Tisdale on 2011 ocean heat content and the GISS-Miss

Bob has done some very interesting work here, I call attention to figure 6 and figure 7  below where he asks:

If the observations continue to diverge from the model projection, how many years are required until the model can said to have failed?

January to March 2011 NODC Ocean Heat Content (0-700Meters) Update and Comments

by Bob Tisdale

INTRODUCTION

The National Oceanographic Data Center’s (NODC) Ocean Heat Content (OHC) anomaly data for the depths of 0-700 meters are available through the KNMI Climate Explorer Monthly observations webpage. The NODC OHC dataset is based on the Levitus et al (2009) paper “Global ocean heat content (1955-2008) in light of recent instrumentation problems”. Refer to Manuscript. It was revised in 2010 as noted in the October 18, 2010 post Update And Changes To NODC Ocean Heat Content Data. As described in the NODC’s explanation of ocean heat content (OHC) data changes, the changes result from “data additions and data quality control,” from a switch in base climatology, and from revised Expendable Bathythermograph (XBT) bias calculations.

The OHC anomaly data is provided from the NODC on a quarterly basis. There it is available globally and for the ocean basins in terms of 10^22 Joules. The KNMI Climate Explorer presents the quarterly data on a monthly basis. That is, the value for a quarter is provided for each of the three months that make up the quarter, which is why the data in the following graphs appear to have quarterly steps. Furnishing it in a monthly format allows one to compare the OHC data to other datasets that are available on a monthly basis. The data is also provided on a Gigajoules per square meter (GJ/m^2) basis through the KNMI Climate Explorer, which allows for direct comparisons of ocean basins, for example, without having to account for surface area.

This update includes the data through the quarter of January to March 2011. The Global and Tropical Pacific OHC anomalies both rose during the first quarter of 2011, as one would expect in response to a La Niña event. This relationship with ENSO is very apparent when OHC data is compared to SST data. Refer to Sea Surface Temperature Versus Ocean Heat Content Anomalies.

GLOBAL

The Global OHC data through March 2011 is shown in Figure 1. Even with the slight rise this quarter, it continues to be remarkably flat since 2003, especially when one considers the magnitude of the rise that took place during the 1980s and 1990s.

Figure 1

################################

TROPICAL PACIFIC

Figure 2 illustrates the Tropical Pacific OHC anomalies (24S-24N, 120E-90W). The major variations in tropical Pacific OHC are related to the El Niño-Southern Oscillation (ENSO). Tropical Pacific OHC drops during El Niño events and rises during La Niña events. As discussed in the update for October to December 2010, the Tropical Pacific had not as of then rebounded as one would have expected during the 2010/11 La Niña event. It finally responded during the last quarter.

Figure 2

################################

For more information on the effects of ENSO on global Ocean Heat Content, refer to ENSO Dominates NODC Ocean Heat Content (0-700 Meters) Data and to the animations in ARGO-Era NODC Ocean Heat Content Data (0-700 Meters) Through December 2010.

BASIN TREND COMPARISONS

Figure 3 and 4 compare OHC anomaly trends for the ocean basins, with the Atlantic and Pacific Ocean also divided by hemisphere. Figure 3 covers the full term of the dataset, 1955 to present, and Figure 4 shows the ARGO-era data, starting in 2003. The basin with the greatest short-term ARGO-era trend is the Indian Ocean, but it has a long-term trend that isn’t exceptional. (The green Indian Ocean trend line is hidden by the dark blue Arctic Ocean trend line.) The basin with the greatest rise since 1955 is the North Atlantic, but it also has the largest drop during the ARGO-era. Much of the long-term rise and the short-term flattening in Global OHC are caused by the North Atlantic. If the additional long-term rise and the recent short-term decline in the North Atlantic OHC are functions of additional multidecadal variability similar to the Atlantic Multidecadal Oscillation, how long will the recent flattening of the Global OHC persist? A couple of decades?

Figure 3

################################

Figure 4

################################

Further discussions of the North Atlantic OHC anomaly data refer to North Atlantic Ocean Heat Content (0-700 Meters) Is Governed By Natural Variables. And if you’re looking into natural impacts on OHC, also consider North Pacific Ocean Heat Content Shift In The Late 1980s.

ARGO-ERA MODEL-DATA COMPARISON

Many of you will recall the discussions generated by the simple short-term comparison graph of the GISS climate model projection for global OHC versus the actual data, which is comparatively flat. The graph is solely intended to show that since 2003 global ocean heat content (OHC) anomalies have not risen as fast as a GISS climate model projection. Tamino, after seeing the short-term model-data comparison graph in a few posts, wrote the unjustified Favorite Denier Tricks, or How to Hide the Incline. I responded with On Tamino’s Post “Favorite Denier Tricks Or How To Hide The Incline”. And Lucia Liljegren joined the discussion with her post Ocean Heat Content Kerfuffle. Much of Tamino’s post had to do with my zeroing the model-mean trend and OHC data in 2003.

While preparing the post GISS OHC Model Trends: One Question Answered, Another Uncovered, I reread the paper that presented the GISS Ocean Heat Content model: Hansen et al (2005), “Earth’s energy imbalance: Confirmation and implications”. (PDF) Hansen et al (2005) provided a model-data comparison graph to show how well the model matched the OHC data. Figure 5 is Figure 2 from that paper. As shown, they limited the years to 1993 to 2003 even though the NODC OHC data starts in 1955. Hansen et al (2005) chose 1993 as the start year for three reasons. First, they didn’t want to show how poorly the models hindcasted the early version of the NODC OHC data in the 1970s and 1980s. The models could not recreate the hump that existed in the early version of the OHC data. Second, at that time, the OHC sampling was best over the period of 1993 to 2003. Third, there were no large volcanic eruptions to perturb the data. But what struck me was how Hansen et al (2005) presented the data in their time-series graph. They appear to have zeroed the model ensemble mean and the observations at 1993.5. The very obvious reason they zeroed the data then was so to show how well OHC models matched the data from 1993 to 2003.

Figure 5

################################

My ARGO-era model-data comparison graph is also zeroed at the start year, 2003, but I’ve done that to show how poorly the models now match the data. Hansen et al (2005) zeroed at 1993 to show how well the models recreated the rise in OHC from 1993 to 2003 and I’ve zeroed the data in 2003 to show how poorly the models match the data after that. I’m not sure why that’s so difficult to accept for some people. The reality is, the flattening of the Global OHC anomaly data was not anticipated by those who created the models. This of course raises many questions, one of which is, if the models did not predict the flattening of the OHC data in recent years, much of which is based on the drop in North Atlantic OHC, did the models hindcast the rise properly from 1955 to 2003?

Figure 6 compares the ARGO-era Ocean Heat Content observations to the model projection, which is an extension of the linear trend determined by Hansen et al (2005), for the period of 1993 to 2003. Over that period, the modeled OHC rose at 0.6 watt-years per year. I’ve converted the watt-years to Gigajoules using the conversion factor readily available through Google: 1 watt years = 31,556,926 joules. The model projection has risen at a rate that’s 7 times higher than the observations since 2003.

Figure 6

I asked the question in Figure 6, If The Observations Continue To Diverge From The Model Projection, How Many Years Are Required Until The Model Can Be Said To Have Failed? It’s really a moot point. Hansen et al (2005) shows that the model mean has little-to-no basis in reality. They describe their Figure 3 (provided here as Figure 7 in modified form) as, “Figure 3 compares the latitude-depth profile of the observed ocean heat content change with the five climate model runs and the mean of the five runs. There is a large variability among the model runs, revealing the chaotic ‘ocean weather’ fluctuations that occur on such a time scale. This variability is even more apparent in maps of change in ocean heat content (fig. S2). Yet the model runs contain essential features of observations, with deep penetration of heat anomalies at middle to high latitudes and shallower anomalies in the tropics.” I’ve deleted the illustrations of the individual model runs in Figure 7 for an easier visual comparison of the observations and the model mean graphics. I see no similarities between the two. None.

Figure 7

COMPARISON OF OHC ANOMALY DATA BY HEMISPHERES

I don’t recall presenting the OHC anomalies for the Northern and Southern Hemisphere on the same graph in earlier posts, so here they are in Figure 8. As always, the Southern Hemisphere data has to be taken with more than a grain of salt. There were very few observations in the Southern Hemisphere prior to the ARGO era, especially south of the tropics. But what I found interesting was the major divergence after the 1997/98 El Niño. The Northern Hemisphere data rises significantly, but there is a minor dip and rebound in the Southern Hemisphere data at that time.

Figure 8

So I subtracted the Southern Hemisphere OHC anomaly data from the Northern Hemisphere. Refer to Figure 9. The timing of the large variations appear to coincide with El Niño-Southern Oscillation (ENSO) events. I checked and found that they did, but the results were surprising.

Figure 9

Figure 10 is a gif animation that compares inverted and scaled (-0.1) NINO3.4 SST anomalies (a commonly used proxy for the timing and magnitude of ENSO events) and the difference between Northern and Southern Hemisphere OHC anomalies. The animation presents the inverted NINO3.4 SST anomalies shifted up and down. I’ve done this to align them with the corresponding changes in the hemispheric difference for the significant 1972/73, 1982/83, and 1997/98 El Niño events. The hemispheric difference leads the NINO3.4 SST anomalies during those ENSO events.

Figure 10

I have not carried the investigation any farther. Hopefully soon.

THE HEMISPHERES AND THE OCEAN BASINS

(11) Northern Hemisphere

#################################

(12) Southern Hemisphere

#################################

(13) North Atlantic (0 to 75N, 78W to 10E)

#################################

(14) South Atlantic (0 to 60S, 70W to 20E)

#################################

(15) North Pacific (0 to 65N, 100 to 270E, where 270E=90W)

#################################

(16) South Pacific (0 to 60S, 120E to 290E, where 290E=70W)

#################################

(17) Indian (60S-30N, 20E-120E)

#################################

(18) Arctic Ocean (65 to 90N)

#################################

(19) Southern Ocean (60 to 90S)

SOURCE

All data used in this post is available through the KNMI Climate Explorer:

http://climexp.knmi.nl/selectfield_obs.cgi?someone@somewhere