Alternate Title Question: Why Are Global Sea Surface Temperatures Falling Short Of IPCC Projections?
In this post I illustrate and discuss the relatively small rise in the Sea Surface Temperature anomalies (SST) of the East Pacific (from pole to pole) since November 1981 (the start of the Reynolds OI.v2 dataset) and the very obvious upward shifts in the Sea Surface Temperature anomalies of the Rest of The World, which is made up of the Atlantic, Indian and West Pacific Oceans, a subset that represents approximately 66% of the surface area of the global oceans. I’ve illustrated the shifts in this subset in an earlier post, but in that post, I corrected the data for volcanic aerosols and smoothed the data with a 13-month running-average filter. I now present the data without the volcano adjustments or smoothing to assure you that these adjustments have not created the effects.
This post also presents the differences between the Reynolds OI.v2 SST data and the IPCC Climate Model Hindcasts/Projections for the satellite-era. The comparisons are for the East Pacific SST anomalies and the SST anomalies of the Atlantic-Indian-West Pacific Oceans.
Note: My usual links to past posts have been updated to my new home at WordPress. Please update your Favorites and Blog Rolls.
For those new to discussions of ENSO (El Niño-Southern Oscillation), refer to the post An Introduction To ENSO, AMO, and PDO – Part 1. And refer to the post An Introduction To ENSO, AMO, and PDO — Part 2 if you are not familiar with the additional multidecadal variability of the North Atlantic Sea Surface Temperatures.
I’ve recently added two graphs to my monthly Sea Surface Temperature (SST) anomaly updates. (February 2011 SST Anomaly Update) I first presented the reason for these additions in the post Sea Surface Temperature Anomalies – East Pacific Versus The Rest Of The World. The first of the datasets to be discussed is the East Pacific Ocean from pole to pole (90S-90N, 180-80W), so it also includes the portions of the Arctic and Southern Oceans encompassed by those coordinates. As shown in Figure 1, its SST anomalies mimic NINO3.4 SST anomalies, a commonly used El Niño-Southern Oscillation (ENSO) proxy. Keep in mind when viewing the variations in SST anomalies in that graph that the East Pacific data represents approximately 33% of the global oceans. (The percentage is based on the NCEP/DOE Reanalysis-2 “Land Mask” data available through the KNMI Climate Explorer).
The second is the Rest of the World, from pole to pole (90S-90N, 80W-180), which is made up of the Atlantic, Indian, and West Pacific Oceans (and corresponding portions of the Arctic and Southern Oceans). The SST anomalies for this portion of the globe have risen in steps in response to the significant El Niño events of 1987/88 and 1997/98. What differentiates these ENSO events from others is that both were significant El Niño events that were followed by major La Niña events. They also were not counteracted by an explosive volcanic eruption, which happened in 1982 when the eruption of El Chichon suppressed the global response to the 1982/83 El Niño event. In Figure 2, I’ve added period average data to highlight the upward steps in the data. Note also that it appears this portion of the globe is preparing to make another upward step in response to the 2009/10 El Niño and 2010/11 La Niña.
But as you will note in the title blocks of those two graphs, the data has been adjusted for volcanic aerosols and it has been smoothed with 13-month running-average filters.
WHAT DOES THE DATA LOOK LIKE WITHOUT THE ADJUSTMENTS?
Figure 3 illustrates the East Pacific SST anomaly data without the Volcano-adjustment and smoothing. The ENSO variations in this dataset are very strong. The trend, on the other hand, is quite low, at only 0.21 deg C per Century.
The SST anomalies for the Rest of the World, which is made up of the Atlantic, Indian and West Pacific Oceans from pole to pole (90S-90N, 80W-180), has a significant trend, Figure 4. A trend line on global temperature data tends to give the illusion of continuous rise with noise caused by ENSO and volcanic eruptions, but the two upward shifts in the data are still easy to discern, even with the trend line in Figure 4.
I tried a number of ways to illustrate those upward steps while preparing this post. One of things I considered: I did not want to be accused of cherry-picking start and end dates of the periods before, between, and after the upward shifts. With that in mind, the way that seemed best was to delete the data during the years in which the Rest of the World SST anomalies were responding to the evolution and decay of the 1987/88 and 1997/98 El Niño events. In Figure 5, the full term of the Atlantic, Indian and West Pacific SST anomaly data is shown in the background, in royal blue. Then the data is divided into three periods. The first, in green, is the period before the response to the 1987/88 El Niño, and it runs from the start of the dataset, November 1981, to December 1986. Recall that the 1982/83 El Niño was counteracted by the El Chichon eruption during that period. The years of 1987 and 1988 are excluded. The mid-period data, green, runs from January 1989 to December 1996. It contains the dip and rebound from the Mount Pinatubo eruption. The years El Niño evolution and decay years of 1997 and 1998 are excluded. The final period, light blue, includes the data from January 1999 to December 2008. The data from January 2009 to present is also excluded, since the Atlantic, Indian and West Pacific SST anomaly data appears to be rising in another step in response to the combined 2009/10 El Niño and 2010/2011 La Niña. Note that the mid and late period subsets both start and end during La Niña events.
