Many around the climate blogosphere have noted that UNISYS’ recent sea surface temperature anomalies were showing radically different values from various NOAA products.
I decided to reach out to UNISYS directly to find out what might be behind the discrepancies, mentioning that it was confusing that UNISYS was showing Hudson Bay cold, water off the East Coast of Russia frigid, and most of the Northern Hemisphere dramatically cooler than 6 weeks ago, when NOAA was showing nothing of the kind.
UNISYS’ weather program manager, Brian Hughes, sent along the following response:
After further thought and additional analysis, I’ve asked that the images be taken down temporarily.
What originally appeared to be a simple color bar/enhancement table issue looks to be an issue with our anomaly product itself. I took more looks at areas where our product is indicating cooler than normal, the corresponding NOAA product appears to show warmer. That tells me something is off with our processing.
In July, we had to switch to the higher resolution RTG-SST product as the input because we had been using a legacy SST product from NOAAPort that NWS discontinued in June. The SST anomaly product may be suffering from amplified cooling as we transition into NH winter, an error not originally seen when we first switched in the summer.
The dataset used to process and create the anomaly appears to also be an issue, perhaps our software is not calculating the correct temp since the switch to the RTG-SST hires.
We are going to evaluate this and work on a solution.