Polar Amplification: Observations versus IPCC Climate Models
Guest post by Bob Tisdale
We’ve illustrated and discussed polar amplification in a few posts in the past. See here and here. Wikipedia has a short blurb about it:
Polar amplification is the greater temperature increases in the Arctic compared to the earth as a whole as a result of the effect of feedbacks and other processes[1] It is not observed in the Antarctic, largely because the Southern Ocean acts as a heat sink and the lack of seasonal snow cover.[2] It is common to see it stated that “Climate models generally predict amplified warming in polar regions”, e.g. Doran et al.[3]. However, climate models predict amplified warming for the Arctic but only modest warming for Antarctica.[2]
Many discussions about polar amplification around the climate-related blogosphere have similar definitions, leading readers to believe polar amplification is a phenomenon that only occurs in a warming world. But if we divide the trends of the global surface temperature anomaly data since 1917 into its cooling period (1944-1976) and two warming periods (1917-1944 and 1976-2011), and present the surface temperature linear trends on a zonal-mean (latitudinal) basis, Figure 1, we can see that polar amplification works both ways. That is, during a period when global temperatures cool, like 1944-1976, there is greater cooling in the Arctic than elsewhere. Note also that, according to the GISS Land-Ocean Temperature Index (LOTI) data, the rate at which the Arctic warmed was higher during the early warming period (1917-1944) than it has been during the current warming period (1976-present).
Figure 1
If you’ve never seen a zonal-mean plot before, they’re not difficult to understand. The y-axis (vertical) is temperature in deg C, just like a time-series graph. But the x-axis (horizontal) is latitude, with the South Pole to the left at -90 degrees and the North Pole to the right at 90 degrees. In Figure 1, and in the other zonal-mean graphs in this post, we’re illustrating trends in deg C per decade. Note: I have not included any data south of 75S (the Antarctic) because the data there starts in the 1950s and it is sporadic early on.
Many of you will find it odd that global surface temperatures warmed at such similar rates during the early and late warming periods—especially when we consider that the net effective forcings during the late warming period rose at a rate that’s about 4.5 times greater than during the early warming period. See Figure 2. The GISS net forcing data is available here.
Figure 2
That’s one of the ways the surface temperature record contradicts the hypothesis of anthropogenic global warming. According to the net effective forcing data (and the model simulations presented later in this post), the rate at which surface temperatures warmed during the late warming period should much higher than during the early warming period. But it’s not. The two periods warmed at similar rates. This was discussed in more detail in Section 2 of my book. In fact, Figure 2 above is an updated version of Figure 2-17from the book.
HOW WELL DO THE IPCC’s CLIMATE MODELS HINDCAST AND PROJECT POLAR AMPLIFICATION?
One-word answer: poorly. I’ll leave it up to readers to come up with a two-word answer.
We’ll compare linear trends (deg C/decade) of the GISS Land-Ocean Temperature Index (LOTI) data and the simulations of global surface temperature by the multi-model ensemble mean of the CMIP5-archived coupled climate models that have been prepared for the upcoming 5th Assessment Report (AR5) of the Intergovernmental Panel on Climate Change (IPCC). I’ve used the RCP 8.5 scenario hindcast/projection since it was simulated by the most models and since it is similar to most of the other scenarios during these periods. See Preview of CMIP5/IPCC AR5 Global Surface Temperature Simulations and the HadCRUT4 Dataset. And again, we’ll use the zonal-mean graphs.
As shown in Figure 3, the models do a good job between the latitudes of 30N to 70N of simulating the rates at which global surface temperatures warmed during the later warming period (1976 to present). They underestimate the warming north of 70N and, for the most part overestimate the warming south of 30S. For example, at the equator, the models are hindcasting/projecting warming that is about 1.6 times higher than the rate that’s been observed since 1976 [(0.186 deg C/decade)/( 0.113 deg C/decade)] .
Figure 3
During the mid-20thCentury “flat temperature” period, Figure 4, the models do reasonably good job between the latitudes of 60S-60N, but fail to capture the observed warming of the Southern Ocean and the polar-amplified cooling over the Arctic.
Figure 4
And for the early warming period, Figure 5, the models are not simulating any polar amplification. They missed the boat during this period too.
Figure 5
And for those interested, Figure 6 compares the trends of the simulated warming rates for the three periods on a zonal-mean basis. The brown curve of the early warming period should be similar to the purple curve of the late warming period.
Figure 6
CLOSING
This post was just another way of illustrating that the climate models employed by the IPCC show no skill at being able to simulate the surface temperatures experienced over (nearly) the last century. We’ve illustrated and discussed these failings numerous ways in past posts. Refer to the other IPCC model-data comparisons at my blog or in my book.