I’ve also added trend lines for the three periods in Figure 5. The trends are basically flat. In other words, without the El Niño-induced rises in 1987/88 and 1997/98, there would be little to no trend in the Atlantic, Indian and West Pacific SST anomaly data. It’s really tough to miss the upward steps, once you know they exist. There are links at the end of this post to past posts that discuss the processes that cause these upward shifts.
The KNMI Climate Explorer Monthly scenario runs webpage allows users to plot the outputs of the IPCC AR4 climate models collected and archived by the Program for Climate Model Diagnosis and Intercomparison (PCMDI). I used the first selection on the table, “all members”, 20c3m / sresa1b, which is a combination of the 20C3M hindcast for the 20th Century and the SRES A1B projections for the 21st Century. And while there are 54 ensemble members shown on the selection table, the output for TOS included 32 of them. I cross checked the mean of these runs against the third selection provided by KNMI, “multi-model mean,” and the differences were insignificant. The linear trend differences for the datasets I’m using were less than 0.01 degree C per Century, so in the future, I’ll save some time and use the “multi-model mean”. The designation “TOS” is Sea Surface Temperature. If memory serves me well, the difference between TOS and SST in the IPCC models is how sea ice is handled.
Figure 6 compares the Reynolds OI.v2 SST data and the mean of the IPCC Hindcasts/Projections for TOS anomalies for the Atlantic-Indian-West Pacific ocean, using the coordinates of 90S-90N, 80W-180. The linear trends are remarkably similar. The trend of the IPCC hindcast/projection is about 9% higher than that of the SST data. Not bad at first glance. However, as shown in Figure 5, between significant ENSO events, the linear trends for this portion of the global oceans are basically flat. The IPCC assumes the Sea Surface Temperatures for this part of the globe rise in response to anthropogenic forcings, where the satellite-based SST data since 1982 indicates they rise in response to significant ENSO events and do not rise between those ENSO events.
The portion of the global oceans where the IPCC hindcasts/projections truly fail is the East Pacific. Keep in mind that this dataset represents about 33% of the surface area of the global oceans. As shown in Figure 7, the IPCC hindcasts/projections for the East Pacific are approximately 1.32 deg C per Century, which is in line with the 1.41 deg C per Century for the Rest of the World oceans, Figure 6. But the satellite-based SST data only shows a linear trend of 0.19 deg C per Century. The linear trend of the IPCC hindcast/projections is more than 7 times higher than the SST data trend for 33% of the global ocean surface area. The IPCC assumes that anthropogenic forcings will cause the rise in Sea Surface Temperatures for the East Pacific to be similar to that of the Atlantic, Indian, and West Pacific Oceans. They clearly erred with this assumption.
BUT SST ANOMALIES RESPOND TO VOLCANIC AEROSOLS
One of the arguments I am anticipating is that the Reynolds OI.v2 SST data for the Atlantic, Indian, and West Pacific Oceans clearly responds to the Mount Pinatubo eruption. This is obvious with the dip and rebound starting in 1991 and ending about 1994. This may seem to some to be proof that the forcings used by IPCC Climate Models are what dictates the rise in sea surface temperature. Unfortunately that train of thought fails to consider the differences in the forcings. Volcanic aerosols reduce the amount of Downward Shortwave Radiation (visible light) that reaches and enters the oceans, and sea surface temperatures decrease in response. But anthropogenic greenhouse gases increase the amount of Downward Longwave Radiation (infrared radiation), not Downward Shortwave Radiation, and there is a significant difference between the two forcings. Downward shortwave radiation from the sun reaches depths of a couple hundred meters, decreasing in intensity with depth, but downward longwave radiation from greenhouse gases only impacts the top few millimeters of the oceans. It has been argued that most if not all of the effects of downward longwave radiation are released by the oceans through evaporation.
Smith and Reynolds (2004) Improved Extended Reconstruction of SST (1854-1997)] stated about the Reynolds OI.v2 data, “Although the NOAA OI analysis contains some noise due to its use of different data types and bias corrections for satellite data, it is dominated by satellite data and gives a good estimate of the truth.” The truth is a good thing. Unfortunately, the IPCC hindcasts/projections show little evidence of “a good estimate of the truth,” for the East Pacific SST anomalies or for the Atlantic-Indian-West Pacific data.
Links to further illustrations and to discussions of the processes that cause the upward shifts in SST anomalies follow.
My first post about the multiyear aftereffects of significant ENSO events was the November 6, 2008 post Another Look at the Saw-Tooth Trends in the Indian Ocean.
The multiyear aftereffects were then discussed in the two-part post Can El Nino Events Explain All of the Global Warming Since 1976? – Part 1 and Can El Nino Events Explain All of the Global Warming Since 1976? – Part 2.
And for those who like visual aids, refer to the two videos included in:
The impacts of these El Nino events on the North Atlantic are discussed in:
There Are Also El Nino-Induced Step Changes In The North Atlantic
More detailed technical discussions can be found here:
The Reynolds OI.v2 SST data is available through the KNMI Climate Explorer Monthly Observations webpage, and the IPCC climate model outputs (hindcasts and projections) are available through their Monthly Scenario Runs webpage.