Weather-related Natural Disasters: Should we be concerned about a reversion to the mean?

by Risk Frontiers July 31, 2017

Professor Roger Pielke Jr (University of Colorado Boulder)


Roger is a long-term Research Fellow of Risk Frontiers and recently it was our pleasure to be able to host him, once again, in Sydney. During this visit, we were rewarded with an insightful seminar entitled Natural Disasters and Climate Change: The Science and the Politics. Below is a brief synopsis of some of the key points raised in Roger’s talk. A pdf of his presentation is available for those wanting further information.

We would also direct readers to Roger’s book entitled: The Rightful Place of Science: Disasters and Climate Change as the most accessible and thoughtful compendium of research in this area and of discussion as to the correct role of science in informing political debate and policy in contentious and important areas like global climate change. (https://www.amazon.com.au/d/ebook/Rightful-Place-Science-Disasters-Climate-Change-Pielke/B00SZ83XMG/ref=sr_1_fkmr0_2?ie=UTF8&qid=1501026563&sr=8-2-fkmr0&keywords=roger+Pielke+jr).


The world is presently in an era of unusually low weather disasters. This holds for the weather phenomena that have historically caused the most damage: tropical cyclones, floods, tornadoes and drought. Given how weather events have become politicized in debates over climate change, some find this hard to believe. Fortunately, government and IPCC (Intergovernmental Panel on Climate Change) analyses allow such claims to be adjudicated based on science, and not politics.  Here I briefly summarize recent relevant data.

Every six months Munich Re publishes a tally of the costs of disasters around the world for the past half year. This is an excellent resource for tracking disaster costs over time.  The data allows us to compare disaster costs to global GDP, to get a sense of the magnitude of these costs in the context of economic activity.  Using data from the UN, here is how that data looks since 1990, when we have determined that data is most reliable and complete.

The data shows that since 2005 the world has had a remarkable streak of good luck when it comes to big weather disasters, specifically:

  • From 2006 to present there have been 7/11 years with weather disasters costing less than 0.20% of global GDP.
  • The previous 11 years saw 6 with more than 0.20% of global GDP.
  • From 2006 to present there has be zero years with losses greater than 0.30% of global GDP.
  • The previous 11 years had 2, as did the 6 years before that, or about once every 4 years.
  • According to a simple linear trend over this time period, global disasters are 50% what they were 27 years ago, as a proportion of GDP.

Why has this occurred? Is it good luck, climate change or something else?

By disaggregating the data phenomenon by phenomenon we can get a better sense of why it is that disaster costs are, as a proportion of global GDP, so low in recent years.

A good place to start is with tropical cyclones, given that they are often the most costly weather events to occur each year.  The figure below shows global tropical cyclone landfalls from 1990 through 2016. These are the storms that cause the overwhelming majority of property damage. Since 1990 there has been a reduction of about 3 landfalling storms per year (from ~17 to ~14), which certainly helps to explain why disaster losses are somewhat depressed.

Even more striking is the extended period in the United States, which has the most exposure to tropical cyclone damage, without the landfall of an intense hurricane. The figure below shows the number of days between each landfall of a Category 3+ hurricane in the US, starting in 1900. As of this writing the tally is approaching 4500 days, which is a streak of good fortune not seen in the historical record.

A very conservative estimate of the effects of this “intense hurricane drought” is that the United States is some $70 billion in arrears with respect to expected hurricane damage since 2006. In fact, it is not widely appreciated but the US has seen a decrease of about 20% in both hurricane frequency and intensity at landfall since 1900. I urge caution placing too much significance on linear trends, as they are quite sensitive to start and end dates, but there is very little to indicate that tropical cyclones are either more frequent or intense.

Data on floods, drought and tornadoes are similar in that they show little to no indication of becoming more severe or frequent.  The IPCC concludes:

  • “There continues to be a lack of evidence and thus low confidence regarding the sign of trend in the magnitude and/or frequency of floods on a global scale.”
  • “There is low confidence in observed trends in small spatial-scale phenomena such as tornadoes and hail.”
  • “There is low confidence in detection and attribution of changes in drought over global land areas since the mid-20th century.”

Read the rest at Risk Frontiers.

HT/ The GWPF

Advertisements

49 thoughts on “Weather-related Natural Disasters: Should we be concerned about a reversion to the mean?

