The Pacific Mean SST – Watts Up With That?

By Andy May

I originally planned to discuss the North Pacific Index (NPI) in this post, but while researching it, I discovered something interesting about Pacific sea surface temperature (SST) and how it relates to the HadCRUT5 global average surface temperature. As a result, this post is about the total Pacific mean SST and its correlation to HadCRUT5.

Of all the Pacific Oscillations I studied the NPI was the most correlated with HadCRUT5, but it only ranked 7^th overall, and three North Atlantic Oscillations ranked above it. This is odd since the Pacific covers 33% of the Earth, as opposed to 8% for the North Atlantic as shown in Table 1.

All the common Pacific Oscillations are useful in explaining past climate and weather events in the Pacific Basin and they also explain many environmental processes, such as the abundance of many fish (Lluch-Belda, et al., 1989), (Mantua, Hare, Zhang, Wallace, & Francis, 1997), and here. But none of them characterize the whole ocean and only a few of them work with Pacific SSTs. Out of curiosity, I tried regressions of the total Pacific mean SST from HadSST4, ERSST5, and HadISST against HadCRUT5 and found that HadSST4 correlated best (see a discussion of these SST datasets here), which is no surprise. It makes up 33% of the HadCRUT5 data. ERSST5 and HadISST use almost the same raw data as HadSST4, but both are interpolated and extrapolated to have complete or nearly complete global SST grids. They also process the data differently, especially in the polar regions, where HadSST4 has many null grid cells. HadISST, at 90%, is almost as well correlated as HadSST. ERSST5 is 80% correlated.

Compare this to the ERSST5 AMO (76%) and the ERSST5 North Pacific alone at 79%. The AMO does quite well considering it is only 8% of Earth’s surface, compared to 33% for the total Pacific Ocean. There are ten commonly cited Pacific Oscillations, teleconnections, and indices as shown in Table 2.

The odd thing about the list in table 2 is that none of them cover the entire Pacific Ocean, although the TPI comes close. An obvious question is how does the mean Pacific SST compare to HadCRUT5? The R² statistics in Table 1 are useful, but as we have seen in this series it is not enough. The acid test of correlation, especially when dealing with time series, is to examine a graph of the data series being compared. We want to see how trend direction changes compare between series. The calculated, and area-weighted series are shown in figure 1.

It is not surprising that the total Pacific HadSST4 and ERSST5 mean SST records have the best correlation with HadCRUT5. It is a little surprising that the ERSST5 AMO, covering only one-fourth the area of the Pacific, does so well and nearly as well as the North Pacific alone (16% of Earth).

But we need to examine the graph in figure 1 more closely, the devil is in the details. All the anomalies in figure 1 are from their respective mean values from 1961-1990. Thus, they closely agree in that period. I find it suspicious that HadCRUT5 is the second highest value from around 1998 to 2024 as well as the lowest value from 1850 to 1905. It only joins the remaining means from 1905 to the mid-1990s.

It is true that land warms and cools faster than water due to its lower heat capacity, but land only occupies 29% of Earth’s surface, less than the area covered by the Pacific. Should this affect the HadCRUT5 trend over periods as long as 1998-2024 and 1850-1905? The fact that the difference is negative in the 19^th century and positive in the 21^st is suspicious, and a little too convenient for the “consensus” by far.

The Pacific is the world’s largest ocean, and one would think it has a huge influence on the HadCRUT5 global mean surface temperature (GMST), but if so, it isn’t clear in figure 1. It also isn’t clear in the commonly cited oscillations and indices listed in table 2 or any other Pacific oscillation. All I see, from a global perspective, when I look at the Pacific oscillations is a confusing mess. They are very important regionally, less so globally. More on this in later posts. These oscillations have a significant impact on North & South American weather and weather in the Far East, but they do not correlate with HadCRUT5 very well.

Could HadCRUT5 be the problem? HadCRUT5 is virtually identical to the BEST global average surface temperature record (Rohde & Hausfather, 2020) relied upon a lot in AR6. HadCRUT5 is also similar to other records of estimated global surface temperature, so we don’t think it is a simple error in data gathering, but it could be due to errors in processing and “correcting” the data as discussed here. Figure 2 shows how the mean total Pacific SST correlates with HadCRUT5.

