COLD SNAP


Tucson recently went through a cold spell, where for 3 days the highs were under 50 F (10 C) and two record lows occurred, the coldest being minus 8. C.  It has been about 40 years since we had 4 straight days under 10 C.

A letter appeared in the paper calling it “global cooling.”  It is a fact that there were no  letters to the editor stating “global warming” when the National Weather Service announced two weeks ago that 2012 tied 1989 as the hottest year on record in Tucson.  Why would three cold days, not a record, constitute global cooling, yet the hottest year on record not be global warming?  For that matter, while the western US is very cold now, the eastern US is warm.  This is typical of jet stream waves to produce such a pattern.  I don’t expect everybody to understand the patterns of the jet stream; however, those who do not understand those patterns should not be writing letters saying this is global cooling, and the newspaper should not be publishing those letters.

A friend, who lives on the East Coast, well educated, although not in science, commented when I visited that 5 days in a row that Philadelphia didn’t break 40 F (5 C) was evidence against global warming.

The letter, the comment from my friend, and those who say that 3 feet of snow in their backyard is evidence against global warming exemplify that too many do not know the difference between weather and climate.  Many are so convinced that there is no global warming that they seize upon any cold weather event and ignore the warm weather ones.  We all are capable of this heuristic:  we hear about something crazy happening, and we say “There must be a full Moon.”  The times the Moon isn’t full (23 days a cycle, since 6 days it is 90% or more illuminated and looks full) are ignored.

We are warmer in June than we are in January.  The weather service has a line trending upward for the average highs and lows during this period.  But we all know that hot and cold days occur.  Once, about a decade ago, April was cooler than March in Tucson.  That had only happened once before in over 100 years.  Yet, nobody would consider thinking that May and June would be cooler, too.  I fail to understand then why a few cold days are treated as evidence against a long term, unheard of in our history as a species, trend.

We know that temperatures change in a jagged fashion trending upward in the early part of the year and trending downward at the end of the year.  One year, fewer than a quarter of Harvard seniors knew why.  This lack of a basic fact taught in elementary schools (my father 60 years ago authored textbooks that included this issue) suggests we have major flaws in educating our children about science.  I am further puzzled by the anger of many of the comments, which are often aggressive with frequent, irrelevant personal attacks.  Why the anger? I consider climate science a matter of models, of ocean level, pH and [CO2], ice core analysis, atmospheric aerosols and other variables, which can be checked for validity.  Al Gore, cap and trade, and so-called radical environmentalists are irrelevant.  The Earth requires storms to balance heat between the equator and poles; a warmer Earth will require frequent, more powerful storms.  I predict this will in fact happen in this century.  I may be wrong, but those on the other side offer very few predictions as to what they think will happen.  If we aren’t truly in the throes of global climate change, then I would like a prediction of the Earth’s temperature range during the next decade as well as to sea level, which currently is rising 3 mm a year. (due to glacier melt and expansion of warmer water).  I want a prediction with confidence intervals, since all predictions are prone to error, and the error must be quantified.

One way to look at this is to split the data in half, from 1930-1970, leave out 1971, a unique middle, and then look at 1972-2012.  We compare 1930 with 1972, 1931 with 1973, and so forth.  This is the Cox-Stuart Test, and it does not require the data to be normally distributed.  It is a non-parametric statistical test, and it is both easy and useful to apply.