  1. Roger Pielke jr is being very cautious, as if the trend is not due to any real long term climate change, but random variation, reversion to the historic average is to be expected.

    • “reversion to the historic average is to be expected.” On what grounds?

      “as a proportion of GDP.”

      as a first step he should explain why this is the most relevant index against which to normalise the data. Sure property gets dearer, prices go up, but GDP is about PRODUCTION . How is that relevant ?

      Thanks again to RP Jr for trying to put some balance on the alarmists hype.

      • The quality of disastrousness is related ti how easily you can afford to repair the damage.

        Hence GDP

      • I believe the idea of scaling to GDP is to normalize how much property damage is done relative to how much property is exposed. For example, the U.S. population in coastal areas From Mexico to Maine has increased from 10 million in 1900 to 50 million in 2010, and the real value of personal and real property held by the average individual has quadrupled over that same period. That means the total value of property at risk of major sea storm damage today is about 20 times higher than in 1900. If we had storms of exactly the same magnitude and frequency today, we would expect fatalities to be 5 times higher and property damage costs/insurance payouts to be 20 times higher. Of course, we see nothing of the kind due to progress in weather forecasting, telecommunications, construction codes, transportation, emergency services, and other mitigating technologies enabled by the harnessing of cheap and abundant energy. Normalizing to population and property values would likely yield a more meaningful result, but either way, the trend is that we are flaunting nature more and more and being harmed by it less and less.

      • GDP is an easily-available proxy for wealth. It is available adjusted for inflation and whilst pretty inaccurate, is more accurate than anything else. Comparing damage in terms of one year’s GDP versus another year’s GDP is a pretty good comparison of how damaging a storm is in term of its effect on people.

      • As I understand what RPjr is writing, if there is no real long term climate change producing this decline in severe weather, then an increase in severe weather would mean nothing more than random variation–ie there is nothing changed. This is a restatement of the null hypothesis.

  2. “European cities will be plunged beneath rising seas as Britain is plunged into a Siberian climate by 2020.” -Paul Harris, 2004, UK Ecojournalist

    “A billion people could die from global warming by 2020.” -John Holdren (Obama’s Science Czar), 1986

    We got barely TWO YEARS for all that to happen, people. For the temperature to go through the roof, for the seas to suddenly rise, and the bad bad weather to hit. It hasn’t happened. In fact it’s the opposite on all fronts. The weather is … calmer!!!. The temperatures … they’re not going anywhere. Arguably they’ve dropped since the thirties.

    And the rate of sea level rise has … stayed the same, and it remains at just the most minimal trifling level:

    • Thanks Eric, nice pic. You seem to have overlooked something called TIDES. Since we do not know what part of the tidal cycle either photo comes from it is meaningless as an indication of sea level rise

  3. Do we have to worry about reversion to the mean? Reversion to the mean assumes that there is a natural state of things from which we may diverge on a temporary basis. link

    All the evidence we have is that the climate is in a constant state of flux. It’s just as likely that things will get better as it is that they will get worse. Things could get worse. We can’t ignore that possibility.

  4. Excellent article which exposes the absurd lies, dishonesty and deception of climate alarmist scientists, mainstream media alarmism propaganda reporters and alarmist politicians in government.

    The Trump Administration needs to continue to push for dismantling the climate alarmism organizations and supporters in government and the media that have lead us so far astray from climate reality based on nothing but ltheir ies and distortions.

  5. From the “leaked” latest version of the U.S. GLOBAL CHANGE RESEARCH PROGRAM
    CLIMATE SCIENCE SPECIAL REPORT (CSSR)

    “Significant advances have also been made in our understanding of extreme weather events and how they
    relate to increasing global temperatures and associated climate changes. Since 1980, the cost
    of extreme events for the United States has exceeded $1.1 trillion, therefore better understanding of the frequency and severity of these events in the context of a changing
    climate is warranted.”

    —————-

    While costs of natural disasters are important, using them to them to measure frequency and severity of extreme weather events is not scientific. It’s pure propaganda. But what else would you expect from progressive SJW government scientists?

  6. Not surprisingly Al Gore in an interview of August 5 published in the Weekend Australian continues the claim that ” over the last decade…climate related weather events are far more numerous and far more destructive.”
    The interviewing reporter was apparently not knowledgeable enough to challenge this claim which is made repeatedly by Gore in the Main Stream Media.

