Extreme precipitation events

This post is partly motivated by something I think I either heard Michael Shellenberger say, or write, but I can’t find it anymore. I have tried reading some of the articles again, and listening to some of the podcasts again, including the Heartland Institute one – where Micheal Shellenberger thanks them for what they’re doing – and the one with Alex Epstein, where Michael Shellenberger suggests that one of the Chapters in the book was motivated by some of what Alex Epstein promotes. If you’re not sure why I’m highlighting this, it might be worth looking up the Heartland Institute and Alex Epstein.

My memory is that Shellenberger was suggesting that precipitation changes would be modest, so it would be relatively straighforward to develop adaptation strategies to cope with these changes. I may have remembered incorrectly, but it is still a topic worth highlighting.

Depending on one’s definition of modest, it is probably true that the overall change in precipitation will be modest; the change in mean precipitation is estimated to be around 2% per K. The problem, though, as this paper points out, is that

the intensity of extreme precipitation increases more strongly with global mean surface temperature than mean precipitation

and that

[g]lobally, the observed intensity in daily heavy precipitation events, i.e. the rainfall per unit time, increases with surface temperature at a rate similar to that of vapour pressure (6–7% per K).

In other words, even though mean precipitation will only increase by about 2% per K, the intensity of extreme precipitation events will increase by more than this, largely (as I understand it) because the vapour pressure increases by 6-7% per K.

Credit: Myhre et al. (2019)

The paper then goes on to point out that it’s not just the intensity of extreme events that changes, but also their frequency. For example, the figure on the right suggests that the most extreme precipitation events could double with every degree of warming, while their intensity also increases by more than 10% per degree of warming.

So, the changes in the frequency and intensity of the most extreme precipitation events are certainly not small. What’s more, this also means that – as we warm – we’ll shift the distribution of precipitation towards more extreme events; i.e., more and more precipitation will occur in what we, today, would regard as an extreme event.

Given that the events that impact us the most are the extreme events, it seems a bit overly optimistic to think that we can easily deal with quite large changes in both the intensity and frequency of such events. As the paper itself even says:

Such large increases are not taken into account by adaptation management, and our findings imply that society may not be adequately prepared for the coming changes in extreme rainfall.

We will, of course, have to adapt to some of the changes we’re facing. However, there is a difference between recognising that some amount of adaptation is unavoidable, and suggesting that adaptation will be sufficient to effectively cope with any possible change we might experience. At the end of the day, it’s going to be a combination of mitigation, adaptation and suffering, and

we’re going to do some of each. The question is what the mix is going to be. The more mitigation we do, the less adaptation will be required and the less suffering there will be.

Frequency of extreme precipitation increases extensively with event rareness under global warming, paper by Myhre et al. (2019).

This entry was posted in Climate change, ClimateBall, Environmental change, Global warming and tagged , , , , , . Bookmark the permalink.

36 Responses to Extreme precipitation events

  1. Ben McMillan says:

    Also worth pointing out that the local changes in local precipitation are much larger than the global averages because the local changes mostly cancel out, and in a large fraction of places rain halves or doubles in typical baseline (unmitigated) simulations.

    e.g. the problem is not ‘worse weather’ but dramatically different weather that local infrastructure/ecosystems can’t cope with. E.g. agriculture depends on monsoon rains that have now shifted 500km north…

  2. As usual, there’s discussion about how AGW may or may not impact extremes while it’s obvious that climate scientists still don’t understand natural variability in regards to the major climate indices. From a week ago, here are Michael Mann and Gavin Schmidt having a clear difference of opinion about whether behaviors such as AMO and PDO are even oscillations

    AFAICT, Schmidt did not take the bait from Mann.

    This lack of understanding is highlighted in a recent paper by Kerry Emanuel “The Relevance of Theory for Contemporary Research in Atmospheres, Oceans, and Climate” AGU Advances 1, (2020), which is in the inaugural issue of AGU Advances https://agupubs.onlinelibrary.wiley.com/toc/2576604x/current

  3. Actually, I found where Michael Shellenberger mentions precipitation and flooding. It’s in the Heartland Institute podcast starting at about 31 minutes (if you wanted to listen). He says

    the precipitation increases that you’re talking about are just like nothing compared to the difference between having adequate flood control systems, or not. So, the issue is having good flood control systems.

    Of course, having good flood control systems is important, but substantial changes in extreme precipitation (both intensity and frequency) that are possible in the coming decades is not – as far as I can tell – going to be simple to deal with. Similarly, there are some regions that become increasingly difficult to protect against flooding.

