Claim: 'climate change' caused more deaths in Stockholm – but it may be due to flawed methodology

This is one of those publications where I look at what was done in the paper and just shake my head in disbelief. For starters, according to the data listed in the SI, the supposed extra deaths due to climate change manifesting itself as increased summer temperatures came from model output; they didn’t actually have health services data/coroner data that showed causes of death. They simply assume the model output is valid. And there are other problems, such as their choice of temperature base period of 1900-1929 to compare against the study period of 1980-2009. See more at the end of the post, I need some reader assistance – Anthony

Press Release from Umeå University

Climate change increased the number of deaths

[2013-10-21] The increased temperatures caused by ongoing climate change in Stockholm, Sweden between 1980 and 2009 caused 300 more premature deaths than if the temperature increase did not take place. In Sweden as a whole, it would mean about 1,500 more premature deaths, according to a study from researchers at Umeå University published in the journal Nature Climate Change.

Global warming does not only give a general increase in temperature, but it also increases the frequency, intensity and duration of heat waves. Previous studies have shown that these changes are associated with increased mortality, especially during extremely hot periods. It also speculated that mortality associated with extreme cold could decrease as a result of a warmer climate.

Researchers at the Department of Public Health and Clinical Medicine, Umeå University, conducted a study in which they examined the extent to which mortality associated with extreme temperatures occurred in Stockholm during the period 1980-2009. In order to assess what can be regarded as extreme temperatures, they compared temperature data from this period with the corresponding data for the period 1900 to 1929.

The study shows that the number of periods of extremely high temperatures increased significantly over the period 1980-2009, all of which contributed to about 300 more deaths during these heat waves than had been the case without climate change.

“Mortality associated with extreme heat during the relevant period was doubled, compared to if we had not had some climate change,” says Daniel Oudin Åström, PhD-student in Occupational and Environmental Medicine, who conducted the study.  “Furthermore, we saw that even though the winters have become milder, extremely cold periods occurred more often, which also contributed to a small increase in mortality during the winter.”

Although the increase in the number of deaths due to extreme temperature overall is quite small over a 30 year period, Daniel Oudin Åström emphasises that the current study only includes the Stockholm area. If the method had been used in the whole of Sweden, or Europe, the increase in the number of deaths would have been much larger. For Sweden as a whole, it is estimated that about 1,500 extra deaths due to climate change had occurred over the past 30 years.

In addition, the researchers only examined mortality in really extreme temperatures. Therefore, the number of premature deaths caused by less extreme temperatures is not included in the study.

Daniel Oudin Åström says that despite the long-standing debate about climate change, Swedes have not changed their attitude and willingness to protect themselves against extreme temperatures.

“The study findings do not suggest any adaptation of the Swedes when it comes to confronting the increasingly warmer climate, such as increased use of air conditioning in elderly housing,” says Daniel Oudin Åström. “It is probably because there is relatively little knowledge in regards to increased temperatures and heat waves on health.”

###

Here is the paper:

http://www.nature.com/nclimate/journal/vaop/ncurrent/full/nclimate2022.html

Attributing mortality from extreme temperatures to climate change in Stockholm, Sweden

Daniel Oudin Åström, Bertil Forsberg, Kristie L. Ebi & Joacim Rocklöv

Nature Climate Change (2013) doi:10.1038/nclimate2022 

Abstract:

A changing climate is increasing the frequency, intensity, duration and spatial extent of heat waves. These changes are associated with increased human mortality during heat extremes. At the other end of the temperature scale, it has been widely speculated that cold-related mortality could decrease in a warmer world. We aim to answer a key question; the extent to which mortality due to temperature extremes in Stockholm, Sweden during 1980–2009 can be attributed to climate change that has occurred since our reference period (1900–1929). Mortality from heat extremes in 1980–2009 was double what would have occurred without climate change. Although temperature shifted towards warmer temperatures in the winter season, cold extremes occurred more frequently, contributing to a small increase of mortality during the winter months. No evidence was found for adaptation over 1980–2009

===============================================================

More than a couple of things stand out that I’m looking into.

1. A paper they cite by Pat Michaels and Chipp Knappenberger found only one US city that had any mortality increase due to heat, and that was Seattle. Michaels opines that this was likely due to the city being such a cool climate that very little cooling infrastructure was in place in the city. This might also be true of the high latitude city of Stockholm.

2. A cursory check of climate data for Stockholm from NASA GISS shows that something curious happened around 1930. Notice the big step change then:

Source: http://data.giss.nasa.gov/cgi-bin/gistemp/show_station.cgi?id=645024640000&dt=1&ds=12

Note also how much more variance there is after 1930. To me this looks like a classic station move signal, though it could be related to something as simple as a building going up/torn down nearby that affected wind patterns near the station. The fact that they use 1900-1929 as the base period for the model comparison is troubling, since it seems to be the coolest, least variable part of the station record.

Also, for some reason, GISS can’t seem to get data updated for Stockholm past 1994, even though the station continues to produce data. I’ve asked Dr. Gavin Schmidt about this, but he has ignored my request. Perhaps he’s too busy on Twitter to bother.

Waymarking notes of the station:

“When the observatory was renovated and extended in 1875 the thermometer was moved to a metal cage outside a window on the first floor. The current observation site, from 1960, is only about 10 metres away. These few small relocations make Stockholm’s long observation series one of the world’s absolute best. The high quality of the series has recently been documented in several scientific studies.”*

*From Swedish Meteorological and Hydrological Institute SMHI website.

With it being supposedly “…one of the world’s absolute best.” you’d think NASA GISS would want to get current data for it. It’s a travesty they have not updated it since 1994:

Source: http://data.giss.nasa.gov/cgi-bin/gistemp/find_station.cgi?dt=1&ds=12&name=stockholm

3. The authors cite the shift in temperature distribution during summer as being proof of more heat which would translate into greater mortality(see figure 2 below).

Figure 2: Temperature distribution of 2-day moving average of mean temperatures during summer months.

Grey distribution, 1900–1929; black distribution, 1980–2009.

Problem is, this data they are plugging into their mortality model appears to come from a single weather station, what I believe is the Stockholm Observatory, though they don’t actually name the station dataset in the paper that I’ve found. The Stockholm Observatory has all sorts of microsite issues that they have not accounted for, such as a brick building nearby and wind shading from rows of vegetation.

Image from Waymarking.com, taken July 30th, 2010 – more here

Here is the aerial view from Google Earth using the lat/lon provided by Waymarking.com You can see how wind sheltered the station is, especially during summer with all those broadleaf trees around it. One wonders what the site looked like in 1929 and if the weather station was in the same location.

4. There’s other cherry picking going on; they cite Stockholm as being representative of the changes in Sweden, yet study no other cities or stations to test that theory. They are using mean temperatures, rather than looking at Tmax. Mean temperatures are sensitive to effects of microsite bias which mostly show up in Tmin. If heat waves are really increasing in Stockholm, affecting mortality, it should show up in Tmax, yet they didn’t test for this that I can find.

I think this paper is seriously flawed because the authors assume the temperature data is “near perfect” and chose an inappropriate base period which exacerbates the comparison differential. Whether this is incompetence or cherry picking remains to be seen.

I’m working on locating metadata for a detailed history of the station in Stockholm to test out what I have observed, but I need help.

Anyone reading who is familiar with the station and the meteorological service there, I ask that you weigh in with a comment below. I need the help since I’m not well versed in Swedish. Any help will be appreciated.

Related articles

• Stockholm heat toll ‘doubled in 30 years’ (climatenewsnetwork.net)