Quiz: Which 30-Year Warming Period Is Recent?

Guest Post by Bob Tisdale

THREE MAPS ILLUSTRATING SIMILAR GLOBAL SURFACE TEMPERATURE CHANGES OVER DIFFERENT 30-YEAR PERIODS

The GISS map-making webpage allows users to create global maps of surface temperature anomalies for specific time periods or to create maps of the change in surface temperatures over user-defined time periods based on local linear trends.

My Figure 1 includes 3 maps from that webpage.  They are color-coded to show where and by how much surface temperatures have changed around the globe over three different 30-year periods, based on the GISS Land-Ocean Temperature Index (LOTI) data.  I’ve highlighted in red the respective global temperature changes in deg C.  They’re basically the same at 0.48 deg C and 0.49 deg C. You’ll note that I’ve also blacked-out the time periods, because I’ve asked the question Which 30-Year Warming Period Is Recent?

Figure 1

Before I provide the answer, here’s some…

INFORMATION ABOUT THE DATA AND CLIMATE MODELS

Global Land-Ocean Surface Temperature Anomaly Data

There are a number of suppliers of global land plus ocean surface temperature data. For this exercise, we’re presenting the GISS Land-Ocean Temperature Index (LOTI) reconstruction, which is a product of the Goddard Institute for Space Studies.  The GISS Land-Ocean Temperature Index (LOTI) is a merger of much-adjusted near-land surface air temperature anomaly data for the continental land masses and much-adjusted sea surface temperature anomaly data for the ocean surfaces.  Starting with the June 2015 update, GISS LOTI uses the new NOAA Extended Reconstructed Sea Surface Temperature version 4 (ERSST.v4), a.k.a the “pause-buster” reconstruction, which infills grids without temperature samples.  For land surfaces, GISS adjusts GHCN and other land surface temperature products via a number of methods and infills areas without temperature samples using 1200km smoothing. Refer to the GISS description here.   Unlike the UK Met Office and NCEI products, GISS masks sea surface temperature data at the poles, anywhere seasonal sea ice has existed, and they extend land surface temperature data out over the oceans in those locations. GISS uses the base years of 1951-1980 as the reference period for anomalies.  The values for the GISS product are found here.

For summaries of the oddities inherent in the new NOAA ERSST.v4 “pause-buster” sea surface temperature data see the posts:

• The Oddities in NOAA’s New “Pause-Buster” Sea Surface Temperature Product – An Overview of Past Posts
• On the Monumental Differences in Warming Rates between Global Sea Surface Temperature Datasets during the NOAA-Picked Global-Warming Hiatus Period of 2000 to 2014

Climate Models

Later in the post we’ll present model-data comparisons.  We’re using the model-mean of the estimates of past global surface warming (1880 to 2015) based on the climate models stored in the CMIP5 (Coupled Model Intercomparison Project Phase 5) archive, with historic forcings through 2005 and RCP8.5 forcings thereafter.  (The individual climate model outputs and model mean are available through the KNMI Climate Explorer.) The CMIP5-archived models were used by the IPCC for their 5^th Assessment Report. The RCP8.5 forcings are the worst-case future scenario.

NOTE: If climate model forcings are new to you, see Chapter 2.3 of my free ebook On Global Warming and the Illusion of Control – Part 1 (25MB pdf), and if you’re new to RCPs (Representative Concentration Pathways), see Chapter 2.4.[End note.]

We’re using the multi-model mean (the average of the climate model outputs) because the model-mean represents the consensus of the modeling groups for how surface temperatures should warm if they were warmed by the numerical representations of forcings that drive the models. See the post On the Use of the Multi-Model Mean for a further discussion of its use in model-data comparisons.

30-Year Periods

In the maps and graphs, we’re presenting temperature changes based on linear trends over 30-year periods.  Why 30 years?  Because climate is typically defined as the average of a weather-related variable over 30 years. See the Frequently Asked Questions webpage from the World Meteorological Organization (my boldface):

Climate, sometimes understood as the “average weather,” is defined as the measurement of the mean and variability of relevant quantities of certain variables (such as temperature, precipitation or wind) over a period of time, ranging from months to thousands or millions of years.