MY FIRST BOOK
As illustrated and discussed in If the IPCC was Selling Manmade Global Warming as a Product, Would the FTC Stop their deceptive Ads?, the IPCC’s climate models cannot simulate the rates at which surface temperatures warmed and cooled since 1901 on a global basis, so their failings on a zonal-mean basis as discussed in this post come as no surprise.
Additionally, the IPCC claims that only the rise in anthropogenic greenhouse gases can explain the warming over the past 30 years. Satellite-based sea surface temperature disagrees with the IPCC’s claims. Most, if not all, of the rise in global sea surface temperature is shown to be the result of a natural process called the El Niño-Southern Oscillation, or ENSO. This is discussed in detail in my first book, If the IPCC was Selling Manmade Global Warming as a Product, Would the FTC Stop their deceptive Ads?, which is available in pdf and Kindle editions. A copy of the introduction, table of contents, and closing can be found here.
SOURCE
The modeled and observed surface temperature data presented in this post are available through the KNMI Climate Explorer:
http://climexp.knmi.nl/selectfield_obs.cgi?someone@somewhere
Related articles
- If the IPCC was Selling Manmade Global Warming as a Product, Would the FTC Stop Their Deceptive Ads? (wattsupwiththat.com)
- Another GISS miss: warming in the Arctic – the adjustments are key (wattsupwiththat.com)
- Guest Post Titled “Decadal Prediction Skill In A Multi-Model Ensemble” By Geert Jan van Oldenborgh, Francisco J. Doblas-Reyes, Bert Wouters, Wilco Hazeleger (pielkeclimatesci.wordpress.com)
- Comments On The Paper “Skillful Predictions Of Decadal Trends In Global Mean Surface Temperature” By Fyfe Et Al 2012 (pielkeclimatesci.wordpress.com)
Discover more from Watts Up With That?
Subscribe to get the latest posts sent to your email.
Julienne Stroeve says: “Bob, I’m curious, if you believe the multil-model ensemble mean provides a reliable estimate of the forced component, I could then argue that 50% of the observed decline in Arctic sea ice cover is externally forced. Do you agree with that type of assessment?”
Julienne, you’ve slightly modified what I wrote above. I wrote, To sum up that discussion, we’re illustrating and discussing the model mean because the model mean provides the best representation of the forced component of the climate model ensemble.
Your statement assumes the models are correct. I haven’t made that assumption.
Julienne Stroeve says: “Remember too that the model spread provides a measure of natural climate variability.”
Actually, it does not. The model spread represents the range of the noise of the ensemble. Some of the noise results from the poor portrayal of ENSO and other natural variables by models, but much of the noise is simply that, noise internal to the models.
Julienne Stroeve says: “Thus the timing of ENSO events in the models is not expected to match that observed.”
The modeling community needs to raise their expectations. The reason: the natural variation in the frequency, magnitude and duration of ENSO events dictates the amount of heat released and redistributed from the tropical Pacific and dictates the rate at which temperatures remote to the tropical Pacific vary in response to ENSO-caused teleconnections. During multidecadal periods when the frequency, magnitude and duration of El Niño events exceed those of La Niña events:
1. more warm water than “normal” is released from the Pacific Warm Pool,
2. the tropical Pacific releases more heat than normal into the atmosphere,
3. teleconnections cause surface temperatures outside of the tropical Pacific to warm more than normal, and
4. more warm water than normal is redistributed toward the poles and into adjoining ocean basins.
As a result, global surface temperatures rise. The opposite holds true when the frequency, magnitude and duration of La Niña events exceed those of El Niño events.
Julienne Stroeve says: “And what models did you use in your ensemble mean? How did you assess continuity in the historical and future scenarios?”
The source of the data (KNMI Climate Explorer) is linked toward the end of the post:
http://climexp.knmi.nl/selectfield_obs.cgi?someone@somewhere
Click on the “Monthly CMIP5 scenario runs” in the “select a field” menu toward the right. That brings you here:
http://climexp.knmi.nl/selectfield_cmip5.cgi?id=someone@somewhere
I selected “TAS” under “CMIP5 mean” using the “rcp85” scenario. That should represent all 21 models listed thereafter, which include 60 ensemble members. KNMI provided any merging of the historical and future scenarios. If there had been an obvious discontinuity as there is for sea surface temperature data (TOS), I would have ended the presentation before the shift and noted it in the post.
Julienne Stroeve says: “Arctic amplification is apparent in the observational record and using the GISS data, the amplification of the last decade is greater than at any other time in the GISS record.”
