It’s been getting rainier across much of the United States over the past few decades, largely due to the planet’s warming. These are not just more frequent pleasant showers. Instead, increasingly intense storms release a deluge of water that cost the U.S. billions of dollars in flood damage each year. Converting increased rainfall caused by climate change to costs from flood damage is a tricky task. But it’s one that scientists have recently completed. The researchers found that around a third of the $200 billion in flood damages over the past thirty years is due to climate change’s effect on precipitation. The study, led by doctoral student Frances Davenport and Professor Noah Diffenbaugh at Stanford University, was published this January in Proceedings of the National Academy of Sciences.
To tackle the cost of flooding induced by climate change, the scientists first dug up and analyzed records of intense rainstorms and flood damage for each state and each month since 1988. The authors then checked their monthly trends with 24 climate simulations that estimate how rainfall has transformed in the past thirty years. To separate humanity’s impact from natural climate variability, they compared two versions of their simulations. The first version is based on recent CO2 trends, and the second assumes much lower CO2 levels associated with the late 1800s, when fossil fuel usage was still in its infancy. Comparing these rainstorm simulations and applying that comparison to recent rainstorm trends and flood damage is the final step to determining how much climate change has worsened flood damage across the U.S.
Their approach takes a more detailed stab at the question of climate change and flooding than previous studies, which have either only looked at annual trends or only looked at the United States as a whole. Since the U.S. isn’t experiencing uniformly heavier rainfall across the entire country or over the entire year, results from less detailed assessments are much less conclusive. By evaluating these more detailed trends, they determined that the increase in rainstorms since 1988 can at least be partly attributed to Earth’s warming. Taking it one step further, they tie climate change to about $73 billion—or 36% of the total—U.S. flood damage during that period.
Higher air temperatures lead to more intense precipitation in many areas because warmer air can retain more water. When clouds finally do swell to the point of raining down on a warmer planet, there is therefore much more water. Higher temperatures also increase evaporation, often leaving soils drier and harder. This further encourages flooding because the hard ground is less willing to absorb water. (Paradoxically, this also leads to increased drought in many areas of the U.S.). Global temperatures between 1988 and 2017 have risen around half a degree Celsius or almost one degree Fahrenheit. As the planet continues warming, floods and the immense damage they cause can only be expected to increase.
Frances Devnport is a PhD student at Stanford University, working under Climate Science Professor Noah Diffenbaugh.
Managing Correspondent: Jordan Wilkerson
Original Science Article: Contribution of historical precipitation change to US flood damages
Image Credit: flickr
Hey, Nick
Below, I’ve responded to your main points. Beyond my response, I’m also providing a link to a report published by the Intergovernmental Panel on Climate Change in 2013. The report discusses the physics of climate change, and I believe it includes discussion of what goes into climate models (they’ll publish an updated report this year if you’d rather just wait for that).
Here’s that link: https://www.ipcc.ch/report/ar5/wg1/
#1 The flooding and precipitation data does tell you, in detail, of the rising trends over the past 30 years. We know that the rise is generally due to climate change. It’s one thing to say something vague like that, but it’s another to ask how much exactly is climate change causing. That’s the reason they coupled this data with climate simulations: to put numbers and error bars on the extent to which the change in extreme precipitation and flood damage in each state can be tied to climate change. The 24 simulations are all used to simulate the same 30 years. Perhaps, it’s better to call them 24 independent simulations. These are all developed by different climate scientists across the world. The benefit is that we essentially take 24 independent understandings of how the climate works and see how well they all agree. So it’s not quite just that they use them all and take the mean. They also check variability and extent of agreeability between the simulations.
