>Including Ishii and Kimoto (2009), Domingues et al (2008), and Wijffels et al (2008)
The Levitus et al (2009) paper “Global Ocean Heat Content 1955-2008 in light of recently revealed instrumentation problems” included a comparison, Figure 1, of Ocean Heat Content (OHC) data from Levitus et al (2009), Ishii and Kimoto (2009), and Domingues et al (2008). All three papers illustrate the same variable, but the data varies significantly between the three datasets. Note the divergence of the Levitus et al data (red curve) in 2003. The comparison graph is presented as their Figure S9 on page 8 (page 13/22 of pdf version) of the supplement to Levitus et al (2009):
The text for the comparison graph (Figure 1) reads, “Yearly time series of ocean heat content (10E+22 J) for the 0-700 m layer from this study and from Domingues et al.  and Ishii and Kimoto . Linear trends for each series for 1969-2008 given in the upper left portion of the figure. Reference period is 1957-1990”
And in the text of the paper, while discussing Figure S9, Levitus et al state, “All three estimates of OHC700 exhibit a similar linear trend but there are differences in the year-to-year variability. These estimates all use different processing methods and there are some differences in the data used. For example D08 do not use any MBT data in their study. We compare the linear trends for the 1969–2003 period since 1969 is a few years after XBTs began to be used in large numbers and 2003 is the last common year of analysis for the three estimates.” They then list the trends for those periods and conclude with, “The general agreement of the three results suggests that the global linear trend is qualitatively robust to processing methods although there are differences in the magnitude of the trend. However there is reason to question the D08 results as we shall show.”
The linear trend listed in Figure S9 for Domingues et al (D08) is 0.41*10^22 Joules/year, which is 71% greater than the Ishii and Kimoto linear trend of 0.24*10^22 Joules/year. How is that a “general agreement” or “a similar linear trend”?
Levitus et al present data from 1955 to 2008, which includes a period (1955 to 1969) when OHC declines, yet they exclude that early period in the listed linear trends. They do provide their reasoning, but it appears they selected the start year to maximize the trends.
The Levitus et al Figure S10 includes yet another OHC reconstruction, that of Wijffels et al (2008). It is included here as Figure 2. Note the text for the graph states, “Linear trends for each series for 1969-2007 given in the upper left portion of the figure. Reference period is 1957-1990,” but linear trends are not illustrated. Since the Wijffels linear trends are missing from the Levitus et al Figure S10, I’ll include them in the following comparisons. Levitus et al discuss the reason for the divergence of the Wijffels et al data. They state, “In particular W08 results clearly document [Wijffels et al., 2008, Figure 9] that the regression technique they use to compute their fall-rate correction equation produces errors.”
THE TRENDS PRESENTED BY LEVITUS ET AL
For Figures 3, 4 and 5, I used the coordinates feature of Windows Paint to reproduce the OHC curves presented in Figures S9 and S10 of Levitus et al (2009). Figure 3 illustrates the trends from 1969 to 2003, the years Levitus et al elected to use for their presentation of trends. The dashed lines represent the data excluded in the trends. Note the agreement with the Levitus et al trends illustrated in Figure 1.
THE TRENDS PRESENTED BY THE DATA IN THE LEVITUS ET AL TREND COMPARISONS
In Figure 4, the trend analyses begins in 1955 and ends in 2003, the last year common to all of the OHC datasets. Again, the dashed lines represent the data that was not included in the trend analyses. As could be expected, over that period, the trends presented by the data are significantly less than the trends Levitus et al chose to present. The dataset with the highest trend, Domingues et al, drops from ~0.41 to ~0.24*10^22 Joules/yr.
Figure 5 presents the linear trends for each of the OHC datasets over their individual terms. The trend of the dataset that extends the farthest, Levitus et al, rises very slightly from ~0.21 to ~0.27*10^22 Joules/yr. (Those with keen eyes will also note the rise in the Ishii and Kimoto trend even though there is a drop in the last few years.)
I’ve provided the comparison graph of all four datasets without the trend lines in Figure 6. With the Wijfells data added, the anomalous rise in 2003 appears to make Levitus et al the outlier. Could it be that the problem Levitus et al claim to be present in the Wijffels data, “the regression technique they use to compute their fall-rate correction equation produces errors,” actually applies to the Levitus et al data?
I originally intended to use gif animations to show the changes in trends, but an unusual characteristic of EXCEL made its presence known in the graphs. I illustrated it in the following post Here’s An Odd Effect Created By EXCEL.