1930-1972      67.3         67.9

1931-1973      67.2         66.4

1932-1974      66.4         67.3

1933-1975      66.8         66.3

1934-1976      69            68.3

1935-1977      66.5         69.4

1936-1978      67.9         69

1937-1979      67.2         68.3

1938-1980      67.5         69.4

1939-1981      68            69.6

1940-1982      68.9         67

1941-1983      67.3         67.8

1942-1984      68.5         68.1

1943-1985      70.2         68.7

1944-1986      67.4         70.3

1945-1987      67.2         69.2

1946-1988      68            70.5

1947-1989      68.1         71.4

1948-1990      67.8         69.6

1949-1991      67.3         69.1

1950-1992      69.4         69.8

1951-1993      68.3         70.2

1952-1994      68.2         70.9

1953-1995      68.5         70.3

1954-1996      70.6         70.6

1955-1997      67.4         69.8

1956-1998      68.8         68.1

1957-1999      69            69.6

1958-2000      68.9         70

1959-2001      68.2         69.7

1960-2002      67            70

1961-2003      67.1         70.7

1962-2004      68.2         69.3

1963-2005      68.6         70.8

1964-2006      66            70.2

1965-2007      67.1         70.5

1966-2008      67.4         70

1967-2009      68.1         71.2

1968-2010      68.1         70

1969-2011      68            69.9

1970-2012      67.3         71.4

We begin by assuming there has been NO change in the temperatures, since the status quo is usually taken to be accurate. If we find a change, which we will, we may determine the probability that change is random or is unusually high or low.  Therefore, we set the probability of 50% that one pair will be warmer/colder than the other.  If there has been no change in the past 82 years, then we treat the data as a simple coin toss 41 times.  We would expect 20-21 heads, but we would not be surprised if we had 22, 20, or frankly anywhere between 14 and 28, if we allowed chance intervals to have a probability of at least 5%.  “Anything is possible” is not an appropriate statement here.  I would consider a probability of  0.000001 or 0.049 unlikely, and that is typically where we draw the line in statistics, at 0.05.  Anything outside this range is considered not to be a chance event.  We define what is possible before we collect the data, so we don’t change around the probabilities to call something significant or not significant because it suits us to do so.

There were 32 instances where the second half was warmer than the first, 7 where it was cooler, and 2 ties.  Could this be a chance event?  Yes, it could, but the probability of its occurring if chance were operating here is far less than 1 in 10,000, much smaller than our 0.05 probability we stated at the outset, so we say this is not a chance event, and in fact the temperatures are increasing.  True, the increase is not linear, but we wouldn’t expect annual temperatures to be increasing linearly any more than we would expect the temperature between January and June to increase linearly, either.  That is one of the biggest concerns I have about people talking about a “cool year”.  There is natural variability, or common cause variation, and there is other variability, called special cause variation.  In Tucson’s case, it is likely that there is both heat island effect and global climate change causing the changes.  It is a fact that Tucson is becoming hotter, and this was just proven.  This being a  chance event is not only highly unlikely,;we defined ahead of time what that chance event interval must contain, and the value we obtained, the test statistic, was outside that interval.

To summarize the data in one more way:

  1. Six of the last 10 years ranked in the top 10 for warmest.  That is a fact.
  2. Since 1995, there have been 6.7 times as many high temperature records set as low temperature records (last year’s ratio was 18  to 2).  I counted them.  There were 161 record highs, 24 record lows.
  3. 3.  Last year, 7 months were in the top 10 for warmest; over the past 15 years. about a third have been in the top 10 for warmest.  I counted them.  
  4. 4. The probability in 2013 of Tucson’s being warmer than normal is greater than 99.9%, since there is a 14 year streak, and that would be expected to occur 1 in 16,384 times if it were random.  Since climate is not completely independent (one year’s temperatures may affect another’s) I decreased the probability slightly.

These are facts.  They are on the NWS Web Site .   There is no question among scientists that heat island effect has raised city temperatures in general and affected rainfall as well, and that by definition is manmade climate change. I’m stating facts.

For January in Tucson:

  1. Since 2003, five January average temperatures have been in the top 10.
  2. None has been in the bottom 10, the coldest January since 1934 was in 1937 with the average temperature 41.3.
  3. As of 16 January, Tucson’s average temperature is more than 43 F.  We are now in a warming trend, and long term models out the next 9 days suggest this will continue, although not be excessively warm.  Tucson will almost certainly not set a record low for January; it may yet be in the top 10 coldest, but that is by no means certain. What is clear is that this January is nowhere near a record OR a trend, for one cannot call a trend from a single point.

