"Our greatest responsibility is to be good ancestors."

-Jonas Salk

Sunday, August 15, 2010

Seeking Precedent for Asian Jet Stream Anomaly

One way to approach this is just through the local temperature anomaly in and around Moscow.

Weather isn't really Gaussian. Physical constraints truncate the tails of temperature distributions. So we can ask this as a relevant objective question: has a prior calendar month on record anywhere ever been 4 standard deviations hot or cold?

Update: There is a 30 day period in a place called Lukojanov that seems to be pushing 6 standard deviations.

That's handy because we can refer to the industrial literature, where "six sigma" is a commonly stated ambitious goal. If the distribution were Gaussian, this would happen 3.4 times per million samples, or once every 300,000 months or about 25,000 years. Maybe there are 250 regions on this scale, so generously we have a slight chance of matching this on the observational record, and a pretty good chance of a historical match. But as I said, that's probably a huge overestimate, because I don't think the tails are occupied.

But, it becomes an objective question.

19 comments:

Vinny Burgoo said...

Coat-trailing, that's what it's called. And I fell for it. Half an hour of Googling. Here you are: In March 1843, temperatures in Fort Leavenworth were 4.52 sigmas below the month's mean for the period 1835-2005. Minneapolis/St Paul and Fort Scott were also more than four sigmas colder that month. Savannah was more than four sigmas colder in November 1845. Extracted from a survey of 15 GHCN stations by Nielsen-Gammon and McRoberts. They're a bit suspicious of the Savannah result but reckon the others are fairly solid. They don't think any subsequent warming trend can have skewed the monthly means for the whole period very far but they don't put a figure on this (or not in the bits I've read).

Unspecified parts of the Arctic and Pacific also had a four-sigma positive temp. anomaly some time in 1982. I can't remember the details. This may have had something to do with the Soufriere volcano. Either that or the volcano gave other parts of the world a four-sigma kick one way or the other at another time.

Then there's Evan Mills of Lawrence Berkeley. He claims that the whole of the 2003 European heatwave was a six-sigma event.

What does all this mean? Not a lot. I could trail a coat of my own but I'm not in the mood.

Michael Tobis said...

Thanks, Vinny. Links please?

Since the Russia event is worse than the Europe event, the Europe event being in six sigma territory doesn't count as reassuring evidence.

Volcanic forcing also can be ignored. The hypothesis is that the system can freely wander into 4+ sigma monthly temperature anomalies.

Hank Roberts said...

So -- the warming predicted in the Arctic: http://www.google.com/search?q=climate+prediction+arctic+warming

Is "slightly less" or "almost as much" warming expected near but not quite in the Arctic?

Any mechanism related to warming the Arctic that also relates to what's happened around the Arctic?

Is there any model for what happened to cause the sudden cooling that gave us mammoths frozen in the ice with green stuff in their stomachs?

Steve Scolnik said...

Re: 1843 extremes

THE UNUSUAL GENERAL CIRCULATION PATTERN OF EARLY 1843
The year 1843 produced some unusually cold months in the upper Mississippi Valley, with March about 25°F below normal for the largest monthly departure of record in the area. A short summary of the winter months of 1842–43 in Wisconsin is included. A check of climatological data for Europe showed that unusually mild weather prevailed in eastern Europe during January and February 1843, while March 1843 was near normal. The effects of snow and ice cover in maintaining a strong and persistent anomalous flow pattern is alluded to. A high frequency of record coldest and warmest months in Wisconsin occurred during the 1830's and 1840's. High dust content of the upper atmosphere from frequent volcanic activity with reduced incoming solar radiation during the 1830's may be partially responsible for the occurrence of the many cold months.

http://journals.ametsoc.org/doi/abs/10.1175/1520-0493%281970%29098%3C0266%3ATUGCPO%3E2.3.CO%3B2

Steve Scolnik said...