  7. As others have noted, a warmer climate, particularly in the Arctic which has the greatest share of the warming period we’ve had in the last decade or two of the 20th century, and the plateau since have reduced differences in temperature between tropical and polar latitudes. For the engine of climate the severity of storms is proportional to the the differences between temperature at ‘intake’ and ‘exhaust’. Lower difference, calmer weather.

    That said, I stopped cautioning in recent years that the return of a cold period will bring back, tropical storms, droughts and floods of the 1950s and given the 60-70year cycle in climate, that time may be drawing nigh. Since warmer proponents won’t accept the cyclic natural variation so they will chime in worsening climate.

    • Gary
      The energy has gone to the poles instead of hurricanes. Something changed that removed the blocking that was preventing atmosphere to migrate into the polar regions, particularly the Arctic.

      Removing the blocking reduced the incedence of hurricane initiation and intensification. Hurricanes form on pressure.

      While sea ice had been decreasing, 2005 was a larger drop. Apart from 2006, Arctic sea ice reduction accelerated.

      The energy did not go away, it has just been diverted. Where has some of the greatest increase in temperature been. The high northern latitudes and Antarctica. In noted corridors.

      The equator to pole temperature reduction occured by forced transport.

  8. Reversion to the mean implies a number of things about the probability distribution of the object. In particular, it presumes ergodicity. While it is convenient to model the distribution as ergodic over short periods, it is almost certainly a bad assumption over longer periods.

  9. I don’t think percentage global GDP gives an accurate view of damage trend because most of the damage occurs along coastlines while GDP is a product of entire land area. I’d rather see it measured with a first world currency normalized for inflation and normalized for something like “representative coastal real estate value”, (which might be as simple as population?).

    Consider that if every country’s GDP but one held constant and physical hurricane damage remained constant, a steady increase in the GDP of say Uganda will produce a declining trend.

  10. “reversion to the historic average is to be expected, On what grounds? ”
    I believe it’s called the Central Limit Theorem.

    • Assuming the processes are stochastic and not chaotic, where the latter can only be reasonably characterized over a limited frame of reference (i.e. scientific domain).

  11. I haven’t done work on this topic for years, but when I did my observations were that a warmer climate appeared to coincide with fewer extreme weather events, and a cooler climate with more extreme events.

    We also predicted in an article published in 2002 that natural global cooling would start by 2020-2030. I am now of the opinion that such cooling will commence by about 2020, but such timing is very difficult to predict with precision.

    If our 2002 prediction is correct re imminent (moderate) global cooling, Earth should also experience more extreme weather events (e.g. major hurricanes, etc. ) starting in the next few years.

    This is not simply a prediction of regression to the mean – it is a prediction of a natural cycle that has been beneficial to humanity in recent decades, and is about to change for the worse.

    If I am correct, it will be truly ironic – our “IPCC climate brain trust” will have been proven wrong on all counts – which is their dismal predictive track record to date. Watch this space…

    Regards, Allan

  12. How does this square with NYT reports that the new US state of the climate report says that there has been (and will be) an increase in extreme weather events? Has Roger commented on this report anywhere? Is it simply more NYT hyperventilating?

    • As far as I can tell the report is mostly projections with “High Confidence” that are based on the IPCC and modeling:
      “The findings in this report are based on a large body of scientific, peer-reviewed research, as well as a number of other publicly available sources, including well-established and carefully evaluated observational and modeling datasets. … scientific assessments (such as the rigorously-reviewed international assessments from the Intergovernmental Panel on Climate Change; IPCC 2013)”

      Most of the paragraphs detail the terrible projections if the “unprecedented” heating continues (“very high confidence”) The temperature data that everything is built on has the same issues as have been discussed here many times And then there are models …

      Trends calculations are based on interesting periods:
      “For the observed trends, 1986–2015 is generally chosen as the most recent 30-year period (2016 data was not fully available until late in our development of the assessment). … The other commonly used reference period in this report is 1976–2005 … This period is used for graphs that illustrate projected changes simulated by climate models.”

      Of course having more rain (if indeed that trend is correct as they start with 1958 and don’t say how much happened before 1980) is a sign of warming even though I’m certain I heard less rain is also a sign of warming (or is that good rain vs. bad rain?)

