Trenberth Still Searching for Missing Heat

Kevin Trenberth is one of the authors of new Balmaseda et al (2013) paper Distinctive climate signals in reanalysis of global ocean heat content.

I find the title of the paper somewhat odd. The paper is based on the European Centre for Medium-Range Weather Forecasts (ECMWF) Ocean Reanalysis ORAS4. That reanalysis is described in detail in the Balmaseda et al (2012) paper (submitted) Evaluation of the ECMWF Ocean Reanalysis ORAS4. (Update:  Sorry. That link no longer functions. UCAR has apparently removed the preprint version of the paper from its website.  Refer to the abstract here.) Basically, the reanalysis is the product of a climate model that has data rolled into it. Since volcanic aerosols and sea surface temperatures are used as inputs, it should therefore come as no surprise that the reanalysis will include the “distinctive climate signals” associated with El Niños and volcanic eruptions.

FIRST: A BRIEF LOOK AT THE EARLIER PAPER THAT DESCRIBES THE ORAS4 REANALYSIS

Figure 1

My Figure 1 (with my note) is Figure 4 from the Balmaseda et al (2012) paper (submitted) Evaluation of the ECMWF Ocean Reanalysis ORAS4. (This is NOT the paper that Kevin Trenberth co-authored. But I want to discuss it before we move on to the more recent paper.) Figure 1 contains three time-series graphs that represent the temperatures of the global oceans at different depths from 1958 to 2009. Each cell contains three variables. The blue “control integration” (CNTL) curves are the outputs of the models that don’t fold in the data. Balmaseda et al describes them as:

It is important to evaluate the impact of assimilation in ORAS4 by comparing it with a simulation that does not assimilate data. This simulation, called the control integration (CNTL), uses the same spin-up, forcing fields, SST/sea-ice relaxation and relaxation to climatology (with 20-year time scale) as ORAS4.

The black curves are the five ensemble members of ORAS4. And the red “NoBias_Crtn” curve “is equivalent to the unperturbed member of ORAS4 but without bias correction.” All of the graphs show how poorly the model, the blue “control integration” (CNTL) curves, simulates the warming. The bottom cell shows no warming at depths below 2000 meters in both the black ORAS4 curves and the red “ORAS4 without bias correction” curve. For depths of 700m to 2000m, the right-hand cell, the red “ORAS4 without bias correction” shows the same temperature in 1958 and 2000, but the black ORAS4 curves show a gradual warming due to bias corrections. Is the upward swing in the red “ORAS4 without bias correction” a result of the introduction of ARGO floats to a dataset that had poor spatial coverage before them?

For the upper 700m, the red “ORAS4 without bias correction” curve shows little to no warming from the mid-1960s to the mid-1980s, when an upward shift takes place. The red “ORAS4 without bias correction” curve basically remains unchanged from the early 1990s to 2000, when another upward shift takes place, leading to another plateau. Upward shifts give the appearance that Mother Nature is the primary cause of warming, and that’s not practical in a world that’s supposed to be warmed by greenhouse gases, so that would definitely need to be corrected. The bias corrections in the black ORAS4 curves smooth out the 2000 upward shift to make it look like a more gradual increase, and the corrections lower the temperature significantly before 1990 to provide a greater long-term warming.

Some of you might think these are yet more examples of inconvenient results being resolved through corrections.

A BRIEF LOOK AT THE MORE RECENT PAPER

Okay, we’re back to the Balmaseda et al (2013) paper Distinctive climate signals in reanalysis of global ocean heat content. That’s the paper coauthored by Trenberth.

The abstract of Balmaseda et al (2013) reads (my boldface):

The elusive nature of the post-2004 upper ocean warming has exposed uncertainties in the ocean’s role in the Earth’s energy budget and transient climate sensitivity. Here we present the time evolution of the global ocean heat content for 1958 through 2009 from a new observational-based reanalysis of the ocean. Volcanic eruptions and El Niño events are identified as sharp cooling events punctuating a long-term ocean warming trend, while heating continues during the recent upper-ocean-warming hiatus, but the heat is absorbed in the deeper ocean. In the last decade, about 30% of the warming has occurred below 700 m, contributing significantly to an acceleration of the warming trend. The warming below 700 m remains even when the Argo observing system is withdrawn although the trends are reduced. Sensitivity experiments illustrate that surface wind variability is largely responsible for the changing ocean heat vertical distribution.

It would be interesting to see just how much the warming trend is reduced when ARGO data is removed.

Figure 2 is, I believe, Figure 1 from Balmaseda et al (2013). Since the paper is paywalled, the illustration is from the World’s oceans are getting warmer, faster post at CarbonBrief. It illustrates the warming of ocean heat content for the depths 0-300 meters, 0-700 meters and “total depth”, but because there has been no warming below 2000 meters in the ORAS4 reanalysis, the “total depth” is kind of misleading. The SkepticalScience post New Research Confirms Global Warming Has Accelerated also presents that same graph. And of course, Joe Romm cross posted Dana1981’s post from SkepticalScience as In Hot Water: Global Warming Has Accelerated In Past 15 Years, New Study Of Oceans Confirms over at Climate Progress.  Curiously, looking back at my Figure 1, the only acceleration appears in the red “ORAS4 without bias correction” for depths of 700m to 2000m, but that should have been excluded from Balmaseda et al (2013).

Figure 2

Do Balmaseda et al (2013) address how much of the long-term warming is also a response to “surface wind variability”? As an example, see Figure 3, which shows the NODC ocean heat content (0-700 meters) for the North Pacific north of 24N, with and without the 1989-1990 shift that’s likely caused by a shift in the “surface wind variability”. Figure 3 was presented and discussed (as Figure 25) in the post Is Ocean Heat Content Data All It’s Stacked Up to Be?

Figure 3

CLOSING

Ocean heat content is at best a make-believe dataset. Refer again to the post Is Ocean Heat Content Data All It’s Stacked Up to Be? Even with all of the adjustments to the NODC’s ocean heat content data, the data still indicates the warming resulted from natural factors, as shown in that linked post.

A reanalysis is an even more abstract form of ocean heat content “data”—one that also requires “corrections” to provide the desired results.

Curiously, Paul Voosen’s October 2011 article Provoked scientists try to explain lag in global warming includes quotes from a handful of well-known climate scientists—including Kevin Trenberth. Voosen had this to say about Trenberth’s opinion of ARGO:

Trenberth questions whether the Argo measurements are mature enough to tell as definite a story as Hansen lays out. He has seen many discrepancies among analyses of the data, and there are still “issues of missing and erroneous data and calibration,” he said. The Argo floats are valuable, he added, but “they’re not there yet.”

A reanalysis didn’t make the ARGO floats any better; it simply provided a way for Trenberth to confirm his beliefs–regardless of whether or not those beliefs are realistic.