In June 2012 the NTSB released this Safety Alert about the actual age of "in cockpit" radar weather displays such as those received by satellite services or through FIS-B. This was later revised December 2015. This is the first time the FAA or NTSB has issued any guidance to pilots on the use of satellite-delivered weather products despite the fact that these systems have been around for the better part of a decade.
They are correct in saying that the date-time presented to the user is based on the time the data was broadcast (prepared) and does not represent the actual age of the product. In this statement, they suggest that the data could be as old as 20 minutes. However, I think this is a gross overstatement for datalink products from providers such as SiriusXM and WxWorx and for the free FIS-B ground-based system. On a visit to their Huntsville, Alabama office, I've spoken directly to the technical staff at WxWorx which supplies the data for the XM-delivered satellite weather mosaic. While they do agree the actual presentation is older than the date-time stamp on the product, it should not be any older than 10 minutes at the worst case during convective outbreaks. I've explained this numerous times, but I'll repeat it below.
There are 3 potential areas of latency...
(1) The volume scan of the individual radars
(2) The time it takes to composite and ship to XM
(3) The time it takes to uplink.
The time it takes to composite the data and ship to XM is just a few seconds. While the satellite-based systems are seriously bandwidth-challenged, the time it takes to uplink the product is about 20 to 30 seconds depending on the actual activity (number of radar returns). The greatest delay is in the volume scan of the actual radars. The image presented is a filtered composite reflectivity mosaic that consists of all elevation scans which encompasses a little more than five minutes of data.
WxWorx and SiriusXM do not use the NWS composite reflectivity image. They grow their own. They arbitrarily undertake compositing at a five minute interval, compositing the latest scan and going back in time so that all previous elevations are included. When convective precipitation is occurring, most radars are set into a mode with 14 elevations. If the radar has just finished the fourth elevation, for example, they will include this elevation scan, plus the previous 13 elevations spanning the previous 5 minutes or so. The composited image from each radar is then "sewn together" to produce a mosaic for the whole U.S. It is therefore possible that while one elevation scan (latest) is only "seconds" old, other elevations can be as old as 5 minutes upon compositing. The majority of the data is 3 minutes or less “old” so what reaches the pilot in the cockpit is typically less than 3 minutes and 30 seconds old on average considering the other delays.
Perhaps what is missing from the discussion is that the pilot will stare at this image for another 2.5 minutes until it is once again updated. So, in the worst case scenario, the mosaic shown to the pilot would be just shy of 8 to 10 minutes old. Now, of course, the pilot must also be aware of missed updates. This can be problematic at times. Therefore, a user's display could be 15 or 20 minutes old if one or two updates are missed, respectively.
I especially liked the NTSB's statement at the end of the document...
"The general issue of latency with in-cockpit NEXRAD is discussed in pilots’ guides, in industry literature, and on service providers’ websites. However, the NTSB has not found that such guidance contains details about the potential time difference between the age indicator and actual conditions."
I guess they are not readers of IFR magazine. I've discussed this numerous times here, here and here dating back to 2005. Just another case where the NTSB and/or FAA is simply out of touch with general aviation.
Most pilots are weatherwise, but some are otherwise™
Scott Dennstaedt, PhD
Weather Systems Engineer
CFI & former NWS meteorologist