I mention these data, because too many seize upon a cold day or a freak snowfall to make their case, and few seem to challenge them. Others think that global warming means every spot on the Earth is getting warmer every year.  It doesn’t.  Publishing letters like that one at the outset allows obvious misperceptions to see the light of day, which they should not.  Precipitation patterns will change in a warmer world; in winter, that will take the form of snow, so heavy snow does not imply a cooler world.  Australia’s temperatures, now over 51 C in places (122 F.), are exactly what was predicted.  The shrinking of the Arctic Ice Cap was predicted.

I wish to be clear: the Earth is a greenhouse.  This is basic science; it explains why Venus is warmer than Mercury, although it is almost twice the distance from the Sun.  Were it not for water vapor and carbon dioxide, we would not be able to survive as a species.  We have changed the concentration of both key gases, as well as having a worrisome contribution from methane, should the permafrost melt, which I have seen in Alaska.  This is also a fact.  If anything, climate models appear to under-predict the severity of ice melt.  We have known acidification of oceans, because they buffer rising CO2 by absorbing it in the surface. The ocean rise will become a huge problem later this century, and warmer water will increase water vapor, the biggest greenhouse gas of all, causing a positive feedback loop.  These are all facts.  Whether specific extreme weather events are due to global climate change is not always clear (the Texas drought was caused, with 95% confidence, by climate change; Bangkok flooding was not), but the trend would certainly appear to be that way.

All this January cold spell will do is make it less likely that in 2013 Tucson will have the warmest year on record.  I predict with 99.9% probability 2013 will still be warmer than normal.

Those who deny climate change may be correct, but they must produce clear science from refereed journals (or refereed Internet sites) including confidence intervals or probability of error.  The IPCC is 95% confident, an admission that they may (unlikely) be wrong, because there is necessarily a degree of uncertainty in predicting the future behavior of a complex system.  I am clear what it would take for me to change my mind: a downward trend of global temperatures and Arctic ice melt through 2020, and the end of sea level rise.

I would request those who disagree to say what would change their mind.  If the answer is “nothing,” then the issue is no longer about science but faith or ideology, for uncertainty must exist about this issue. If Tucson had two consecutive years with below normal temperatures and normal rainfall resumed, I would reconsider my views.  We are in a severe long term drought with a 2 year deficit in rainfall in the last decade.  The normal rainfall of 12 inches was changed to a new “normal” of 11.5, and the annual average for the past 19 years has been fewer than 10 inches (fact).  That is why I call it climate change and not global warming; it’s more than temperature that is changing.

I am glad I am old and have no children to ask me, “Why didn’t you do something about climate change, Dad?  You knew it was occurring, and you could have changed the direction in your lifetime.”

Writing a letter, or even a column, as I did for the Medical Society (getting many negative comments from doctors), or posting on my blog seems like a pretty weak answer.  “I’m sorry” is even weaker.  And “I told you so, but you wouldn’t believe me” doesn’t help, either.  I hope I am wrong, but the probability of that is unlikely, and I know how unlikely.

Note on 21 January:  We are now looking a record heat for 2 days this week and the first 27 C (80F) day about 3 weeks earlier than normal.  The current January temperature is now 10th coldest and may remain below normal, although that remains to be seen.  I do not expect to see letters about the end of “global cooling”.

 

Final note:  January was 2.8 F below normal.  This is about in the 33rd percentile for January.  While the first half of the month was the 4th coldest, the second half was the 10th warmest.

Note for comments:  I went to great pains to try to avoid pejorative statements and to remain factual where there are facts and probabilistic where there were not.  Since I have the ability to delete comments, I will delete those that personalize the issue, which fail to meet my standards of decency, fail to show sound science, using peer-reviewed scientific journals, with p-values, confidence intervals, and models that show disagreements. Trend lines are unacceptable as proof of warming or cooling unless accompanied by regression diagnostics that show validity of the assumptions of normality and equal variance.  It is  expected that these terms are understood by the commenter.  If that sounds like a tall order, it is.  

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


%d bloggers like this: