July 22, 2013

Are Nighttime Heat Waves Increasing in the Northwest?

On Saturday, the Seattle Times ran a front page story about how nighttime heat waves are increasing here in the Northwest during the summer.     Based on an article in the Journal of Applied Meteorology and Climate, the story was repeated by media outlets throughout the region.

But the more one looks at these results, the more questions come to mind.  This blog will highlight some of my concerns.

Before I go further, let me note the region they considered was western Oregon and Washington.

Interestingly, the authors found no trend in the extremes of maximum temperature, but only in minimum temperature.  That seems a bit odd in itself, but some studies have suggested this is a general finding (more increase in min than max temps), with the idea that more cloud cover or other effects might be the cause.   Around here, daytime warming is what really put folks under acute heat stress and the "heat waves" examined in this study were associated with nighttime temperature falling to roughly the mid 60s or more for three days or more.  Not very dramatic and wimpy by heat wave standards for the rest of the country.

But the lack of correspondence between serious daytime and nighttime heat waves IS concerning, since most major heat waves around here (like July 2009)  are observed in both maximum and minimum temperatures.   There is a reason for that...our biggest heat waves are associated with strong offshore flow and subsidence down the western slopes of the Cascades, and this phenomenon tends to raise both maximum and minimum temperatures.

But what REALLY bothered me was their plot showing heat wave occurrence over time (see figure displayed in the Seattle Times):

There really seems to be a sudden increase in heat waves around 1990.   That is why their paper and Seattle Times are talking about a trend in heat waves.   Why would heat waves rev up in the years around 1990?   Such a variation is not consistent with human-caused global warming, because that the warming increases  slowly and would not start abruptly like shown in this figure.  Natural variability can cause a sudden warming, but the big Kahuna of multi-decadal variability around here, the Pacific Decadal Oscillation (PDO), switched phase from cold to warm around 1975--which doesn't fit this figure.

So we are left with finding the origin of the sudden increase in nighttime heat waves around 1990.  To explore this issue,   I started playing around at the wonderful web site supported by the Office of Washington State Climatologist (OWSC) that allows you to plot temperature from stations around the region.  And as I plotted station after station a disturbing pattern was apparent:  many stations showed abrupt temperature jumps, from one plateau level to another around 1990.  Let me show you a few examples, displaying summer minimum temperatures.  Try to ignore the red lines, the trend line for the entire period...really is useless.  Try covering the figures before and after the jumps..that really drives the point home.

First, there is McMillin Reservoir near Tacoma..big jump by1-2F around 1986.

 Or Astoria, a jump around 1989.
Everett, a jump in the early 1990s.

 Forks moved to a new, higher plateau around 1989.

So what was producing the jumps to higher plateaus?  Well, it turns out that NOAA made a major change in temperature sensors, from old style alcohol/mercury thermometers in white slotted shelters to electronic sensors called MMTS in plastic housings during the late 1980s and early 1990s..  Here is how they look.

Big difference.   A number of studies have shown that the MMTS sensors tend to read high for minimum temperature..by about .5F.   But there is another issue.  Since MMTS sensors must be wired in, they have generally been moved in closer to buildings and structures than the sensors they replaced.   Here is an example--the MMTS  at  Forks (near a deck!)
 and this one in Conconcully, above hot rocks!

A number of studies have shown the buildings and urbanized settings tend to influence the minimum temperatures more than the maximum temperatures.  For example, rocks, concrete and structures tend to absorb heat during the day and release it at night.  And  the atmosphere is generally more mixed during the day, so local effects tends to be lessened.

NOAA converted many of the climate sites to MMTS between 1985 and 1995.  And my brief examination of the record from each station suggest that many of the jumps were associated with the conversion to electronic thermometers.  For example, take McMillin Reservoir.  The official NOAA metadata (metadata is a description of sensor changes and moves) shows the conversion to MMTS in 1986.  For Everett 1991-1992.  And there was another sensor change at the airports to the HO83 sensor, which had its own biases.

Further proof of the MMTS conversion issue has been noted by Mark Albright, past state climatologists.  He compared McMillin Res against Olympia since Olympia didn't undergo the MMTS conversion and is only 20 miles SW of McMillin Res.  HE  selected 2 years prior to the discontinuity and 2 years following the jump in temperature at McMillin Res for July.  In 1984-85 McMillin Res and Olympia were both equally cold in July.    Something had changed by 1990-91 to make McMillin Res +2.5 deg F warmer than Olympia. 
              July 1984-85       July 1990-91
McMillin Res    49.7                  53.0
Olympia         49.7                  50.5
Difference       0.0                  -2.5

And there is another problem.  A number of the minimum "heat wave" days would not be considered heat waves at all by any reasonable person.   One of their periods started on July 21, 2007.  Take a look at the official record at Seattle Tacoma Airport:

Several days with minimum temperatures in the mid to upper 60s.  But the maximum temperatures are only in the seventies and light rain was observed.   During the 3-day "heat wave" of 21-23 July 2007 at SeaTac Airport, rain fell each day totaling 0.60 inches and the high temperature averaged BELOW NORMAL 72 degrees.  The average high temperature at Olympia was an even cooler 71 degrees. This is no heat wave.