Also:

The Extreme Temperature Anomalies of March 1843 and February 1936
The month of March 1843 has been characterized as exhibiting the greatest temperature anomaly of any month during the period of instrumentally measured meteorological data in the United States. The March 1843 departure patterns from recent temperature normals are compared to those of February 1936 which were of the same magnitude (−30°F/−16.7°C), though perhaps of lesser areal extent. The frigidity of the air during February 1936 set some all time low temperature records that still stand today.

http://journals.ametsoc.org/doi/abs/10.1175/1520-0493%281979%29107%3C1688%3ATETAOM%3E2.0.CO%3B2

Vinny Burgoo said...

The standard deviation calculations for March 1843 are from 'March 1843: The Most Abnormal Month Ever?' by John W. Nielsen-Gammon and Brent McRoberts. It's a chapter in _Historical Climate Variability and Impacts in North America_; generous previews available at Google Books.

A half-hour WebEx presentation by one of the authors is available here:

http://ams.confex.com/ams/88Annual/wrfredirect.cgi?id=8029

Answering a question at the end, he says that volcanic activity wasn't implicated.

Michael Tobis said...

Maybe John will comment on this, since he's been reading here.

However, Vinny has done a good job finding a candidate. If last month was only the second weirdest month, or tied for first, that is a bit reassuring.

Another example before 1970 or so would be great; it would give us a hint of the repeat interval.

Note also that the 1943 example is a winter cold outbreak so we don;t really have an analogous situation, but I don't mean to be moving goalposts.

Vinny, I don't understand the coat-trailing complaint (nor do I understand the analogy, but never mind that). I appreciate your discovery. I learned something. You learned something and taught it to us. Is that a problem?

Aaron said...

You want 6 sigma? Prior to 1970, rain was essentially unknown on Greenland. People could live and travel around Greenland all their lives without ever seeing rain.

Now, it has been more than 44 months since there has been a month without recorded rain somewhere on Greenland. Further, rain has occurred in all settlements, at least once per year.

Looks to me like the post 2005 Greenland climate is 6 sigma off of the 1920 to 1970 Greenland climate

Michael Tobis said...

Aaron, seriously? Rain in Greenland in February?

Reference, please?

n-g said...

Vinny has the citation correct. This was an event that my predecessor as Texas State Climatologist, Prof. John Griffiths, was alerted to by folks at NCDC a long time ago. He passed the microfiche printouts of the old Army weather logs to me when he stepped down. Cary and Lesley-Ann putting their historical climate variability book together was the incentive necessary to finally write up the event.

To post a copy of the paper would be to deny them royalties (it's a very good book), but here's the abstract:

Weather observations taken four times daily across the present-day central and eastern United States permit a detailed reconstruction of monthly climate and daily weather during March 1843. Based on normalized departures of temperatures from the historic monthly mean, March 1843 may be regarded as the most anomalous month in recorded history for the central and eastern United States. In places, the average temperature for the month was more than 25 °F (14 °C) below normal, and the expected return frequency for such an anomaly is thousands of years. Using the detailed observations and recent analogs, the daily weather maps for March 1843 can be reconstructed. They show a storm track displaced well to the south of its normal position, with occasional snowstorms bringing winter weather to the Deep South, the Mid-Atlantic States, and New England. The extended severe winter weather caused hardship throughout the area.

Constructing weather maps from 165 years ago (surface by hand, upper air by analog) was lots of fun.

For the statistics, we analyzed 15 representative GHCN stations in the central and eastern US for 1835-2005, computing normalized anomalies (subtract the mean, divide by the standard deviation). Most of the top events were from the first 15 years of that period and only affected individual stations, implying a problem with early record-keeping. But the 1843 event was spatially coherent and undoubtedly real. The largest anomalies were on the edge of the network, so its true intensity may have been slightly greater.

