Climate Models Are NOT Simulating Earth’s Climate – Part 1

This post will serve as part 1 of the 2015 update of the model-data comparisons of satellite-era sea surface temperatures. The 2014 update is here. I’ve broken the update into two parts this year.

INTRODUCTION

The locations, the timings and the magnitudes of the naturally occurring variations in the surface temperatures of our oceans are primary factors that drive weather and, in turn, climate on Earth.  In other words, where and when the surfaces of the oceans warm, or cool naturally and by how much—along with other naturally occurring factors—dictate where and when land surface air temperatures warm and cool and where precipitation increases or decreases…on annual, decadal and multidecadal timeframes.

Unfortunately for the climate science community, the spatial patterns of the modeled warming rates for the global ocean surfaces from 1982 to 2015 (the era of satellite-enhanced sea surface temperature observations) show no similarities to the spatial patterns of the observed warming and cooling…no similarities whatsoever. This is blatantly obvious in Figure 1.  The map on the left includes the simulated sea surface temperature trends from 1982 to 2015 based on the average (multi-model mean) of the climate models stored in the Coupled Model Intercomparison Project Phase 5 (CMIP5) archive. Those models were used by the IPCC for their 5th Assessment Report. The multi-model mean basically represents the consensus of the climate modeling groups for how the surfaces of the oceans should warm if they were warmed by the factors (primarily manmade greenhouse gases) that drive the climate models. (For more information on the use of the multi-model mean, see the post here.)

Figure 1

The map to the right shows the observed warming and cooling rates of the ocean surfaces from 1982 to 2015 based on NOAA’s satellite-enhanced Optimum Interpolation (Version 2) sea surface temperature data (a.k.a. Reynolds OI.v2).  This is the standard 1-deg resolution (weekly, monthly) version of the Reynolds OI.v2 data…not the (over-inflated, out-of-the-ballpark, extremely high warming rate) high-resolution, daily version of NOAA’s Reynolds OI.v2 data, which we illustrated and discussed in the recent post 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.

Figure 1 is one of the best examples of a simple reality: that climate models are not simulating Earth’s climate as it exists.  The models show the greatest warming near the equator and at mid-latitudes of the Northern Hemisphere, while in the real world, the greatest warming has occurred at mid and high latitudes, with little warming for much of the eastern Pacific Ocean.  The models show warming at the high latitudes of the Southern Hemisphere, where the data show cooling.

The North Atlantic in the real world warmed at the highest rate.  That warming of the North Atlantic is associated with the Atlantic Multidecadal Oscillation.  The model mean does not present that additional warming in the North Atlantic, which indicates that the recent additional warming of the North Atlantic has occurred naturally; that is, the Atlantic Multidecadal Oscillation is not a process forced by factors that are used to make the oceans warm in the models.  We’ll discuss and illustrate this further in Part 2 this post.

The observed “C-shaped” pattern of warming in the Pacific is the result of the dominance of El Niño events during this period.  El Niño events release sunlight-created warm water from below the surface of the western tropical Pacific. That warm water temporarily floods into the eastern tropical Pacific, primarily along the equator, during El Niño events.  At the end of the El Niños, the leftover warm water is driven west by the renewed trade winds and by other ocean processes. Ocean currents carry the leftover warm water poleward to the Kuroshio-Oyashio Extension (east of Japan) and along the South Pacific Convergence Zone (east of Australia and New Zealand).  As a result of those processes, the observed sea surface temperatures of the East Pacific Ocean (from the dateline to Panama) and of the tropical Pacific (24S-24N, 120E-80W) show little warming in 34 years.  Because the climate models do not properly simulate El Niño and La Niña processes, they do not create the spatial patterns of warming and cooling in the Pacific. Keep in mind, the Pacific Ocean covers more of the surface of the Earth than all of the continental land masses combined, and the modelers show no skill at simulating how and where or why the surface of the Pacific Ocean warmed.