      The quoted person in the NYT article is a “political scientist” which is an interesting oxymoron.

      They should fear that this will not make it through the Trump administration, however I’m certain it will get just as much hype regardless of what the government decides to do

  13. “The IPCC concludes:
    1.There continues to be a lack of evidence and thus low confidence regarding the sign of trend in the magnitude and/or frequency of floods on a global scale.
    2.There is low confidence in observed trends in small spatial-scale phenomena such as tornadoes and hail.
    3. There is low confidence in detection and attribution of changes in drought over global land areas since the mid-20th century.

    The third of these is the only thing that makes sense. Assuming that climate scientists keep track of the magnitude and frequency of floods around the globe, and I can’t imagine that they don’t, then the trend is what it is. Simply plot the data and calculate the damned trend. If it’s not statistically significant, the balance of evidence is that there is no trend. Saying “low confidence regarding the sign of the trend” is Orwellian doublespeak for “the trend in the actual data isn’t yet as high as we think it will eventually be if we just wait a little bit longer.”

    Ditto for number 2.

    If they were honest, number 3 is a foregone conclusion with respect to attribution. No one will ever be able to confidently attribute changes in drought to any particular cause absent the ability to experiment on the climate system. Simply observing droughts as they occur cannot possibly give the kind and amount of information you’d need.

    • The article you link has “Weather-related losses and damage have risen from an annual average of about $50 billion in the 1980s to close to $200 billion over the last decade, according to the Munich Re insurance group” under “Story highlights”, but provides no link — nor is that text or supporting information found elsewhere in the article. A curious choice given your criticisms.

      Munich Re is given as a source both for this “highlight” and the graphs above, but they don’t actually contradict each other. Your link is to losses in total dollars, unadjusted by inflation. Roger’s graph shows losses as a proportion of GDP. In a world that contains both inflation and real GDP growth, we would naturally expect the raw dollar amounts from “weather-related losses and damage” to continually rise — it says nothing at all about whether weather-related incidents are increasing in intensity or number. (I do note that Roger claims that the Munich Re damages are only complete from 1990 on, so he does not compare to numbers in the 80s — and if he is correct, neither should the world bank.)

      I believe economic impact as a percentage of GDP is a much more relevant measure for policy than raw dollar totals. Of course, that also says nothing at all about whether weather-related incidents are increasing in intensity or number, but it *does* say something meaningful about the collective impact of weather-related incidents.

      • Probably fewer structures were insured globally in the 1980s than now. that should be adjusted for, if possible, to get an apples-to-apples comparison.

  14. Roger, have you fit the data (number of days between major hurricanes) to a distribution and/or looked at year to year independence? Without the actual data, I couldn’t look.

  15. The funniest thing about this prediction of “More Extreme Weather Coming!” is that ordinarily, this should be a hands down easy bet to win. Statistically, since “exactly normal” weather years almost never happen, you should have a 50% chance of being right each year, if you make that prediction at the beginning of every year. Then you just ignore the years it didn’t work out, and trumpet the years that you were “Right” (the years in which your statistical bet payed off)

    What’s so funny is that this “can’t lose” bet has now lost for more than 10 years in a row. heh heh.

    Kind of like Al Gore trying to time his movie so that it would be playing during the “hottest” weather all year.

  16. Roger Pielke Jr. has not considered the possibility that there have been recent changes in human civilization which reduce losses compared to GDP. One could simply be that cell phones are now carried by a large part of the adult population, which reduces weather surprises. We are all safer now. Another could be the tougher building standards in locations subject to storm damage, such as along the Gulf Coast in the United States. One goal of those tougher standards is to achieve lower levels of storm damage from future storms.

  17. Monetary impact may not be the best measure of severe weather. If Hurricane Andrew (1992) had gone 45 miles to the right of its actual path, it would have hit Miami and New Orleans instead of Homestead and Houma. I believe the cost of it hitting Key Biscayne instead of trailer parks in Homestead would have been significantly higher. This is with no increase in frequency or severity.

    Hurricanes are down in any event. I am not really concerned about storms that landfall in Iceland.

    • OTOH, if things had been a little different, Sandy’s cost would have been halved. Over time, hurricanes’ monetary impact should average out.

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 )

Google+ photo

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

Connecting to %s