our [$1.2 trillion] Bipartisan Infrastructure Law, 46,000 new projects have been announced all across your communities. And by the way, I notice, some of you [who] strongly voted against it are there cheering on that money coming on. I’m with you. I’m with you. If any of you don’t want that money in your district, just let me know. [We'll] modernize our roads and bridges, ports and airports, public transit systems.

This should be good news for everyone. Federal infrastructure investment flowing through state and local governments has climbed sharply over the last two years, and McKinsey says that extensive federal planning efforts carried out in 2023 will be reflected in skyrocketing infrastructure expenditures in 2024.

Here's the White House map of public spending right now. It’s a thicket of green “formula” (predictable grants to states based on Congressional allocations) and blue “discretionary” (competitive) investments:

Surely everyone has something to look forward to.

Public works like roads and airports are crucial not just because they are essential elements of the country's economy but also because they provide visible evidence that government exists and is doing its job. It's essential that these projects get built.

It's also essential that they are designed to withstand the rapidly accelerating climate effects that scientists know are going to happen. Otherwise, their quick failures will further amplify and entrench existing pessimism about the role of government in American life—as well as endanger lives. Trust in government is low these days, with 73% of Americans saying they think the country is going in the wrong direction.

That's why I'm worried that the current standards broadly used by just about everyone involved in designing, building, and operating water-related infrastructure in America still assume a stable climate. 

Remember the prolonged heavy rainfall in Vermont in July 2023?

Among a host of other debacles, it caused several wastewater treatment plants to fail. That meant that untreated wastewater that was supposed to be handled by combined sewer systems—including everything that goes down the drains of homes and businesses—flowed right into Lake Champlain.

How about the extreme rainfall and flash floods in Middle Tennessee in 2021? Dozens of roads, including major highways, were flooded.

Or Hurricane Harvey in August 2017? That storm just sat dumping rain on Houston, and Harris County's drainage infrastructure collapsed.

We know that storms are intensifying and we know they're not moving along the way they used to: Of the 28 billion-dollar disasters that struck the US last year, 17 included major storms. We keep hearing that existing drainage systems, bridges, and other structures may not have been designed to handle the intensity of the onslaught.

What most people don't know is that the precipitation frequency estimates used by engineers, planners, and government agencies for roads and stormwater drains and everything else, called Atlas 14, are still out of date because they assume "stationarity"—they use the past to predict the future. When it comes to planning infrastructure, we are blind to what science tells us is about to happen.

There's a strange history behind this situation when it comes to funding and methodology. 

Funding: Until 2021, dedicated, reliable funding for the Office of Water Prediction, the part of NOAA that generated this central set of predictions, was uncertain. When NOAA or other federal sources supported particular work, often with support from a particular state that wanted an expedited update, OWP would do it. This resulted in separate "volumes" of predictions based on state boundaries being completed independently from one another and often years apart. It meant that there were weird differences drawn along state lines because different methodologies were chosen for different updates. It also meant that some states weren't updated at all: as of 2020, NOAA precipitation frequency estimates for the five Northwestern states were more than 50 years old, and estimates for 19 states hadn't been updated in 20 years. (NOAA put out a video about all this in February 2020.) The Bipartisan Infrastructure Law changed this awful situation, directing direct federal funding to the Office of Water Prediction for the first time. The new update, Atlas 15, will cover the whole country. 

Methodology: The Atlas 14 approach that is still in effect was developed in the 1990s. One of its assumptions was that extreme precipitation characteristics are stationary. That assumption is, we now know, bananas.

This is why people say "Why are they calling these storms that happen all the time '500-year storms?'"

The OWP has done its best, based on historical data, to estimate how likely it is that a storm exceeding a given amount of precipitation for a specified length of time (five minutes, an hour) will happen in a given location in a given year. A "500-year storm" means there's a 0.2% chance in any given year that this will happen. That's a very low chance.

The trouble is that things are speeding up, and what the OWP labels as a 500-year storm is probably already a 100-year storm—much more likely. The probability differences could be enormous:  A June 2023 piece in Scientific American by Thomas Frank reported that the First Street Foundation was asserting that the likelihood of a 100-year (1% likelihood) storm predicted by Atlas 14 for Baltimore was actually likely to occur every 14 years (more like a 7% chance).

These are all estimates, but it's worrisome that stationarity is still in place.

This situation is being fixed. Slowly. NOAA knows that Atlas 14 needs to be updated to account for climate change, and Congress has directed federal funding for this task. NOAA predicts that Atlas 15 will be complete by 2027. 

Look, this process is difficult and needs to be done carefully. In the meantime, though, we'll be building a lot of infrastructure that will be at risk of being outmoded in short order. I've heard different things about Singapore's much-hyped "digital twin" work to create a detailed fine-grained planning model of micro-climates, so I'm not pointing to it as an aspiration for the US. But surely we could do a better job at understanding, predicting, and publicizing what's about to happen. 

It's likely a question of money. If we applied the funding to this standards process (and other climate data work) that it deserved, everyone involved would have a far clearer sense of the risks we face and could plan—and spend—accordingly.

That’s one way to bolster trust in government.