Phrased differently, the observed warming pattern in the Pacific is one associated with El Niño and La Niña events (a.k.a. El Niño-Southern Oscillation or ENSO).  ENSO helps the Pacific distribute heat (created by sunlight) from the tropics to the mid latitudes, and into adjoining ocean basins. The differences between the modeled and observed warming patterns should be caused by the failures of the models to properly simulate basic ENSO processes. Those failings are well known to the climate science community.  Two papers that present model failings at simulating ENSO are:

• Guilyardi et al. (2009) Understanding El Niño in Ocean-Atmosphere General Circulation Models, and
• Bellenger et al (2012) ENSO representation in climate models: from CMIP3 to CMIP5

ALL SEA SURFACE TEMPERATURE DATASETS SHOW SIMILAR SPATIAL PATTERNS

There, of course, will be persons who believe I cherry-picked the standard version of the NOAA Reynolds OI.v2 sea surface temperature data for the trend map in Figure 1.  Figure 2 shows the trend maps for three NOAA and one UKMO sea surface temperature products for the period of 1982 to 2014. (The HadISST data have not been updated through December 2015, so I’ve ended those trend maps in 2014.)  All four datasets show the same basic warming and cooling spatial patterns.

Figure 2

Datasets included:

• NOAA ERSST.v4 (NOAA’s recently introduced “pause buster” data, infilled, in situ only, adjusted for ship-buoy biases), Cell a
• NOAA ERSST.v3b (NOAA’s former dataset, infilled, in situ only, not adjusted for ship-buoy biases), Cell b
• UKMO HADISST (UKMO’s interpolated/infilled product, satellite-enhanced, not adjusted for ship-buoy biases), Cell c
• NOAA Optimum Interpolation/Reynolds OI.v2 – “Original” (NOAA’s original satellite-enhanced data, 1-deg resolution, infilled, presented weekly and monthly, not adjusted for ship-buoy biases), Cell d

Notes: The notation “in situ only” means the dataset includes only observations from ships (buckets and ship inlets) and from buoys (moored and drifting). The “satellite-enhanced” datasets also include in situ observations and the satellite-based data are also bias adjusted with the in situ data. “Infilled” means that data suppliers use statistical devices to create data for ocean grids without observations. [End notes.]

TRENDS ON A ZONAL-MEAN (LATITUDE-AVERAGE) BASIS

Another way to illustrate how poorly models simulate the warming and cooling rates of ocean surfaces is using graphs that show the 1982-2015 trends on a latitude-average basis.  And we’ll return to the original (weekly, monthly) 1-deg resolution version of NOAA’s Reynolds OI.v2 sea surface temperature data for these graphs.

Figures 3 through 7 are model-data trend (warming and cooling rate) comparisons of sea surface temperatures for the global oceans and for the Pacific, Atlantic, and Indian Oceans.  But they aren’t time-series graphs.  The horizontal (x) axis is latitude. The South Pole (“-90”) is to the left, the equator (“0” latitude) is center, and the North Pole (“90”) is to the right. The units of the vertical (y) axis are degrees C per decade—based on the calculated linear trends.  Each data point represents the linear trend (warming or cooling rate) in degrees C per decade for a 5-degree latitude band. For example, the data point at -82.5 (82.5S) latitude represents the linear trend of the high latitudes of the Southern Ocean surrounding Antarctica (85S-80S). The data points representing the trends then work northward (left to right) in 5-degree increments through each of the ocean basins (80S-75S, then 75S-70S, then 70S-65S, and so on) using the longitudes for each ocean basin. The average temperatures of latitude bands are called the “zonal mean” temperatures by climate scientists; thus the use of that term in the title blocks.

I’ve highlighted zero deg C/decade on the trend graphs.  Above zero deg C/decade, the trends are positive, indicating warming ocean surfaces, and below zero deg C/decade, the trends are negative, indicating cooling.  The greater the positive (negative) values, the faster the ocean surfaces have warmed (cooled) at those latitudes for 1982 to 2015.

