The Basics of Extreme Weather and Global Warming

January 2, 2014 3:23 pm7 comments

An oft-heard statement is that the weather is getting more extreme because of global warming.  As opposed to, say, the benign, mild weather conditions associated with the Last Glacial Maximum 20,000 years ago.

Many long-term records of extreme weather events are fraught with problems, usually involving changes in the way that extreme events are detected and measured.  If you go beyond the extreme weather itself to consider human disasters caused by extreme weather, additional complications such as changes in population exposure over time arise.

Tornado damage by PhotoJunkie! via Flickr

In order to lay the groundwork for future discussions of extreme weather events, I’d like to discuss the basic scientific expectations for changes in extreme events.  What is extreme weather?  What ought to happen with global warming?

What is extreme weather?

Much confusion arises from the fact that “extreme weather” has two different meanings.

One meaning is that of “statistically extreme weather”, weather that happens extremely rarely.  An example would be the record high temperature for a particular day and place.  By definition, it has only happened once (except for ties) in the history of weather records, and so is extremely rare.  For this definition to make sense, some reference period such as the “period of record” must be chosen.

The other meaning is that of “extremely dangerous weather”.  An example would be a tornado.  Though there are hundreds of tornadoes each year, each one is an extreme weather event, whether or not it happens to cause damage.  Other examples are tropical cyclones, floods, and droughts.  Sometimes the term “weather” is defined loosely enough to include wildfires.

There is overlap between these two types of extreme weather.  For example, a record hot day in the summer is statistically extreme, and it can also be extremely dangerous.  A record hot day in the winter is usually not so dangerous, unless it causes dangerous snow or ice melt.

What ought to happen?

For statistically extreme weather, the answer, at least on the surface, is straightforward.  If the average value of some weather variable (temperature, wind speed, etc.) has changed compared to a reference period, then one extreme ought to become more common while the other extreme becomes less common.  Thus we expect (and see) more maximum temperature records and fewer minimum temperature records.

The expected change is asymmetrical.  In other words, the average number of records set during a given year ought to increase rather than staying the same, no matter the direction of the change.  Consider a location with a 100-year record of a stable climate.  On average, a new daily high temperature record will be set 3.65 times a year and a new daily low temperature record will be set 3.65 times a year, for an average of 7.3 new records per year.  Now suppose the temperature suddenly goes up five degrees.  The next year, there might be 40 maximum temperature records, but probably zero minimum temperature records.  One kind of record has become more frequent, while the other has become less frequent.  But the total number of annual records has increased by 32.7!

It is also conceivable that the variability of weather might change.  Maybe temperatures will become more erratic, or maybe less erratic, for example.  Such a change would not be an obvious consequence of global warming, while an increase in the frequency of maximum temperature records by itself is an obvious consequence of global warming.

Changes in the frequency of extremely dangerous weather must be considered on a case by case basis.  Very few changes are obvious.  This is because most dangerous weather arises from various types of localized dynamic instability in the atmosphere, and it’s rarely clear whether the generation of atmospheric instability ought to speed up or slow down in a warming atmosphere, let alone whether other environmental changes will make that instability easier or harder to respond to.

A common oversimplification goes that “a warmer atmosphere has more energy, and therefore more energetic weather.”  Scientists sixty years ago showed that this was wrong.  Sure, a warmer atmosphere has more energy, but most of that energy can’t do anything but radiate away to space.  It is the variation of energy (temperature) from place to place that drives the planetary winds.  And in the Northern  Hemisphere, as the difference in temperature between the equator and pole decreases, the circulation ought to become less energetic.

Perhaps the most obvious tendency for extremely dangerous weather is that the strength of the heaviest downpours ought to increase.  This is because rainfall intensity is a product of the rate of upward motion of the air and the amount of water vapor it contains.   Since warmer air can contain more water vapor, about 7% more for each degree Celsius of warming, a given updraft will tend to produce heavier rain.  This effect might even be felt at the storm scale, more than making up for the decline of temperature variations.

Note that this is different from saying that there will be more rain on average.  The general consensus is that there will indeed be more rain on average, but this conclusion is not obvious from simple principles.  Sure there will be more water vapor in the air, but what if the air rises more slowly?  Might that not compensate for the increased water vapor?

Also, “more rain on average” doesn’t mean “more rain on average everywhere”, it just means that the global average amount of rainfall (and snowfall) is expected to increase.  It doesn’t even necessarily mean that more than 50% of the Earth’s surface will see increased rainfall.  Along similar lines, increased drought frequency (another expectation) doesn’t mean that drought frequency will increase everywhere, nor even that drought frequency will increase over a majority of the globe.