Figure 2 compares the mean total Pacific SST from HadSST, HadISST, and ERSST to HadCRUT5. Figure 3 plots the difference between HadCRUT5 and the mean Pacific SST. The largest positive difference (HadCRUT5 larger than the Pacific mean) occurs from 2000 to the present and the largest negative difference occurs in the 19^th century, which has the result of inflating the global surface warming rate.

Surprisingly the largest standard deviation of the total Pacific mean SST since 1941 occurs during the modern era when we have ARGO float data, which are the highest quality SST measurements. The only period of comparable standard deviations is from 1870 to World War I when SST measurements were very sparse and of low quality. There is one spike in 1941 that is as bad as the peak in 2022, but otherwise every yearly standard deviation since 1923 is below the values since 2016, very odd.

This makes little sense, the equipment used from 2016 to today is the best that has ever been deployed, further the modern era has satellite temperature measurements, which were unknown before 1978. It is well known that the worst period for SST data is during World War II (WWII), yet the standard deviation is minimal in that period. Why should HadCRUT5 and the total Pacific mean SSTs agree best during World War II? The collected data was awful then, it can only mean the three reconstructions (HadISST, HadSST, and ERSST) used the same methods to deal with the bad data in WWII, it cannot mean the estimates are more accurate.

Many will remember figure 4, which compares HadCRUT3, 4, and 5. This illustration first appeared when HadCRUT5 first came out and it illustrates how different processing methods have increased the global average surface temperature for nearly every year from 2000 to 2014. The data used by the Hadley Centre didn’t change between 2000 and 2014, only the processing and “error corrections.”

Figures 2 to 4 show that the warming rate from HadCRUT5 is very suspect. It also shows that modern estimates of SST are probably getting worse as the data is getting better. It is highly unlikely that three different estimates of Pacific mean SST would be more different since 2005 when data from thousands of highly accurate Argo floats became available. It is also dubious that the modern era is less accurate than the World War II period when the data was awful. Something is wrong.

Discussion

As figures 2 and 3 make clear, most of the time from around 1910 through 1975, the total Pacific mean SST anomaly, and its standard deviation track well with HadCRUT5. Before 1910 and after 1975 HadCRUT5 is outside the Pacific mean standard deviation and before 1910 it is below the Pacific mean temperature and after 1975 it is above. I suppose that there could be some climate influence causing this, but it seems unlikely. Considering the huge heat capacity in the Pacific, relative to the global atmosphere, the difference in warming trends between the Pacific and the global surface for these multidecadal periods is not credible.

As I discussed and documented here, the World War II period was a period of great error in SST data, both because of the war itself and because of the transition from measuring SST in insulated buckets dipped into the ocean to measuring it with instruments in ships’ engine cooling water intake ports. This should be the period when different estimates of mean total Pacific SST are maximally different, not minimally different.

The modern era, when we have Argo floats and abundant tethered ocean buoys, should have the best data and the least uncertainty, but figures 2 and 3 show the opposite. The whole issue of how well or how poorly SST is estimated is discussed in more detail here. It seems likely that there is a problem with the HadCRUT5 reconstruction of global surface temperature. There are also problems with estimating Pacific mean temperature, but why should the comparisons in figures 2 through 4 indicate that the error in these estimates is increasing? It seems very odd.

Useful references related to this post are listed below, See here and here for more information on the topics discussed and the references below. In the next post I will discuss the North Pacific Index (NPI), as originally planned.