  4. angech says:

    “Of course, having good flood control systems is important,”
    ATTP, I live in a flood plain. 30 year cycle, just above the 100 year level.
    Darn close in 1991 and 2012 or there about. My nippers were drawing chalk marks on the footpath as it covered the road and came up the driveway to the slab.
    Old records indicate at least 6 this high or higher in the last 100 years so that 100 year level looks a little sus.
    Rainfall usually 18 inches a year. Darwin where I grew up was 60-80 inches a year but no real flooding risk.
    Thanks for your comments.

  5. verytallguy says:

    “So, the issue is having good flood control systems”

    One of the reasons why adaptation makes less sense than mitigation is that we don’t know what to adapt to.

    Regional climate change is what needs to be adapted to, and, as I understand it, our ability to model and predict regional climate is really very poor.

    So where does Michael propose we build these “flood control systems”, and to what specification?

  6. vtg,
    Indeed. I’m certainly no expert at this, but my understanding of the situation in the UK is that it’s also extremely challenging, even if you have some idea of what to expect. If you try to protect one region from flooding, you end up moving the problem somewhere else. We already have regions that are prone to flooding. How will they cope if you increase the intensity of the extreme precipitation events by ~10% and have these events occur twice as often?

  7. Willard says:

    Perhaps this:

    MS: I think humility is the most important thing. I didn’t even include it in the book because this is one of many details, but for a long time, they thought that more warming would increase rainfall in the Amazon. Now, everybody agrees that increased warming is going to reduce rainfall in Amazon. That’s kind of important. And then, does it really change what you do very much? Most of us want to save the Amazon, not out of some prediction of future rainfall, but just because we want to save the wildlife and the biodiversity. So I just think sometimes, the debates about modeling can get a little sectarian when I think there’s often some broader agreement about kind of basic stuff — masks and social distancing in the case of the coronavirus — and then you work out how much we are going to impose masks and social distancing. How much do we want them? And that’s often determined by things other than the models.


    After compassion, humility.

  8. verytallguy says:


    I’m not sure for the uk, but at least as I understand it, the level of certainly on regional charges in precipitation is very low.

    Table 14.2 in AR5 has a summary of the confidence we have in models to predict current precipitation and future changes from it.

    Current skill features many “Low”. Future changes features many “Medium”. Sceptics might query the latter in the light of the former, but either way we clearly can’t make confident predictions as to how we should adapt.

    Whilst the overall level of extreme precipitation events might go up, those in a given region could even go down. Or equally go up by far more than predicted.

    That’s my extremely amateur understanding FWIW. I look forward to being schooled by those more expert!

  9. vtg,
    Thanks. I wasn’t suggesting that it is certain (I suspect – as you suggest – it isn’t). I was more suggesting that even if it was, designing effective flood defenses in the UK is still very challenging.

  10. izen says:

    A topical example –

    “Japanese weather forecasters had expected heavy rains. The Japan Meteorological Agency issued its highest warning before 5 a.m. Saturday, according to Kyodo News. The agency said the downpour that came was greater than any seen before in the region, the BBC reported. At one point, part of Kumamoto prefecture saw almost 4 inches of rainfall per hour.”

  11. Chubbs says:

    Adequate flood control is like a nuclear powered economy – made even more unlikely by discounting climate change.

  12. izen says:

    I am sure that those better informed on this subject will correct me if in error; but my understanding is that while it is possible to construct flood controls that can mitigate rainfall that exceeds the average by an order of magnitude over a month, it is not realistically possible to have flood control infrastructure that can cope with rainfall that is x10 the average over a day.

  13. MarkR says:

    I’ve looked into heavy precipitation frequency with some colleagues recently. Will wait for review to finish before saying anything new, but the “standard” understanding seems pretty solid.

    Frequency of heavy events has gone up, and in % terms faster for more extreme events. E.g. pick an event that typically happened once per year some time ago. If it now happens twice per year, then it’s like that the event that was heavy enough to happen once per decade some time ago now happens plenty more than twice per decade.

    It’s very regional and sensitive to internal variability, but we have some large scale stats to confirm that the frequency of heavy precip has, on average, increased.

    That could cost a lot of money in insurance, adaptation and abandonment assuming markets haven’t properly worked this out yet.

  14. MarkR says:

    With the spatial pattern, it looks like there are some typical features that seem real.

    Very heavy rainfall in the convective zones is most likely increasing. So if you look inside the Intertropical Convergence Zone (ICTZ) in any model, you’ll probably see it get stronger with warming. But if you look at it on a map the agreement would be worse because models don’t all put their ITCZ in the same place.

