>Over the past few months, I’ve been posting weekly or monthly Global SST anomaly maps that I create at the NOAA NOMADS website from the OI.v2 dataset.
Figure 1 is the SST anomaly map for the week centered on July 19, 2009.
Another provider of SST anomaly maps that appears to use the OI.v2 SST anomaly data is Unisys.
To confirm the dataset, I had emailed Unisys asking for the source of their data a number of months ago. They chose not to reply. But there are a limited number of satellite-derived SST datasets, and as you will see, Unisys does not appear to use NESDIS data. Note that Unisys excludes the extreme high-latitude data. Figure 2 is a gif animation of the two reading dates available in their archive…
…for the same week.
I originally wrote, Note that Unisys does not keep the temperature scale constant, which creates problems when trying to compare changes from one date to the next.” But Lawrence du Plessis wrote in the comments, “I think you are mistaken to say that ‘Unisys does not keep the temperature scale constant, which creates problems when trying to compare changes from one date to the next’. The number labels move around a bit, but it looks to me that the colors are for the same numbers on each chart.”
I’ve run back in time through the Unisys archivesand I believe Lawrence is correct.
Regularly on blogs, someone posts a link to an NESDIS SST anomaly map. Many times the blogger is someone wondering about the difference between the Unisys and NESDIS maps when the two datasets assumedly use the same technology satellites, the AVHRR satellites, and both are compared by their providers to in-situ data from ships and buoys. But there are other bloggers who use the NESDIS maps for other reasons. A gif animation of the two NESDIS maps that cover the same period as the OI.v2-based map above are shown in Figure 3. The NESDIS uses vibrant oranges and reds that indicate SST anomalies in the Arctic in excess of 4 deg C. Curiously, those elevated anomalies are there every July and August in the Arctic, if one bothers to go back in time for the past three to four years. But the OI.v2 SST anomaly maps, Figures 1 and 2, shows only a few areas with SST anomalies over 3 deg C. Compare the areas north of the Bering Strait and west of Greenland in the three presentations.
Granted, the map I’ve posted illustrates SST anomalies averaged over a whole week, while the NESDIS maps are snapshots on two days of that week. But the locations of the elevated SST anomalies in the Arctic aren’t shifting over the week in the NESDIS maps, so the averaging in the weekly data is having little to no effect on the Arctic SST anomaly presentation. Also, the Unisys data confirms the lower SST anomalies in the daily snapshots.
WHAT ARE THE DIFFERENCES?
On the non-technical side, there are obvious differences in the color scale. The NESDIS graph uses the “warm” yellows to indicate a temperature range of 0 to +0.5 deg C. The Unisys data “zeroes” their data in cool-appearing blues. And the OI.v2 SST anomaly map I provide shows white from -0.5 to +0.5 deg C. But there are differences in the methodologies used to create the two datasets.
The NESDIS SST anomaly data is described under a webpage titled “NOAA Coral Reef Watch – Methodology, Product Description, and Data Availability of NOAA Coral Reef Watch (CRW) Operational and Experimental Satellite Coral Bleaching Monitoring Products.”
What’s so special about the NESDIS dataset? It uses only nighttime SST data. Why? The answer lies within the description of the data.
“Nighttime-only satellite SST observations are used to eliminate diel variation caused by solar heating at the sea surface (primarily at the “skin” interface, 10-20 um) during the day and to avoid contamination from solar glare. Compared with daytime SST and day-night blended SST, nighttime SST provides more conservative and stable estimate of thermal stress conducive to coral bleaching.”
So the NESDIS data excludes the daytime observations to eliminate the variations that occur over a 24-hour period and to eliminate solar glair. The OI.v2 SST data uses both nighttime and daytime SST measurements and as they explain, “The AVHRR instrument has three infrared (IR) channels. Due to noise from reflected sunlight (sun glint), only two channels can be used during the day. However, at night the three IR channels are used because the residual noise is lower.” The OI.v2 data accounts for the glair problem and samples SST anomalies over the full course of the day.
WHAT ABOUT THE HIGH LATITUDES?
The NESDIS webpage cautions users about high latitudes of the “Coral Reef Watch” SST dataset: “Note that these anomalies are somewhat less reliable at high latitudes where more persistent clouds limit the amount of satellite data available for deriving both accurate SST analysis field and climatologies.”
NOAA’s OI.v2 SST anomaly data on the other hand goes to special lengths to fill in the high latitude SST anomalies. Refer to page 7 of the Reynolds et al 2002 paper “An Improved In Situ and Satellite SST Analysis for Climate.”
It reads, “A large potential error occurs near the sea ice edge where in situ observations tend to be sparse because of navigation hazards and satellite observations tend to be sparse due to cloud cover. Thus, using sea ice data to generate simulated SSTs in the marginal ice zone (MIZ) helps fill in a region with limited data.” They then go on to describe those adjustments.
So what does the additional step the providers of the OI.v2 data take to infill the high latitude SST data actually do? Apparently it lowers SST anomalies in the Arctic.
A NOTE ABOUT THE SEASON CYCLE IN SST ANOMALIES
A point that many bloggers miss is the seasonal cycle in SST anomalies. It is very obvious in animations.
Note how the warm anomalies “cycle” into the Northern high latitudes during June, July, and August, but then shift down to the Southern high latitudes during December, January, and February. It happens every year. Also, as Arctic ice melted after the 1997/98 El Nino, the area of elevated Arctic SST anomalies increased, so this adds to the illusion of higher and higher Arctic SST anomalies, when Arctic SST anomalies have been dropping for the past few years, Figure 4.