My post on the Hay Springs surface station a few days ago has attracted the attention of Eli Rabbett. I was drawn to this particular surface station by a post by Anthony Watts about the microsite bias that would be introduced at this site due to siting issues. I noticed that immediately after the surface station switched to a MMTS thermometer, there was a jump in the temperature readings at this station compared to surrounding stations. I attributed that to the MMTS device; Eli disagreed. What made this even more interesting is that the Hay Springs surface station is considered to be rural, which means that the temperature trend at this location is used to calibrate those around it. In fact, Hay Springs is one of the 272 best surface stations.

In the comments of my previous post on Hay Springs surface station, Eli points out that surrounding stations switched to MMTS thermometers at around the same time. He says,

Alliance got an MMTS in 86, Lodgepole in 89, Bridgeport in 89, Rapid City in 83, Wray in 84, Hettinger in 88, and you get the idea. So whatever you have found is not a systematic effect from the MMTS since all the other stations were MMTS

That’s interesting, but did not disprove my points. It is certainly possible that all of the surrounding stations had the switch to MMTS either not corrected or incorrectly corrected. I should point out here that due to a stupid programming error, the stations I used in the previous post were not the closest 10 stations. They were all within 500 km, and the updating of the software does not change the conclusions of the previous post at all.

I decided to make a very simple model of how the MMTS switch could manifest itself in my plots if none of the stations had been corrected. To do this, I made use of the Multi-network Metadata System (MMS). For each of the ten surrounding stations, I found the date that the surface station made the switch to the MMTS thermometer.

• Hay Springs: August 1989
• Alliance: December 1986
• Merriman: June 1990
• Harrison: August 1987
• Bridgeport: August 1989
• Hot Springs: February 1997
• Torrington: N/A
• Lusk: November 1986
• Lodgepole: August 1989
• Kimball: May 1987
• Cottonwood: August 1986

At first glance it would appear Eli is right. Most of the stations around Hay Springs switched to MMTS at around the same time. What does the litereature say about the MMTS adjustment? Qualye et al. (1991) showed that the average maximum temperature measured by the MMTS was -0.4C (-0.72F) lower than the previous thermometer. Doesken and McKee (1995) reported that the MMTS read about 0.5C cooler in the maximum temperatures. They reported no change in minimum temperatures.

Pielke Sr., et al. have this to say about the MMTS adjustment of Quayle:

In the mean of many stations over broad areas, the conclusions of Quayle et al. (1991) and the subsequent procedure for adjusting temperature time series to accommodate the change from liquid-in-glass thermometers may be appropriate. But, at any given station, a single adjustment factor may not be appropriate

But let’s assume that when each station switched to a MMTS, it introduced a +3F bias in its maximum temperature measurement. Note that this adjustment has the opposite sign as suggested by Quayle et al. and Doesken and McKee. What would this look like in our data for these 11 stations. Remember that the way I defined the regional average was just the average of the anomalies at the 10 surrounding stations. That means that the MMTS switch due to this station will be 10 times as large.

(Click for full size)

The highly variable thin black line is the difference between the temperature at the Hay Springs surface station and the regional average. The red thin line represents the moving average of the black line. But the important point of this figure is the think purple line. It is the amount of bias introduced into this analysis based upon the simple assumption that all the MMTS switches had a positive bias of +3F.

Therefore, when the surrounding stations switched to the MMTS thermometer, the purple line will decrease in this plot. But it will be a small decrease of only 0.3F. When the Hay Springs surface station switched to the MMTS, it shows up as a large increase in temperature. The change is not 3F because two surrounding stations also changed to MMTS the same month.

This simple model does a fair job of describing the difference between the local temperature at Hay Springs and the surrounding regional temperature. It does not do a good job of describing what happens after Hay Springs has switched to the MMTS thermometer. While they might be interesting, describing these variations after the Hay Springs surface station switched to the MMTS is beyond the scope of this post.

Conclusions

This simple model seems to demonstrate that the Quayle MMTS adjustment is incorrectly applied or not applied at all to the Hay Springs as surrounding surface stations. This was shown using a simple model that assumed a positive bias in the MMTS as opposed to the negative bias of Quayle. Each station will have its own bias (positive or negative) that will be different that its neighbors, and will need to be accounted for individually.

References:
Doesken N.J., and McKee T.B., 1995: Colorado Climate water-year series (October 1994-September 1995). Climatology Report 95-1. Department of Atmospheric Science. Colorado State University, Fort Collins, CO; 147 pp.

Pielke, R.A. SR, T. Stohlgren, L. Schell, W. Parton, N. Doesken, K. Redmond, J. Moeny, T. McKee, T.G.F. Kittel, 2002; Problems in Evaluating Regional and Local Trends in Temperature: An Example from Eastern Colorado, USA. Int. J. Climatol., 22, 421–434.

Quayle, R.G., D.R. Easterling, T.R. Karl, and P.Y. Hughes, 1991: Effects of Recent Thermometer Changes in the Cooperative Station Network. Bull. Amer. Meteor. Soc., 72, 1718–1723.

(No Ratings Yet)

 Loading ...

Related Posts:

Trackback URI |