As a general rule, most changes related to rainfall will be very place-specific.  This is because the production of precipitation requires ascending air, and air ascending in one place requires air to be descending in another place.

Let us suppose, for the sake of example, that 60% of the globe will see more and heavier rain, i.e., more floods, and likewise that 60% of the globe will see more droughts.  Presumably those will tend to be different parts of the globe, so these numbers would guarantee that only 20% of the globe would see an increase in both drought and flood.

In summary, some aspects of extreme weather are easy to predict, others less so.  We’ll have to consider them each in turn.

Tags:

THE FORUM'S COMMENT THREAD

  • Hello, thanks for this helpful post. Am I correct in understanding that although studies have shown an increase in record temperature setting (on account of the rising trend of average temperatures), that simulations/surveys of the prevalence of “dangerous weather events” has been inconclusive?

    • Each type of dangerous weather event needs to be considered separately. For some, present and future trends are inherently inconclusive (example: tornadoes), but for others, present and future trends are fairly solid (example: droughts).

      I doubt we’ll have the time to discuss all types of extreme weather in CCNF. In the meantime, the IPCC SREX (Special Report on Extreme Events) from a couple years ago seems to be widely respected as a good summary. The latest IPCC report (AR5 WG1) includes a summary table showing where the AR5 opinion differs from SREX, but I haven’t yet looked at those differences in detail.

  • Here is the link to an open access article in PNAS by Rahmstorf and Coumou (2011), that expands upon, and lays out the theoretical basis of, John’s discussion of the expected frequency of record events. I found it to be a very helpful paper.

    http://www.pnas.org/content/108/44/17905.full.pdf

  • I received the following comment from Bob Koss:

    John your math is correct for that station with 100 years of records in a stable climate. But that isn’t the situation we have. It is known the PDO alters the climate and all station records are not equal in length.

    I find the chronic announcements such as 250 new highs were established vs only 50 new lows to be a disingenuous presentation of the reality of temperature history. I contend the reason for the current disparity of reported high vs lows is that many of the stations setting records are doing so due to having only been established in the last 60 or fewer years and barely being around for about one full cycle of the PDO if even that much. Stations with records starting during the negative phase of the PDO would likely have established the majority of their low records early in their history and most of their high records in more recent times during the positive phase of the PDO.

    A more realistic view of record temperatures would be to use long records of an equal length. Below I link to a graph I made last January of annual state temperature records through 2012 where you can plainly see most records, both high and low, are still held by years 1885-1950. The 50 states still have 32 high and 26 low records during that period. I even credited just the most recent year of record if it happen to tie an already established prior record for a state. That is about two full PDO cycles for all states and avoids the ambiguity of different record lengths.

    When discussing temperature extremes no decade has been more extreme than the 1930s which still holds 29% of all state records with 19 record highs and 10 record lows. Additionally, I see now that they have completed tropical cyclone reanalysis for the 1930s. The hurricane season of 1933 now holds the record value for ACE at 259 surpassing the 2005 season. That was one extreme decade for both temperature and storms.

    http://i46.tinypic.com/9gy2hx.gif

    I also have graphs of monthly state records which show a roughly similar distribution and would provide more data points, but I don’t want this post to get sent to spam for having too many links. Those only go to 2008 tho’.

    • Bob –

      What I said applies no matter why the climate has changed. If CO2 has made things warmer compared to the period of record, we’ll see more maximum temperature records. If the PDO has made things warmer compared to the period of record, we’ll see more maximum temperature records.

      There are a couple of problems with what you wrote. First, the statement of “250 new highs vs. 50 new lows” is usually made with respect to the United States, and the United States doesn’t automatically become warmer with a positive PDO. Generally, the northern US gets warmer and the southern US gets cooler. Also, the PDO is now negative, so if you are correct in your implication that the PDO is driving things in the United States we would have seen a flip in the ratio of record highs to record lows starting several years ago, and we did not.

      I do agree that the 1930s had the most dramatic extremes in the United States in the past century, and most statewide temperature extremes were set before 1950. Considering the possible reasons for that would get us far off topic.

      I also think that anybody who looks at trends in some local-scale or regional-scale meteorological variable over 60 years or less and assumes that those trends are entirely caused by global warming is making a serious mistake.
      – John

Leave a Reply

You must be logged in to post a comment.

PUBLIC COMMENT THREAD

  • Lulu Liu

    Hello, thanks for this helpful post. Am I correct in understanding that although studies have shown an increase in record temperature setting (on account of the rising trend of average temperatures), that simulations/surveys of the prevalence of “dangerous weather events” has been inconclusive?

    • http://atmo.tamu.edu/profile/JNielsen-Gammon John Nielsen-Gammon

      Each type of dangerous weather event needs to be considered separately. For some, present and future trends are inherently inconclusive (example: tornadoes), but for others, present and future trends are fairly solid (example: droughts).