    Similarly, you expect regions to get drier as the Hadley Cells expand. If you didn’t used to be under the descending branch but now you are, then you’re going to see big precip changes, most likely dryer typical days. But again, if you look at it on a map, models don’t all have the same sized Hadley cells, so it looks more confused.

  15. Bob Loblaw says:

    Let’s get one thing clear. Infrastructure design for precipitation is not based on averages.


    It’s so prevalent, it just gets referred to as IDF. Precipitation data are analyzed to determine the expected maximum precipitation over a specific time period (minutes, hours, days, etc.). You end up with IDF curves. These are typically published so that engineers can do The Proper Thing. The Canadian example:

    You can’t just add up the record and say “the maximum 100-year precip is the biggest value in this 100-year record”, though. You need to fit the existing data to an acceptable distribution function. Why? A 100-year record will not necessarily have a 100-year storm. It may have a 1000-year storm. You don’t know, so you have to presume a statistical distribution and fit it.

    In a random sequence, the probabliity of a storm with return period X in a record Y is given by the binomial theorum.

    In a 100-year record:
    – the probability of seeing one (and only one) 100-year event is 0.37.
    – the probability of seeing one (and only one) 500-year event is 0.16.
    – the probability of seeing one (and only one) 1000-year event is 0.09.

    You need to account for exceeding a storm of a certain size to do it properly, though.

    Anyone that starts talking about the behaviour of averages in a discussion of extreme precipitation is barking up the wrong tree.

  16. Bob Loblaw says:

    Hmmm. A rather lengthy comment has failed to appear after posting. If caught in a spam folder, please review and release. I have a copy of the text, so I can repost if needed.

  17. Bob Loblaw says:

    …and it has now appeared after I posted a short one…. Go figure.

  18. Mark,
    Thanks, interesting comment.

    I think it all appeared. I find myself getting confused by the probability of 100-year, 500-year, etc events. However – as you indicate – you don’t plan for averages, you need to have some idea of the distribution of events and plan accordingly. As I understand it, we’re already getting caught out on occasion, given that the distributions are already starting to shift.

  19. Bob Loblaw says:

    Yes, the long comment is complete. I don’t know why., but it seems that sometimes after posting a comment it takes a few minutes to appear – even after refreshing the page, visiting another page and returning, etc.

    In the absence of a better explanation, I’ll blame it on some bizarre combination of the WordPress web page structure and caching in my browser (firefox 68.10.0esr on linux).

    For anyone interested, the page I linked to for the binomial calculator also has a short explanation of the binomial probability distribution.


  20. Chubbs says:

    Below is US flood risk assessment recently published by First Street Foundation. Covers flood experience, risk distributions, and climate change on a state-by-state basis. About 10% of US properties are currently at substantial risk (1 in 100 years) with an 11% increase projected due to climate change by 2050. The number of properties at substantial risk is 1.7 times higher than the current FEMA estimate, which drives decision making, due in part to more up-to-date climate change information.

    Click to access first_street_foundation__first_national_flood_risk_assessment.pdf

  21. anoilman says:

    Pretending that we can just gloss over all this seems pretty silly to me.

    In regions where there is relatively little impact to precipitation, there are severe and extreme problems. Even if annual precipitation remains relatively steady, the rain is arriving at the wrong seasons. That means farmers can’t grow their crops.

    And of course if a place is receiving an extreme precipitation event, remember it has to average out over the year? So a drought for the rest of the year should average that out pretty well.

    BC Hydro (most power in BC Canada is hydro) has been suffering from droughts, and have been watching their reserves dip pretty low in recent years.

  22. Dave_Geologist says:

    When Hemingway was asked how he went bankrupt, he’s supposed to have said “slowly at first, then all at once”.

    It’s like that with extreme events. Adaptations and mitigations work, until they don’t. Forests (like a third of the Klamath) which are already out of equilibrium with the current climate but persisting because the forest makes its own microclimate will have zero damage until the next big fire. Then they’ll have infinite damage. Not just the resource value times the fifty years or whatever it takes to regrow – they’ll never grow back. When it comes to humans dying in city heatwaves, it’s when you get three or four days or a week above a certain night-time temperature that mortality skyrockets. You can’t calculate that by multiplying the casualties from a one-night heatwave by three or four. A flood meeting a flood barrier will cause zero damage until it over-tops. Then massive damage. People will rebuild on a site that gets flooded once or twice a century, but not on one that gets flooded once or twice a decade. Etc.

