Short Splice, Long Splice

Guest Post by Willis Eschenbach

When I was a kid I had the great fortune to be taught to splice rope by a grandson of Richard Henry Dana. He taught me how to do a long splice and a short splice … but I never learned how to do either the long or short data splice. For that, I had to wait for climate science.

Anthony recently highlighted a paper about reconstructed Greenland temperatures called “Past 4000 Years of Greenland Temperatures“. I don’t really have a dog in the fight, but I was greatly amused by their description of what they call “in situ observations”.

Here’s the top panel of their Figure 1.

Figure 1. ORIGINAL CAPTION FROM THE PAPER (top) Reconstructed Greenland snow surface temperatures for the past 4000 years and air temperature over the past 170 years (1840–2010) from three records. The thick blue line and blue band represents the reconstructed Greenland temperature and 1s error, respectively (this study). The reconstruction was made by two different methods before and after 1950. The “gas method” is as described in section 2, and the “forward model” is described by Kobashi et al. [2010]. Thick and thin black lines are the inversion‐adjusted reconstructed Summit annual air temperatures and 10‐year moving average temperatures, respectively [Box et al., 2009]. Thin and thick red lines are the inversion adjusted annual and 10‐year moving average AWS temperature records, respectively [Stearns and Weidner, 1991; Shuman et al., 2001; Steffen and Box, 2001; Vaarby‐Laursen, 2010]

I thought “hmmm, data from the summit of the Greenland ice cap, didn’t know about that” … so I went to find out what their data consists of, the famous “thin red line” in the Figure 1 above. Here’s what they say about the “in situ observations”:

5.1. Present Greenland Temperature

To place the Greenland temperature proxy reconstruction into a historical context, we incorporate two additional Summit temperature records. One record is obtained from a compilation of Summit Automatic Weather Station ∼2 m surface air temperature (SAT) observations (hereafter AWS or in‐situ record) that spans 23 years (1987–2010). The AWS were situated within 20 km of the GISP 2 coring site and within 25 m elevation of the ice sheet topographical summit (Figure 1, top, red line).

The series begins in May 1987 with Automatic Weather Station data after Stearns and Weidner [1991].

Shuman et al. [2001] merge this record with data from the Greenland Climate Network (GC‐Net) AWS data [Steffen and Box, 2001] to produce the first 12 years of this compilation.

Gaps before June 1996 are in‐filled using daily passive microwave emission brightness temperatures.

GC‐Net data then comprise the period spanning June 1996 to December 2003 with gaps in‐filled by Danish Meteorological Institute (DMI) Summit AWS data [Vaarby‐Laursen, 2010].

The DMI data exclusively form this data series from January 2004 through December 2010.

Man, I thought, that is a curious provenance for the summit temperature data. First, 12 years of observations, with gaps in-filled using satellite microwave observations. Then from 1999 to 2003 the summit temperature was estimated using surrounding stations. Gaps in the surrounding station data are infilled from the DMI Summit station data. Then from 2004 on, we have DMI summit station data only.

So I looked a bit deeper. It gets better. Remember the Shuman et al. observations that form the “first 12 years of this compilation”? It turns out that it was not one station, but three stations … and there was no overlap between the stations to compare temperatures. The Shuman paper is here. Their Figure 2 shows those first 12 years of the record, from three AWS stations: CATHY, GISP2, and SUMMIT.

Figure 2. Automated Weather Station records near and at the Greenland Summit. 

Note the lack of overlap and the number of gaps, covering days, weeks, or months. Here’s Shuman’s description of what they did:

In order to complete a temperature record from the Greenland Summit (May 1987 to October 1999), it will be necessary to adjust the AWS Cathy temperature record to account for the difference in its location as well as to complete the AWS GISP2 and AWS Summit records across multiple periods of missing data .

The methods detailed in Shuman et al. (1996) or in Shuman et al. (1995), which rely on appropriately located and contemporaneous SSM/I brightness temperature data (Table 1), will be used to achieve a complete and consistent temperature record.

Inconveniently, SSM/I data are not available from 4 May 1987 to 10 July 1987 and are missing from 3 December 1987 to 13 January 1988 (Table 3). Smaller gaps in the SSM/I or AWS record of less than 5 days will be dealt with by interpolation.

So for the 23 year record we have a long data splice as follows:

2 years of AWS CATHY data,

7 years of AWS GSIP2 data,

3 years of AWS SUMMIT data,

5 years of estimated data, and

6 years of AWS SUMMIT data.

All of this is “infilled” from a couple of sources.

Now, remember that this data is what they are using to calibrate their algorithm that converts ∂O18 data into temperature data …

The authors of the study cite Ellen Vaarby-Larson 2010, which is here. It shows the Summit temperature for the period 1998- Feb 2010:

Figure 3. Air temperature from the 04416 Summit AWS station. From Vaarby-Larson 2010.

Note that even in the modern period there are gaps of up to about a year in the record. (Also of note is that there are only about six three-hour periods in the record where it is above freezing, and that the summer/winter spread is about 60°C [108°F]. Yikes!) Here’s an example of why the record contains so many gaps and spaces:

Figure 4. Condition of the Summit station 04416 during the 2007 annual visit. From Vaarby-Larson 2010

Let me conclude by looking at some other problems endemic with AWS records. I cannot improve on the words of Ms. Vaarby-Larson, who said (op. cit.):

Recommendations

Generally, great care should be taken when using the observations from station 04416 Summit, since the observations are influenced by:

• Extreme climatic exposure of the station measurement equipment (the extreme cold causing e.g. low availability of wind observations during winter)

• Non static barometer elevation above sea level, due to the Greenland Ice sheet flow-patterns

• Non static height of measurements above ground, due to burial of the station by snow falling

None the less, the observations of temperature, humidity and wind (see Figure 10 – Figure 16 ) show no great, obvious shift in level or variance, at the individual station relocations in 2005, 2006, 2007 and 2009.

Any existing bias due to change in measuring height above ground, might be difficult to identify since:

• 04416 Summit only issues synop every three hours.

• There’s no direct measurement of the ongoing change in actual measuring height above ground, in-between station relocations.

• Change in weather conditions might produce the same signal in the observations

• Natural variability shows e.g. great temperature variance during winter compared to summer, please confer with the early attempt at investigation of temperature variance in Figure 17.

Hmmmm …

In a recent post called “A Modest Proposal—Forget About Tomorrow“, I discussed a paper that showed why even perfect “fitted” or “calibrated” models may not have predictive (or reconstructive) capabilities. In the current case, they are reconstructing the Greenland temperature based on a collage of actual observations, nearby observations, estimates, and satellite microwave brightness.

Let me be clear that there is nothing inherently wrong with putting together a pastiche composed of a variety of estimates and local observations and satellite data. There may even be something to learn from such a collation of disparate elements.

My point is that it doesn’t engender confidence when said pastiche is used to calibrate an algorithm designed to transform an ice core ∂O18 record into a temperature reconstruction. Even with the best of data, that’s a tough sell.

My conclusion? Only have one.

The confidence intervals in the original paper on historical Greenland temperatures are way too narrow.

w.

PS—Ya gotta love the Google Earth image of the Greenlad Summit temperature measuring site (Station ID# 04416), found here:

When I first looked that that I thought “They left out the satellite part of the image” … then I realized that they hadn’t left anything out. Anthony, I do believe we’ve actually finally found a truly rural temperature station …