Brönnimann, S. (2003). A historical upper air-data set for the 1939–44 period. International Journal of Climatology, 23(7), 769-791. doi:10.1002/joc.914

Brönnimann, S., & Luterbacher, J. (2004b). Reconstructing Northern Hemisphere upper-level fields during World War II. Climate Dynamics, 22, 499-510. doi:10.1007/s00382-004-0391-3

Brönnimann, S., Luterbacher, J., & Staehelin, J. (2004). Extreme climate of the global troposphere and stratosphere in 1940–42 related to El Niño. Nature, 431, 971–974. doi:10.1038/nature02982

Freeman, E., Woodruff, S., Worley, S., Lubker, S., Kent, E., Angel, W., . . . Smith, S. (2017). ICOADS Release 3.0: a major update to the historical marine climate record. Int. J. Climatol., 37, 2211-2232. doi:10.1002/joc.4775

Hegerl, G. C., Brönnimann, S., Schurer, A., & Cowan, T. (2018). The early 20th century warming: Anomalies, causes, and consequences. WIREs Climate Change, 9(4). doi:10.1002/wcc.522

Huang, B., Thorne, P. W., Banzon, V. F., Boyer, T., Chepurin, G., Lawrimore, J. H., . . . Zhang, H.-M. (2017). Extended Reconstructed Sea Surface Temperature, Version 5 (ERSSTv5): Upgrades, Validations, and Intercomparisons. Journal of Climate, 30(20). doi:10.1175/JCLI-D-16-0836.1

IPCC. (2021). Climate Change 2021: The Physical Science Basis. In V. Masson-Delmotte, P. Zhai, A. Pirani, S. L. Connors, C. Péan, S. Berger, . . . B. Zhou (Ed.)., WG1. Retrieved from https://www.ipcc.ch/report/ar6/wg1/

Kennedy, J. J., Rayner, N. A., Smith, R. O., Parker, D. E., & Saunby, M. (2011). Reassessing biases and other uncertainties in sea surface temperature observations measured in situ since 1850; 1. Measurement and sampling uncertainties. Journal of Geophysical Research, 116. Retrieved from https://agupubs.onlinelibrary.wiley.com/doi/full/10.1029/2010JD015218

Kennedy, J. J., Rayner, N. A., Smith, R. O., Parker, D. E., & Saunby, M. (2011b). Reassessing biases and other uncertainties in sea surface temperature observations measured in situ since 1850: 2. Biases and homogenization. J. Geophys. Res., 116. doi:10.1029/2010JD015220

Kennedy, J., Rayner, N. A., Atkinson, C. P., & Killick, R. E. (2019). An ensemble data set of sea-surface temperature change from 1850: the Met Office Hadley Centre HadSST.4.0.0.0 data set. JGR Atmospheres, 124(14). Retrieved from https://agupubs.onlinelibrary.wiley.com/doi/abs/10.1029/2018JD029867

Lluch-Belda, D., Crawford, R. J., Kawasaki, T., MacCall, A. D., Parrish, R. H., Schwartzlose, R. A., & Smith, P. E. (1989). World-wide fluctuations of sardine and anchovy stocks: the regime problem. South African Journal of Marine Science, 8(1), 195-205. doi:10.2989/02577618909504561

Mantua, N. J., Hare, S. R., Zhang, Y., Wallace, J. M., & Francis, R. C. (1997). A Pacific Interdecadal Climate Oscillation with Impacts on Salmon Production. Bull. Amer. Meteor. Soc, 78, 1069-1080. Retrieved from https://journals.ametsoc.org/view/journals/bams/78/6/1520-0477_1997_078_1069_apicow_2_0_co_2.xml

Rayner, N. A., Brohan, P., Parker, D. E., Folland, C. K., Kennedy, J. J., Vanicek, M., . . . Tett, S. F. (2006). Improved Analyses of Changes and Uncertainties in Sea Surface Temperature Measured In Situ since the Mid-Nineteenth Century: The HadSST2 Dataset. J. Climate, 19, 446-469. doi:10.1175/JCLI3637.1

Rohde, R. A., & Hausfather, Z. (2020). The Berkeley Earth Land/Ocean Temperature Record. Earth System Science Data, 12(4). doi:10.5194/essd-12-3469-2020

Trenberth, K., & Hurrel, J. (1994). Decadal atmosphere-ocean variations in the Pacific. Climate Dynamics, 9, 303-319. doi:10.1007/BF00204745

Related