Figure 3 presents the modeled and observed warming and cooling rates of the global oceans on a latitude-average (zonal-means) basis. For the period of 1982-2015, the climate models underestimate the observed warming at high latitudes of the Northern Hemisphere, but they overestimate the warming in the tropics and in the mid-to-high latitudes of the Southern Hemisphere. And the models do not capture the cooling of ocean surfaces at the high latitudes of the Southern Hemisphere.  There should be little wonder why models cannot simulate sea ice losses in the Arctic Ocean and sea ice gains in the Southern Ocean surrounding Antarctica.

Figure 3

Figure 4 shows the observed and modeled sea surface temperature trends for the Pacific Ocean (longitudes of 125E-90W) on a zonal-mean basis.  At and just south of the equator in the Pacific, sea surfaces show almost no warming since January 1982. And the highest observed warming occurred at the mid-latitudes of the North and South Pacific.  The models, unfortunately, do not create that spatial pattern. The models show much more warming in the tropics than observed.  The models also overestimate the warming at the high latitudes of the North Pacific, and they show warming in the Pacific portion of the Southern Ocean, while the observations show cooling there over the past 34 years.

Figure 4

Figure 5 shows the modeled and observed trends in sea surface temperature anomalies for the Atlantic Ocean (longitudes 70W-20E) from Jan 1982 to December 2015.  The models overestimate the warming in the South Atlantic and underestimate it North Atlantic, especially toward the high latitudes.   In fact, the models show just about the same warming trends from 40S to 70N—that is, the models show the Atlantic Ocean should have warmed at about 0.15 to 0.2 deg C/decade for the last 34 years for the latitudes of 40S to 70N—while the observed trends vary greatly over those latitudes.  Again, how can the climate scientists/modelers hope to create the warming and precipitation patterns on adjoining land masses when they can’t simulate the warming pattern of the surface of the Atlantic?

Figure 5

The last of the trend graphs on a zonal-mean basis is for the Indian Ocean, Figure 6.  The models, basically, show way too much warming at most latitudes.  As a result, the same problem problems exist in the models for the warming and precipitation patterns on land masses adjacent to the Indian Ocean.

Figure 6

Figure 7 includes two comparisons.  The top graph includes the model-simulated trends for the period of 1982 to 2015, on a zonal-mean basis, for the Atlantic, Indian and Pacific basins, and the bottom graph includes the observed trends for those ocean basins.

Figure 7

The modelers apparently believe the ocean basins should show similar warming rates as we progress from the Southern Ocean toward the high latitudes of the Northern Hemisphere.  But, because there are different well-known coupled ocean-atmosphere processes taking place in the ocean basins in the real world (like the Atlantic Multidecadal Oscillation in the North Atlantic, like El Niño-Southern Oscillation or ENSO in the Pacific), the observed warming rates show few similarities north of the mid-latitudes of the Southern Hemisphere.

PART 2

The next post in this series will present time-series graphs of the model simulations of sea surface temperatures and data in absolute form (not anomalies) for 1982 to 2015, the satellite era of sea surface temperature data. For a preview, refer to last year’s post here.

ADDITIONAL READING

This post presented evidence that the climate models that serve as the foundation for the hypothesis of human-induced global warming are flawed…fatally flawed.  You can find much more evidence of climate-model flaws in my free ebook On Global Warming and the Illusion of Control (25MB, .pdf).

CLOSING

The differences between modeled and observed warming and cooling rates of the surfaces of the global oceans strongly suggest two things: (1) that ocean circulation processes in climate models are flawed and (2) that the sensitivity of climate models to carbon dioxide and other forcings is too high.

The spatial patterns of the warming of the ocean surfaces dictate the spatial patterns of warming of the surface air over land, and those patterns of ocean warming and cooling contribute to the precipitation patterns on the continents.  Because the climate models cannot simulate the spatial patterns of the warming of sea surfaces, one wonders how the modelers could hope to properly simulate the warming of land surface air or the precipitation that occurs there.

For almost two decades, the IPCC has claimed that they have found the “fingerprints” of human-induced global warming.  Because they’re using climate models as the basis for those claims, it looks like they need a new method of fingerprint analysis. There are no similarities between the modeled and observed fingerprints shown in this post.

SOURCE

The maps, data and climate-model outputs presented in this post are available through the KNMI Climate Explorer.