I’ve moved to WordPress. This post can now be found at Part 3 of Comparison of GISTEMP and UAH MSU TLT Anomalies###############
Or The Comparison of OI.v2 Sea Surface Temperature Anomalies and the UAH MSU TLT Anomalies for the Same Ocean Surface Areas
In the first part of this post, Part 1 of Comparison of GISTEMP and UAH MSU TLT Anomalies,
I illustrated the linear trends of GISTEMP Surface Temperature Anomalies and UAH MSU TLT anomalies for specific sections of the globe. Some were remarkably close. Others differed greatly. In Part 2 of Comparison of GISTEMP and UAH MSU TLT Anomalies, the linear trends of GISTEMP Land Surface Temperature Anomalies and UAH MSU TLT anomalies over the same continental land areas were examined. As in the first post, a few of the areas presented trends that were remarkably similar, while others differed greatly. Many of the datasets with the substantial divergences were those where the land surface areas had poor surface station coverage, requiring infilling by GISS.
In this post, I’ll compare the SST anomalies of the dataset GISS has used since December 1981 (OI.v2 SST) and the UAH MSU TLT Anomalies for the same ocean areas. This post differs from the other two in a very significant way. For the areas selected, the sea surface area coverage is complete in both datasets.
AREAS FOR THE COMPARISONS
Along with many others, the KNMI Climate Explorer website provides access to OI.v2 SST data. The other dataset used in this post, UAH MSU TLT data, is also available through KNMI. Unfortunately, as explained in Part 2 of this post, it is not separated into land and ocean subsets. And due to the shape of the oceans, there is no way to gather the TLT anomalies over a complete ocean without also retrieving TLT anomalies over land. Because of this, for this comparison of SST anomalies and the TLT anomalies over those same ocean areas, I subdivided the data into smaller areas of the oceans to minimize any influence from land data.
Figure 1 illustrates the global grids used in the comparisons of OI.v2 SST and UAH MSU TLT anomalies. The coordinates are listed on the graphs that follow. The SST and TLT data in the following graphs represent major portions of the North and South Pacific, North and South Atlantic, and the South Indian Oceans.
SST and TLT for three of the five areas track to a reasonable extent. The first of those illustrated is the North Atlantic, Figure 2. The UAH MSU TLT anomalies exaggerate the 1997/98 El Nino, and there’s a short divergence beginning in 1990, before the Mount Pinatubo eruption. For the North Atlantic, the UAH MSU TLT anomaly linear trend (0.255 deg C/decade) is less than the SST anomaly linear trend (0.298 deg C/decade).
The South Pacific SST and TLT data, Figure 3, also show general agreement over the term of the data. Again, the TLT anomalies rise much more than the SST anomalies during the 1997/98 El Nino. And again, there is a divergence between the two datasets around the time of the 1991 Mount Pinatubo eruption. The TLT anomalies drop in response to the volcanic aerosols, but the SST anomalies rise, appearing to be impacted more by the 1991/92 El Nino. The UAH MSU TLT anomaly linear trend (0.104 deg C/decade) for the South Pacific is greater than the SST anomaly linear trend (0.063 deg C/decade).
The SST anomalies for the South Indian Ocean, Figure 4, show a higher response to the 1986/87/88 El Nino than the UAH MSU TLT anomalies. The SST anomalies also make a pronounced dip and rebound from 2004 to mid-2006 that does not appear in the TLT data. These two differences are the likely reasons why the SST anomaly linear trend (0.058 deg C/decade) in the South Indian Ocean is significantly less than the UAH MSU TLT anomaly linear trend (0.118 deg C/decade).
Figure 5 is the comparative graph of the SST anomalies and TLT anomalies for a large portion of the South Atlantic. The first thing that stands out is the limited response of the SST anomalies to the 1997/98 El Nino. The second: the reaction of both the TLT and SST anomalies to the 1991 Mount Pinatubo eruption is pronounced. The two datasets diverge on occasion. Between 1984 and 1990, TLT anomalies dropped below SST anomalies twice, and between 2001 and 2007, TLT anomalies rose above SST anomalies. The end result: the UAH MSU TLT anomaly linear trend (0.1 deg C/decade) is significantly greater than the SST anomaly linear trend (0.022 deg C/decade).
The SST anomalies for the portion of the North Pacific used in this post, Figure 6, had a counterintuitive decline during the 1997/98 El Nino, while the TLT anomalies rose as one might expect. The exaggerated responses of the TLT anomalies to the El Chichon and Mount Pinatubo eruptions also bias the TLT anomaly trends downward during the first half of the time span. These result in a significantly higher UAH MSU TLT anomaly linear trend (0.255 deg C/decade) than the linear trend of the OI.v2 SST anomalies (0.132 deg C/decade).
I’ve provided Figure 7 to illustrate the cause for the unusual decline in the North Pacific SST anomaly data during the 1997/98 El Nino. It’s a screen capture from the JPL Sea Surface Height animation I’ve used in many of my videos. Most of the elevated SST anomalies were east of the area used for this post, but the dataset included a good portion of the Pacific Warm Pool where SST anomalies dropped.
And here’s a link to the video on YouTube:
The GISTEMP Land Surface Temperature and UAH MSU TLT data (with the exception of the Antarctic Land TLT data) are available through the KNMI Climate Explorer website:
The screen capture was taken from an animations available from the NASA Jet Propulsion Laboratory:http://sealevel.jpl.nasa.gov/gallery/videos-ssh-movies.html
Specifically, it was taken from: http://sealevel.jpl.nasa.gov/gallery/tiffs/backup/videos/tpj1global.mpeg
There seems to be a greater correlation with UAH and the ocean indices than I had noticed before.
Do you think the newer ocean datasets have been tuned towards the LT observation datasets or were the correlations there all the time.
I charted the monthly AMO versus RSS awhile ago and the correlation is (I have to say, unexpected) high.
Maybe all the datasets look like this on a monthly basis. Can you tone down the smoothing and see if something is there.
Bill Illis: I'll try to post the raw SST vs TLT comparisons in a couple of days.
Post a Comment