How are you defining Arctic amplification? The reason I ask: The following graph compares running global and Arctic (65N-90N) 120-month surface temperature anomaly linear trends based on GISS LOTI data. The Arctic decadal trend was much higher in the 1920s according to the GISTEMP data than it is today, while the corresponding global temperature trends were slightly lower at that time than they are today.
http://i47.tinypic.com/2a9nvyh.jpg
Should I divide the Arctic trends by the global to represent Arctic amplification?
Regards
wmconnolley: Regarding your April 25, 2012 at 12:55 am comment, I and the majority of the visitors here understand the source for the HADISST sea surface temperature data is very scarce in the Southern Ocean prior to the satellite era and that the HADISST data is infilled. We’ve discussed the dataset a number of times here at WUWT. But just as long as GISS treats it as a spatially complete dataset, I will do the same.
And to help you cast aside your doubts as to whether or not GISS would present Arctic and Southern Ocean surface temperature data as “crudely as [I’ve] done”, I suggest you take a look at Hansen et al (2001):
http://pubs.giss.nasa.gov/docs/2001/2001_Hansen_etal.pdf
Scroll down to page 23, and notice the zonal temperature change (trends) plots below the maps. Similar time periods, similar graphs. They even showed the trends seasonally.
Wanna try again?
armidalehigh62to67: Thanks for buying my book. I’ve started writing a new one just about ENSO and need some direction from readers. When you get to the Section on ENSO (Section 6), note where you think I need to make the explanations more detailed or easier to understand. Where can I add more illustrations–graphs, maps, cartoons? And what did I fail to discuss? I know I need to discuss the differences between minor ENSO events and the major ones, but what else? Should I add cartoons that show the transition from El Nino to La Nina events and vice versa?
Bob Tisdale says: April 25, 2012 at 4:36 am>
Much better, well done. The paper is a good ref, but theirs is a different (longer) period, and they don’t over-interpret it.
> Wanna try again?
Not really. But I’d add that comparing one world to the multi-model mean is probably wrong, too.
The reason IPCC models cannot correctly reproduce polar amplification is that there is no polar amplification. Polyakov et al. attempted to detect it observationally in 2002 and came back empty handed. There is Arctic warming indeed but no thanks to polar amplification as IPCC pseudoscience would have it. Arctic warming is not greenhouse warming at all but is caused by Atlantic Ocean currents carrying warm water from the Gulf Stream into the Arctic. Proof of this is available in this article: http://curryja.files.wordpress.com/2011/12/arno-arrak.pdf . The article also works out the consequences of this fact to the rest of the climate system.
Bob: Hansen and his pals at GISS keep calling theoretically calculated absorption changes in CO2 via concentration a forcing on temperature. This is nonsense.The only way to know whether there is ANY effect from the absorption of CO2 on earth surface temperature is to calculate the spectrally integrated outgoing long wave radiation, or the OLR. It has not been done. My bet is that because of the earth’s hydro cycle, there has been no change, meaning water vapor has adjusted accordingly because of the increased emission from CO2 near 15 microns that causes upper tropospheric cooling. Miskolczi’s work computed a decrease in water vapor column amount by .649% of just over 2.5 prcm’s. Hansen has the water vapor/CO2 relationships wrong. The vapor is not a positive feedback on the system, it is negative. That was demonstrated years ago through the founding works of Emden who showed the absorbing and emitting properties of water vapor alone are sufficient to generate an enormous hydro cycle and define tropopause temperatures near 10 Km from the ensuing convective overturn just from water vapor alone. It is no surprise the modeling is dicked up.
Arno Arrak says:
April 25, 2012 at 8:57 am
The reason IPCC models cannot correctly reproduce polar amplification is that there is no polar amplification. Polyakov et al. attempted to detect it observationally in 2002 and came back empty handed. There is Arctic warming indeed but no thanks to polar amplification as IPCC pseudoscience would have it. Arctic warming is not greenhouse warming at all but is caused by Atlantic Ocean currents carrying warm water from the Gulf Stream into the Arctic. Proof of this is available in this article: http://curryja.files.wordpress.com/2011/12/arno-arrak.pdf . The article also works out the consequences of this fact to the rest of the climate system.
Arno, Arctic amplification has been detected in recent years and several papers have been published on this topic. Note that the amplified warming is something you see in autumn and winter, not in summer as surface temperatures are constrained near 0C as the energy from the sun is used to melt the ice. Dr. Screen’s most recent estimates are that 50% of the autumn warming in the Arctic is a direct result of the increases in more open water during the summer and that the rest is an increase in heat transport to the Arctic. Note that you can see evidence of this by the positive anomalies in latent and sensible heat transfer from the ocean to the atmosphere as the ocean freezes up and releases the heat it gained during summer. You also see positive anomalies in sensible and latent heat transfer in the North Atlantic, which is likely a result of warmer SSTs.