The classical period is 30 years, as defined by the World Meteorological Organization (WMO).  Climate in a wider sense is the state, including a statistical description, of the climate system.

Back to the quiz:

ONE IS RECENT; TWO ARE NOT    

The bottom map (Cell c) shows the changes in temperature (based on local linear trends) for the most recent 30-year period of 1986 to 2015, the center map (Cell b) covers 1964 to 1993 (ending more than 2-decades ago) and the top map (Cell a) shows the temperature changes for the 3-decade period of 1916 to 1945, ending 70+ years ago.

There are aspects of the maps that give away the answers.  There’s a lot of temperature-change data missing in the high latitudes of the Southern Hemisphere in Cell a, which suggest the period is before the 1950s. There are few to no surface temperature observations in Antarctica before the 1950s. For the other two, there is more warming in the Arctic in Cell c than in Cell b, and that suggests that Cell c is the newer of those two. Why? Naturally occurring polar amplification, which impacts the Arctic, has been stronger recently than it was a couple of decades ago.

NOTE: For those new to polar amplification, see Chapter 1.18 of On Global Warming and the Illusion of Control – Part 1.  As illustrated and discussed in that chapter, not only is polar amplification a naturally occurring phenomenon that works both ways (amplifying warming during global warming periods and amplifying cooling during global cooling periods), it is a phenomenon that is not simulated properly by climate models…but that’s not surprising because there are many naturally occurring phenomena that climate models are still not capable of simulating correctly. [End note.]

In Figure 2, I’ve uncovered the time periods, and, to the right of the maps, I added time-series graphs of the global GISS LOTI data that correspond to those time periods.

Figure 2

You’ll note that the temperature changes shown on the maps are slightly different than the temperature changes that MS EXCEL derived from linear-trend analyses of the time-series data. Why?  I don’t know for certain so I won’t speculate.  Regardless of whether we look at the temperature changes from the maps or from the data, they are similar for the three time periods.

If you’re a newcomer to the ongoing debate about human-induced global warming, you may have run across a phrase similar to “global warming is worse than we thought”. There may have been some truth to that statement a number of decades ago. But comparisons of model outputs with even the newly revised Land-Ocean Temperature Index data from the Goddard Institute of Space Studies show that is no longer true. That is, recently global warming over 30-year periods is occurring at a rate that is slower than predicted by climate models.

CAN THE CLIMATE MODELS USED BY THE IPCC SIMULATE THE TEMPERATURE CHANGES OVER THOSE THREE PERIODS?

Answer:  The models used by the IPCC are do a reasonable job for the period of 1964 to 1993, but they come nowhere close to being able to simulate the warming that occurred from 1916 to 1945, and climate models are over-predicting warming during the most-recent 30-year period.

Figure 3 includes model-data comparisons of global surface temperature anomalies using time-series graphs for our three periods of 1916-1945, 1964-1993 and 1986-2015. The data are represented by the GISS Land-Ocean Temperature Index and the models are the multi-model mean (average) of the climate models stored in the CMIP5 archive.  (See the preliminary notes about the data and models.)

Figure 3

The comparison in the center graph shows the model change in surface temperatures almost matching the data for the period of 1964 to 1993. In other words, the models appear to be performing reasonably well during this period.  And the bottom graph shows that, according to the GISS Land-Ocean Temperature Index and the CMIP5-archived climate models, global warming over the most-recent 30-year period is occurring at a rate that’s noticeably lower than simulated by the consensus of the climate-modeling groups.  The consensus of the modeling groups indicates that global warming should have occurred at a rate of 0.255 deg C/decade but the data indicate global surface warming has occurred at a rate of only 0.166 deg C/decade from 1986 to 2015.

Let’s focus on the early warming period of 1916-1945 because it’s critical to the evaluation of model performance and to our understanding of what causes global surfaces to warm.  According to the GISS LOTI data and the CMIP5-archived climate models, global surface temperatures warmed at a rate from 1916 to 1945 that was substantially higher than was hindcast by the models…more than 3 times faster than determined by the models.  That means the forcings used to drive the models can’t explain most of the surface warming from 1916 to 1945.