I want to be careful here--my study was a brief one.  But I believe there is real reason to wonder whether this sudden nighttime "heat wave storm" is real.  And to assume the increased frequency of nighttime heat here in the Northwest is due to anthropogenic global warming is triply doubtful.  Mankind will cause substantial warming of the planet later in this century...but I suspect that this nighttime anomaly has a less profound origin.


  1. 1. None of these graphs can visually distinguish between an overall trend and a changepoint. If you have some hypothesis about a particular type of sensor change impacting the readings, you could build a hierarchical regression model to see if that changepoint has any significance above the trend.

    2. The OWSC interactive map states that it is using the adjusted data from USHCN, and the documentation on the USHCN dataset says that they have applied adjustments based on changes in siting and sensor type. I was wondering if you thought it was an issue with adjusting mean temperatures versus daily minimums, but the description of adjustment procedure specifically gives examples of a series of minimums. What do you think is wrong with the adjustment procedure?

  2. Anthony Watts has been making this claim for a long time. Go to http://surfacestations.org/ to review some of his research on weather stations and their siting and how this affects our temperatures.

    Also, if these sitings are changing the low temperatures, just think about how they are affecting the high temperatures.

  3. It would be interesting to see if the low temps jumped for the non summer months as well. Thanks for the analysis. Its a great blog post.

  4. Great post. I would also guess continuing urban sprawl and the resulting "urban heat island" effect is also contributing to high nighttime mins in metro area locations.

    Also, how does the Seattle Times classify their heat waves used in that chart?

  5. Urbanity is hardly sprawling, especially in the Puget Sound area. It grows in small restricted areas in Seattle and hardly anywhere else. 'Suburban sprawl' certainly is sprawling, however (thus the name).

    Semantics aside, the knee-jerk reaction to blame everything on global warming is troubling. It should be fairly obvious to meteorologists that there are many possible reasons that perceived trends exist that can be correlated to many factors. It's manipulative to play upon people's fear of global warming and ignorance of the complexity of the issue to sell a good sounding story.

    It's unfortunate that the Seattle's very own Seattle Times continues to mislead the public on so many issues.

  6. As I read that article, I was having a "Cliff moment"... "Wasn't that around the time they changed the temperature sensors and moved many of them?" I questioned the article as I read it and dismissed it out of hand, figuring the "heat wave increase" was likely to be statistical noise.
    Thanks for confirming my suspicions (and for educating me well over a number of years so that I knew to question the findings in the first place).

  7. LOL - Only in the PacNW would these be called heat waves.

    As for data bias (at least for SeaTac), how about the vast expansion of SeaTac that has occurred since the in 1980s and the relocation of the instruments? I believe Cliff looked into this before.

    Agree very much that AGW is very likely for the planet as a whole, particularly the polar regions. There will be some areas that don't warm or even cool. Given our proximity the ocean and other factors, the PacNW could well be one of the regions that is not impacted. No one really knows.

  8. In the actual paper you link to (and not the Seattle Times article), the authors do go into a fair amount of detail about the changeover from one sensor type to another, even including plots that showed how the identification of these "heat wave" events changed depending on whether or not you accounted for the changes in the sensor type. But I agree that these statistics with small sample sizes demand close scrutiny...

  9. Luke,
    They do mention it and did a small adjustment. But clearly that adjustment is insufficient and I suspect most of it comes from the fact that the sensors were moved closer to building because they require a wire...cliff

  10. Cliff, I am starting to hear a lot of grumbling about too much sun from native Pacific North westerners. Headaches, allergies, general malaise, and I find myself, too, searching the forecasts for when "our" rain will be back.
    Any idea?
    Lori m

  11. Is the UW study peer reviewed?

  12. I feel like this last summer we had a lot of heat waves in Chicago. Our AC broke so we had to rush to hire an air conditioning repair company before we all got heat stroke because of how hot it got. Hopefully it won't be like this next year.


Why do Temperatures Sometime Warm with Height? Inversions Explained. Plus the Weekend Forecast!

During the autumn, the Northwest is often visited by a strange condition: temperatures WARMING with height.  These features are called inv...