During spring you can also see an effect from reduced winter sea ice cover in the Eurasian sector, but you also see a strong signal from increased heat transport as well as a signal from earlier melt of the snow cover over land..
Arctic amplification is not a psuedo-science, it’s a direct result of losing snow and ice cover.
Bob I do find it curious that figures I’ve seen at the IPY meeting in Montreal differ from what you are showing. Perhaps you could show a hovmoller plot by time (on x-axis) vs. lat on the y-axis of temperature anomalies from GISS. I saw a plot like that today which does counter what you are showing. Both show enhanced warming during the same time-periods, but the warming in the last decade is larger than that during the 1940s. I tend to work with the reanalysis data sets which also show different results than you presented.
You could look at deriving an amplification factor, the arctic temperature relative to the global temperature.
Also, note that arctic amplification is a feature mostly during autumn and winter.
Finally if you want to make the argument that the models are not getting the temperature trends right then you really need to look at the individual ensemble members, their standard deviation (after taking into account autocorrelation in the time-series) and testing the hypothesis of whether or not the models show statistically different trends than the observations. If I were to compare the multi-model ensemble mean sea ice extent to the observations, I would conclude that the models remain conservative in regards to the observed ice loss. But, if I include the 2 sigma variance, that is not the case.
Julienne Stroeve says: “Bob I do find it curious that figures I’ve seen at the IPY meeting in Montreal differ from what you are showing. Perhaps you could show a hovmoller plot by time (on x-axis) vs. lat on the y-axis of temperature anomalies from GISS.”
I wish I had the capabilities to create Hovmollers, Julienne. They are a wonderful way to present data–very enlightening.
Do you have a link to the figures from the IPY meeting? Maybe they used different timeframes.
Regards
Bob, I don’t have a link to the plot but I saw the hovmoller in the presentation listed below.I agree that hovmollers are great ways to view variability both seasonally, interannually and by latitude.
G. Lesins1, T.J. Duck1, J.R. Drummond1
1Dalhousie University, Halifax, Canada
Surface and upper air radiosonde measurements starting from the 1950s are presented for Canadian Arctic stations to show that surface warming trends are enhanced compared to the global average (Arctic Amplification) in the presence of a surface-based boundary layer temperature inversion. The high static stability suppresses boundary layer turbulence and prevents the sensible and latent heat fluxes from transporting the positive surface forcings to the free troposphere. The perturbation form of the surface energy balance equation is used to show that Arctic amplification has three contributing factors: 1) the total surface forcing, 2) changes in the latent and sensible heat fluxes that respond to the forcing, and 3) the skin temperature which determines the response of the upward longwave irradiance. The predominance of very strong surface-based temperature inversions during the polar night helps to explain why the recent warming trend is highly seasonal. The enhanced sensitivity of the surface temperature as a result of the inversion will operate regardless of the feedback mechanisms that are operating. The analysis can be applied equally well to amplification from melting sea ice, increases in the meridional heat flux or changes in the local cloud and aerosol properties.
A new measure of Arctic Amplification is presented that is defined as the ratio of surface to upper air monthly averaged temperature anomalies linearly regressed over the entire instrumental record of the station. The Arctic Anomaly Amplification (AAA) is shown to produce similar values to the conventional definition of amplification but with the advantage of being independent of a particular warming period and independent of using a global average as a reference. Hence it will characterize the enhanced amplitudes of Arctic surface fluctuations regardless of whether it is cooling or warming. The geographic variation of the AAA, which can be applied to any station world-wide, provides further confirmation of the importance of the static stability of the boundary layer in determining the sensitivity of the surface temperature to various climate forcings.
Julienne Stroeve: Thanks for the exchange.
Regards
Remember too that the model spread provides a measure of natural climate variability.
Models and the spread between models aren’t a measure of anything. Measures come from measuring real things. And I thought we were repeatedly told the models don’t simulate natural climate variability.
Multi-model ensemble mean provides a robust metric for which to compare against the observations, because the inherent noise in each model simulation is uncorrelated from one simulation to the next, and therefore averaging over many ensemble members reduces noise levels and improves estimates of any overall forced trend.
There is no noise in a model. The models may simulate noise (and I’ll skip over climate science’s abuse of the term ‘noise’), but I’ve never heard that they do. I have seen many times that variability between model runs (what you call ensemble members) results from differing initial conditions.