What caused the additional warming in the 30-year period of 1916 to 1945 that the models can’t explain? If the forcings used to drive the models can’t explain the warming, then the warming obviously must have occurred naturally.  There are naturally occurring, coupled ocean-atmosphere processes like the Atlantic Multidecadal Oscillation and the El Niño-Southern Oscillation that can cause global surfaces to warm naturally over multidecadal timeframes.

Note: For those new to the Atlantic Multidecadal Oscillation, see Chapter 3.3 of On Global Warming and the Illusion of Control – Part 1.  El Niño events, when they dominate, can also cause long-term global surface warming. See Chapter 3.7 of that ebook. You’ll also discover that climate models do not—cannot—simulate the naturally occurring processes that can cause global surfaces to warm over multidecadal periods…and that means the modelers try to explain the naturally occurring portion of the warming in recent decades by using carbon dioxide and other greenhouse gases. [End note.]

Back to performance of the models:  Those differences between model and data trends from 1916 and 1945 also call into question the performance of the models where they align better later in the 20^th Century. During the 30-year period from 1964 to 1993 the warming rate of the models aligns well with the observed trend shown by the data. Is this proof that the models are properly simulating the effects of the manmade greenhouse gases and other forcings that are used to drive the models?  Of course not.  We’ve already seen that global surfaces can warm naturally at a rate that’s much higher than hindcast by the models, and that additional natural warming from 1916 to 1945 suggests that most of the warming seen in the latter part of the 20^th Century could also have occurred naturally. Further, that means it’s likely the climate models are far too sensitive to manmade greenhouse gases, which in turn means the model projections of future warming are much too high.

We already have an indication the model forecasts are too high: It’s plainly obvious in the over-estimated warming for the most-recent 30-year period of 1986 to 2015 shown in the bottom graph of Figure 3.

The fact that the models better align during the latter part of the 20^th Century also suggests the models are simply tuned to perform well at that time. Unfortunately, as far as I know, only two of the dozens of climate modeling groups around the globe have documented how they tune the models they submitted to the CMIP5 archive.

CLOSING

There are a number of take-home points to this post. First, during the three global warming periods discussed in this post—1916-1946, 1964-1993, and 1986-2015—there were similar observed changes in global surface temperatures. It’s tough to claim that the recent global warming is unprecedented when surface temperatures rose at a comparable rate over a 30-year timespan that ended about 70 years ago.

Second, climate models are not simulating climate as it existed in the past or present.  The model mean of the climate models produced for the IPCC’s 5^th Assessment Report simulates observed warming trends for one of the three periods shown in this post. Specifically, during the three global warming periods discussed in this post, climate models simulated three very different rates of warming (+0.050 deg C/decade for 1916-1946, +0.155 deg C/decade for 1964-1993, and +0.255 deg C/decade for 1986-2015), yet the data from GISS indicated the warming trends were very similar at +0.16 deg C/decade and +0.166 deg C/decade.    If climate models can’t simulate global surface temperatures in the past or present, why should anyone have any confidence in their prognostications of future surface temperatures?

Third, the models’ failure to simulate the rate of the observed early 20^th Century warming from 1916-1945 indicates that there are naturally occurring processes that can cause global surfaces to warm over multi-decadal periods above and beyond the computer-simulated warming from the forcings used to drive the climate models.  That of course raises the question, how much of the recent warming is also natural?

Fourth, for the most-recent 30-year period (1986-2015), climate models are overestimating the warming by a noticeable amount. This, along with their failure to simulate warming from 1916-1945, suggests climate models are too sensitive to greenhouse gases and that their projections of future global warming are too high.

Fifth, logically, the fact that the models seem to simulate the correct global-warming rate for one of the three periods discussed does not mean the climate models are performing properly during the one “good” period.

Sixth, is there a maximum rate at which global surface temperatures can warm?  We’ll discuss this further in an upcoming post.