#2 I think I misunderstood your original comment then, as it sounded like you were saying that the air doesn’t necessarily take in more moisture (to which I explained how we know, on average, it does). Yes, there are many factors that can affect relative humidity. The most fundamental one is air temperature. The reason it’s called relative humidity is because that number is reporting the current humidity relative to the maximum amount of water vapor the air can hold at that given temperature. As it gets hotter, the relative humidity actually decreases for the same amount of water vapor because the air can hold more water. I think you’re describing how humid it feels to you at certain temperatures across the country, but that’s not the point. The key is not that it feels humid to you. The key is that clouds eventually saturate and rain down. That threshold for when they saturate is higher in a warming climate. But more water vapor entering into the atmosphere in a warming planet helps clouds meet that heightened threshold. Now once they start precipitating, there’s more water being unleashed onto the surface miles below. So you’re right; the air is not always saturated with water. The point is that when it finally exceeds that saturation (i.e. rain clouds), the rainfall tends is likely to be much more intense. I want to emphasize that I’m speaking on average effects here. Regional effects of climate change have heightened probability of extreme events like drought and floods; no specific extreme event can be confidently connected to climate change. The increase in extreme rainfall events and flooding is the connection in this paper (it’s why they needed 30 years of information through data and simulations to put numbers on this).
#3 They did this by examining the insurance claims associated with each flood. Since the baseline level of flooding also have increased cost over time from higher population and infrastructure, those factors don’t really relate to a % contribution of damage from climate change. I only made that clarification because that’s the novelty of the paper. They push forward a strategy that can help put dollar amounts on the effects of increased extreme precipitation from climate change. This helps more concretely connect the science we’ve learned to the policy conversations. For example, states could use this approach to better constrain their future budget allocations for natural disasters.
Finally, I want to express gratitude for your reading my story and engaging with it. Please continue reading my stories and others at SITN. Thank you!
Sincerely,
Jordan
Was their simulation of the past 30 years accurate? They had the actual records of flooding for the past 3 decades, so how did the simulation compare to what really happened? That would be very interesting and I’d think very valuable to know that their model was able to accurately predict the increased flooding with an acceptable margin of error.
Also, in their late 1800’s simulation how did they estimate the amount of plant life and evapotranspiration?
Warmer air is capable of containing more water, but of course that doesn’t mean it will. I’m also somewhat amazed that 1 degree Fahrenheit is being attributed to 1/3 of the floods. Since is there is no actual evidence that says, for example, floods are more frequent in 88 degree weather than 85 degree weather.
Why is it so difficult to produce a climate change report without making assumptions that favor the desired outcome? Why not create a climate model that when the last 20 years of CO2 and other atmospheric gasses are input, it is able to simulate, with a great degree of accuracy, some material change that we can match with reality. (no looking at the data before building your model please, we don’t need a model that’s gamed so it only works for the last 20 years). Then we might have something useful that one might have reason to have some confidence in. If you can’t xo that, then you obviously either don’t have, or don’t know all the data and data types required to produce a working model – in which case, less talking, more learning please. The sheeple out here are having panic attacks that their 2 year olds will never make it to prom – and that’s on you.
Hey, Nick:
Thanks for your comment. The reason they focus on the last 3 decades is unrelated to their models’ capabilities. It’s because that’s when reporting on flood damages was most consistent; the authors looked at stats from the National Climatic Data Center and the National Flood Insurance Program to look for consistent reporting between the two (methods of reporting change over time, and you don’t want to be dealing with inconsistently collected data).
Their models simulate precipitation changes, which includes probability increases of extreme precipitation, in each state. They’re comparing changes in observed extreme precipitation trends to the average response from climate forcing. They are not simulating floods or flood insurance claims… However, floods are almost entirely the result of extreme precipitation events (coastal flooding isn’t covered in this study).
“Warmer air is capable of containing more water, but of course that doesn’t mean it will. I’m also somewhat amazed that 1 degree Fahrenheit is being attributed to 1/3 of the floods.”
A couple points in response to this. First, extreme precipitation events have increased in the past few decades, as have floods. Atmospheric water vapor levels are indeed increasing as well; academic institutions and the US government monitor all of this regularly. Warmer air is both capable of containing more water and it has been doing so as the planet gets warmer. Second, 1 degree Fahrenheit is the average temperature increase. The characteristic of recent climate change is that both the average global temperature increases, and the variability in temperature increases. This second point helps explain why extreme weather events are increasing, including increased rainstorms that can cause flooding. The final point is minor: the authors attribute extreme precipitation from climate change to 1/3 of flood damages, not 1/3 floods.