    I’ll give my mandatory warning about what the geological record tells us. It goes ‘way beyond Clausius-Clapeyron. During the Carnian Pluvial Event and the PETM there was 3-4°C of rapid warming, albeit on top of a higher baseline. In both cases the hydrological cycle was supercharged, with much, much more intense rainfall; in the PETM at least, interspersed with multi-decade droughts. In both cases we know that wasn’t just in special sites where the land record is fortuitously preserved, because it changed so much that the global oceans recorded clay-mineral and trace-element anomalies demonstrating that weathering, soil-forming and erosional processes were fundamentally different across the world. It was, literally, a whole different world. One where anything remotely resembling modern agriculture would be impossible across much of the globe. Modern cities too, in places (no rain for thirty years in northern Spain, then a storm which rolls car-sized boulders 100 km).

    Extreme times do have their upsides. The CPE is what kick-started the diversification of the dinosaurs. All those empty niches to move into. So we can thank it for some cool movies, documentaries and museum exhibits. And the diversification of primates dates to just after the PETM. There are always winners from environmental catastrophes. It’s just usually not the previous top dog.

  23. Chubbs,

    I don’t trust 1st street whatever. I know that I am outside of the 1:500 year flood plan now or in 2100. yet floodfactor.com suggests …
    “This property has minimal flood risk.
    This property has a Minimal Flood Factor™

    . Although flood risks across the country are changing because of the environment, this property is unlikely to flood over the next 30 years.”

    I’ve got some land to sell you plus sell you some flood insurance.

  24. Left out this …
    “xxxx xxxxxx xxx, Vicksburg, Mississippi
    FEMA Zone (est.): X Flood insurance: recommended”

    Whuch is 100% BS!

  25. JCH says:

    Back in the day before extreme precipitation events my brother packed his stuff and serendipitously drove out of river bottom just before an earthen dam, an example of flood infrastructure, broke during an extreme precipitation event and drowned 100’s of people. Water is powerful. A lot of water is badass.

  26. Chubbs says:


    Tried floodfactor.com for my current house and received same answer you did. Ditto at previous house and house before that. Guessing “minimal” is the lowest risk available.

    Then checked two higher risk beach properties we had stayed at in New Jersey and Florida – “information for XXX is unavailable please try another location”

    So not the best experience.

  27. From DBB link …

    “In Chongqing, in Sichuan province, more than 100,000 people were evacuated as dozens of homes were destroyed.”

    That must have been some quarantine, 100,000 people stacked up like cordwood in those dozens of homes.

    “They never disclose how the disaster is made or why it has happened,” said Wang Weiluo, a Chinese hydrologist and outspoken critic of the giant Three Gorges hydroelectricity plant.

    “Most people think floods are caused by extreme weather but it is mainly caused by the discharge of reservoirs and the result of flood control works,” he said.”

    Who knew the Chinese government did flood control bass ackwards? The reservoirs are normally meant to hold back flood waters. Once the drainage channels (e. g. open channel flow or rivers) overtop their banks, the hydraulic grade line (HGL) goes way down.

  28. David B Benson says:

    Even worse rain in southwestern Japan.

  29. an_older_code says:

    @ATTP – on a brighter note my eldest daughter was notified this morning that she achieved a First Class degree from your esteemed institution – sadly in History and Spanish not Physics

  30. an_older,
    That’s really good news. Congratulations to your daughter.

  31. anoilman says:

    Extreme precipitation evens also come in other forms.

    Here we have hail storms, but in the face of increasing damage the city and insurance companies have been seeding clouds to prevent\reduce the impacts. Never the less we keep having extreme precipitation events;
    (Personally I wonder what kind of armor plating global warming deniers think will keep grapefruit sized hail out of the house. Steel Plating?)

    In fact we’re still dealing with the damage from the thousand year flood we had a few years ago.
    As a result, this kind of damage is now considered a hundred year flood. Adaptation has been to build a lot of barriers, and they plan to divert as much water as possible and ‘sacrifice’ some farms. The farmers were not amused at being sacrificed.

    And then there’s the joy of forest fires (caused by record setting hot dry weather);

  32. angech says:

    Expecting an extreme weather event with flooding east coast Australia in 2 days for up to 4 days. Hope it stays east enough or Moyhu and I might get a little wet.

    These are not unknown as records of up to 400 mls have occurred in the early distant past of records

  33. KenH says:

    “In fact we’re still dealing with the damage from the thousand year flood we had a few years ago.”

    I don’t think it was a “thousand year flood”. Two in the late 1800’s had larger flows.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.