      I doubt we’ll have the time to discuss all types of extreme weather in CCNF. In the meantime, the IPCC SREX (Special Report on Extreme Events) from a couple years ago seems to be widely respected as a good summary. The latest IPCC report (AR5 WG1) includes a summary table showing where the AR5 opinion differs from SREX, but I haven’t yet looked at those differences in detail.

  • http://glacierchange.wordpress.com/ Mauri Pelto

    Daily heavy precipitation in the United States trend. BAMS state of Climate assessment from 2012 also useful.

  • http://ecologicallyoriented.wordpress.com/ Jim Bouldin

    Here is the link to an open access article in PNAS by Rahmstorf and Coumou (2011), that expands upon, and lays out the theoretical basis of, John’s discussion of the expected frequency of record events. I found it to be a very helpful paper.

    http://www.pnas.org/content/108/44/17905.full.pdf

    • http://atmo.tamu.edu/profile/JNielsen-Gammon John Nielsen-Gammon

      The theoretical analysis in Rahmstorf and Coumou (2011) is sound. There’s some dispute about its specific application to the Russian heat wave, but that’s a side issue.

  • http://atmo.tamu.edu/profile/JNielsen-Gammon John Nielsen-Gammon

    I received the following comment from Bob Koss:

    John your math is correct for that station with 100 years of records in a stable climate. But that isn’t the situation we have. It is known the PDO alters the climate and all station records are not equal in length.

    I find the chronic announcements such as 250 new highs were established vs only 50 new lows to be a disingenuous presentation of the reality of temperature history. I contend the reason for the current disparity of reported high vs lows is that many of the stations setting records are doing so due to having only been established in the last 60 or fewer years and barely being around for about one full cycle of the PDO if even that much. Stations with records starting during the negative phase of the PDO would likely have established the majority of their low records early in their history and most of their high records in more recent times during the positive phase of the PDO.

    A more realistic view of record temperatures would be to use long records of an equal length. Below I link to a graph I made last January of annual state temperature records through 2012 where you can plainly see most records, both high and low, are still held by years 1885-1950. The 50 states still have 32 high and 26 low records during that period. I even credited just the most recent year of record if it happen to tie an already established prior record for a state. That is about two full PDO cycles for all states and avoids the ambiguity of different record lengths.

    When discussing temperature extremes no decade has been more extreme than the 1930s which still holds 29% of all state records with 19 record highs and 10 record lows. Additionally, I see now that they have completed tropical cyclone reanalysis for the 1930s. The hurricane season of 1933 now holds the record value for ACE at 259 surpassing the 2005 season. That was one extreme decade for both temperature and storms.

    http://i46.tinypic.com/9gy2hx.gif

    I also have graphs of monthly state records which show a roughly similar distribution and would provide more data points, but I don’t want this post to get sent to spam for having too many links. Those only go to 2008 tho’.

    • http://atmo.tamu.edu/profile/JNielsen-Gammon John Nielsen-Gammon

      Bob –

      What I said applies no matter why the climate has changed. If CO2 has made things warmer compared to the period of record, we’ll see more maximum temperature records. If the PDO has made things warmer compared to the period of record, we’ll see more maximum temperature records.

      There are a couple of problems with what you wrote. First, the statement of “250 new highs vs. 50 new lows” is usually made with respect to the United States, and the United States doesn’t automatically become warmer with a positive PDO. Generally, the northern US gets warmer and the southern US gets cooler. Also, the PDO is now negative, so if you are correct in your implication that the PDO is driving things in the United States we would have seen a flip in the ratio of record highs to record lows starting several years ago, and we did not.

      I do agree that the 1930s had the most dramatic extremes in the United States in the past century, and most statewide temperature extremes were set before 1950. Considering the possible reasons for that would get us far off topic.

      I also think that anybody who looks at trends in some local-scale or regional-scale meteorological variable over 60 years or less and assumes that those trends are entirely caused by global warming is making a serious mistake.
      – John

  • Pingback: FORUM REPORT 01

  • Pingback: ‘Risky Business’ report on the economic risks of climate change for general commentary; predictions of sea level rise and additional 56-108 extremely hot days per year by 2100 under a “business as usual” scenario for fact checking.

  • Pingback: ‘Risky Business’ report on the economic risks of climate change for general commentary; predictions of sea level rise and additional 56-108 extremely hot days per year by 2100 under a “business as usual” scenario for fact checking.

  • Pingback: Are there going to be an additional 56-108 extremely hot days for year 2100 under a “business as usual” scenario? [Fact Checking & General Commentary]