When we averaged the normalized anomalies across all 15 stations, the extreme events seem to be randomly distributed throughout the period of record. But none of the top 15 events were in June, July, August, or September. Also notable is that 14 of the 15 were negative (cold) anomalies, implying a skewed distribution in the cooler months.

Anything else you'd like to know?

- John Nielsen-Gammon

P.S. I have a great hour-long general-interest seminar on the event. Have invitation, will travel.

Hank Roberts said...

Chameides may be, or may know, the right person to ask about weather in Greenalnd. I think he's on his way there now. Here's an interim report:
http://www.nicholas.duke.edu/thegreengrok/baffinisland#comments

".... I present for your edification two non-scientists from northern Canada — Billy Arnaquq and Robert Joamie.
Billy, our tour guide
Billy, one of guides, has lived in this area his whole life and he's seen big changes.

Our two last stops on our way to Greenland were at the town of Pangnirtung (population about 1,300) and the neighboring Auyuittuq National Park, both located on Baffin Island just south of the Arctic Circle. (See map below.)..."

The 'greengrok' page has some staff help and while they take a day or two they usually get comments posted and sometimes reply to email questions, if you want to ask about the Greenland reports they might respond.

Steve Bloom said...

Here's a different type of extreme event from earlier in the year. 93F surface waters sound awfully hot, but I have no idea how it fits with prior such events.

Michael Tobis said...

The question here is not to find scary events; it is to put them into context with pre-AGW scary events.

Steve Bloom said...

Right. It sounds like that may be tough for SSTs.

Steve Scolnik said...

The following excerpt from a conference summary in the Bulletin of the AMS may be of interest, particularly the last bullet:

• Do extreme events have different physics than normal events?
• Are there “dynamical ceilings” that limit extreme events and prevent the reliable extrapolation of statistical results?
• To what extent, and how far in advance, can extreme events be predicted either deterministically or statistically?
• When can we say with confidence that an apparent increase in the occurrence of extreme events is due to a shift in the underlying system?

These and other questions were the theme of a workshop convened in 2007. The proceedings of the meeting are available online (www.soest.hawaii.edu/
PubServices/AhaHulikoa.html).

Vinny Burgoo said...

MT: Vinny, I don't understand the coat-trailing complaint (nor do I understand the analogy, but never mind that). I appreciate your discovery. I learned something. You learned something and taught it to us. Is that a problem?

Sorry, MT. Half-cocked jocularity, that's all it was. Hitting the wrong note after a long evening.

Hank Roberts said...

as noted at kk's:

"... Research has suggested that the persistence of blocking ridges in the upper atmosphere will increase in a doubled carbon dioxide situation (2xCO2) (Lupo et al. 1997)”*
*Lupo AR, Oglesby RJ, Mokhov II (1997) Climatological features of blocking anticyclones: a study of northern hemisphere CCM1 model blocking events in present-day and double CO2 concentration atmosphere.Climate Dynamics 13,181–195. doi:10.1007/S003820050159

Vinny Burgoo said...

Lupo has published several papers on blocking trends since 1997. His latest published update offers only partial support for his 1997 hypothesis. After several decades with no significant long-term trend in the NH and a downward trend in the SH, the frequency of blocking events in both hemispheres increased during 2000-2007. He doesn't ascribe this recent increase to more CO2, saying only that it is consistent with his 1997 paper. However, that paper also predicted that blocking events would become weaker and more persistent and there has been no appreciable change in either strength or persistence in either hemisphere since 2000.

See 'The Global Increase in Blocking Occurrences', a conference write-up, available from this page (which also links to updated archives of blocking events in both hemispheres):

http://solberg.snr.missouri.edu/gcc/

(Tony Lupo is a Republican and a member of the NRA. A conservative climate scientist! MT's right. Something weird is happening to the world.)

Hank Roberts said...

> http://solberg.snr.missouri.edu/gcc/
Thanks, good find!

That's an amazing collection -- both presentations written for Heartland/Icecap and quite an extensive collection of blocking/climate papers going way back.