Guest post by Craig Loehle
On June 2, 2009 at WUWT in Anomalous Spike in Ocean Heat Content I commented on what looked like a data discontinuity in ocean heat content data. In this follow-up post, I show that the recent update to the OHC data at NOAA adjusts the recent data down, as I suggested nearly 2 years ago (though I doubt I had anything to do with it). The original post is in italics followed by my update plus an update on trends.
In the paper
Levitus S., J. I. Antonov, T. P. Boyer, R. A. Locarnini, H. E. Garcia, A. V. Mishonov (2009), Global ocean heat content 1955–2008 in light of recently revealed instrumentation problems, Geophys. Res. Lett., 36, L07608, doi:10.1029/2008GL037155
the long term trend of ocean heat content is reanalyzed in Levitus to attempt to correct for bias in instrumentation, and the record is extended. The graph below depicts the result.
The most recent period, from 2003, uses the ARGO profiling floats, whereas earlier periods use a variety of instruments with various biases. Patching all these data together is a challenge. I draw your attention to the strong spike in the red line from 2002 to 2003. This line is the point at which the earlier data is joined up with the ARGO data. The magnitude of the jump is the largest in the entire record. The transition to the ARGO data can not be said to have been accomplished with a long cross-calibration period. It thus looks to me like there may be an error in how the different data sets are stitched together. I in no way am implying malfeasance here. I have discussed this situation with Roger Pielke Sr. and Josh Willis and they agree it looks odd and merits further investigation. Dr. Pielke points out that there is not a comparable jump in the SST data. I think this example illustrates that if there is a big jump in the data right when you change your instrumentation, it is perhaps good to look a little closer.
Below is the change notice ftp://ftp.nodc.noaa.gov/pub/data.nodc/woa/DATA_ANALYSIS/3M_HEAT_CONTENT/PDF/heat_content_differences.pdf
from NOAA showing that the recent data (post-2002) have been adjusted downward based on many QA/QC issues. There is still a large jump in mid-2003 when the Argo float data join the other data in the database. I believe there is still a data continuity/calibration issue here and that recent data need further adjusting down.
There is a famous quote in the Climategate emails from Trenberth about the “travesty” that there is missing heat which we can’t account for. This refers to the fact that the radiation balance as understood and as modeled implies that the Earth should be absorbing radiation and warming up. The observed rate of warming, including heat stored in the ocean, is significantly less than predicted. This is the “travesty”. Please note that the adjustment of the OHC data in the above plot (black line after 2002) makes the missing heat term larger by a significant amount. The problem of balancing the Earth energy budget remains unsolved. If further adjustments are warranted, as I believe, then the problem grows further.
It is useful to plot just the data since the Argo float data were added to the database, in mid-2003 (note the scale as given from the NOAA download which is in per square meter differs from the Levitus scale which is a global sum).
For this period, the slope is 0.00160 GJ/m2/yr or a slight uptrend. In contrast the slope for the previous 36 years (the steepest part of the above plot, ending in 2002) is 0.01075, which is 6.7 times as fast of a rise. Either there has been a sudden deceleration in OHC rise, or there is still a discontinuity at the point where Argo data are added in (I think both are likely). Note that flat periods are present in earlier periods of the data, such as the 1980s, so a flat period is not unprecedented. However, it does match up with the satellite data.
It is useful to compare the above graph to the Argo-only data ( Loehle, C. 2009. Cooling of the Global Ocean Since 2003. Energy & Environment 20:99-102) available at
below which shows a cooling trend.
Commenters have pointed to the Levitus paper above to claim that my result is “wrong” which is interesting since this plot is exactly the Argo data supplied by Josh Willis of NASA JPL. The quality of the Argo data is evident in that the annual cycle with a period exactly 365 days (see my paper cited) is evident in this data but not in the full dataset above it. The Argo data as of my publication showed a cooling trend, but the Argo data are being updated to deal with quality issues so this may no longer be true (data not yet available).