Jordan, thanks for your response and for taking the time to explain some things. There are still elements that aren’t making sense to me – but it is very possible I am prescribing a layman’s meaning to sentences and that’s causing my confusion. If you don’t mind bearing with me and helping to clarify a bit more I’d really appreciate it as it’s not often someone is willing to help clarify rather than just assume I’m only looking to shoot something down (just the opposite, I’m trying to understand, and if I can then I’d only want to support the Truth). So here is what I’m not understanding:
1. “the scientists first dug up and analyzed records of intense rainstorms and flood damage for each state and each month since 1988. The authors then checked their monthly trends with 24 climate simulations that estimate how rainfall has transformed in the past thirty years.”
If the authors analyzed monthly records of intense rainstorms going back to 1988 (for specific states), I would assume that data alone would tell them how rainfall has transformed over the past 30 years. At that point I don’t understand what the 24 simulations are for. Are the simulations based on a model of less Carbon Dioxide in the atmosphere, so that they are intending to show what the rainfall would have been if not for the atmospheric changes? If so, I can understand that, but before accepting their model/simulation as valid I would expect they’d need some kind of control to prove its accuracy – for example simulating the rainfall for the past 30 years with current atmospheric gasses. If their simulation isn’t accurate when fed all the known data points of the last 30 years , then their model is overlooking some relevant inputs and their simulations would not be useable right? I’m just confused because this was an opportunity to validate their model and its simulations, but I don’t see the data around that? Also do you know why they performed 24 simulations? Was it 1 simulation for every 15 months? Was each simulation covering 30 years and they took the mean? I understand there could be unknown variables that require multiple simulations, but even taking a mean doesn’t mean much if they haven’t shown the simulation can come close to mirroring the facts they spent all that time gathering. Or what am I missing?
#2 I am not arguing that extreme precipitation events haven’t increased – I’m on board that they have. I also don’t have any reason to disagree that there has been increased water vapor levels. The thing is, I don’t see where the data shows that the extreme precipitation events occurred in synch with the times of increased water vapor. I also don’t see where there data shows that the increased temperature resulted in increased water vapor (let’s say there are variables and sometimes things aren’t instant, so do we even have the data that shows water vapor levels rose higher within 3 weeks of the hottest days – and that the extreme rainfalls happened within 3 weeks of that?) Florida is generally much more humid when it’s 90 degrees than when it’s 74 – but it’s also generally more humid when it’s 74 than when New York is 75 (and definitely when Los Angeles is 75). I’m being a bit silly with the Los Angeles part – but my point is there are a lot of factors that go into relative humidity such as the % of land covered by plant life as well as things like wind speed. So I would be curious to see the data that showed the correlation from a timing perspective between the hotter day > increased water vapor > resulting extreme rainfall. Again, I’m not doubting any of these things – I’m just trying to solve a problem – and if the data only shows increased water vapor at a point 3 months before the extreme rainfall (assuming the water vapor had gone back to normal levels by then) – then I feel like we are oversimplifying..
#3 I was confused by the clarification at the end of your explanation. The authors attributed the actual damage from the floods to climate change instead of the floods themselves? How did they do that? Did they account for increased population in an area and then determine how that would affect the wear to the roads, possibly causing them to shift in varying directions by degrees which would enable additional flooding (not to mention the additional people and their belongings that would be damaged). Did they account for the gentrification of areas or increased wages that resulted in more expensive items being destroyed? I just would have thought it would be much easier to estimate the percentage of additional floods than it would be to estimate the additional damage caused by those floods – simply because the variables are so much greater for the latter. Anyway, I really don’t know, and I don’t mean to sound argumentative if that’s how I’m coming off. I might just misunderstand some core principles that are making me confuse the whole thing. It just feels like every opportunity to be methodical and tie the data to reality was not done, – but maybe it was and I just don’t understand.
Thanks again for the dialogue and for letting me drill in even though it could just be from my own ignorance.
Nick