[EAS] IPAWS and NOAA

[Sean Donelan]

On Fri, 6 Jul 2018, Dave Turnmire wrote:
> all the relevant counties for that alert.? But... a given NWR transmitter 
> will only include the counties within its coverage area.

Not only with IPAWS, but also all WRSAME transmitters with overlapping 
alert areas have the same problem. In the 1980s, when SAME was designed, 
there fewer weather radio transmitters and almost none with overlapping 
alert location areas. With EAS, all the participants in the same 
area need to monitor the same WRSAME transmitter. If the LP-1/LP-2 
stations and local broadcasters monitor different overlapping NWS 
transmitters, you will broadcast duplicate alerts.

For example, around Washington DC EAS area, all the EAS stations monitor 
the same NOAA KHB36 transmitter in Manassas, VA. NWS transmits weather 
warnings for the entire EAS area on KHB36. Other WR stations might 
transmit the same alert, but with different, overlapping SAME location 
lists. Because all the EAS stations get the KHB36 version, they all 
de-dup the identical WRSAME ZCZC string (except the Washington DC LP-1 
station, but that's a different discussion).

https://www.weather.gov/lwx/nwrmap

For 99% of short-fused weather alerts, its not a problem because tornados, 
flash floods, severe thunderstorms, etc. have only one or a few counties 
in each alert. Much less chance of overalpping alerts. Regional alerts 
like hurricanes, tsunami, winter storm have more overlaps and more than 
31 counties.

Without changing the WRSAME and EAS protocol and breaking millions of 
weather radios, the only way to guarantee CAP/EAS/WRSAME duplicate 
matching is separate IPAWS queues (i.e. URL or API parameter) retrieving 
different CAP content matching a specific WRSAME transmitter.

Because all EAS/CAP boxes support multiple CAP sources, an EAS box could 
support a specific URL to retrieve weather warnings formatted for the 
local WRSAME transmitter. Duplicate detection across EAS/WRSAME/CAP 
protocols would work because the resulting ZCZC string (including location 
list) matches the specific WRSAME transmitter and EAS area; even though 
the same alert might be transmitted by dozens of other WRSAME 
transmitters and via WEATHERWIRE.

I prefer putting most of the conversion work at the NOAA/IPAWS interface 
instead of thousands of EAS boxes in the field. Its easier to fix, when 
something needs fixing. I know, programmers will freakout the first time 
they hear this, because that is not how it is normally done.

New alert channels should be able to support full digital CAP 
distribution, and use the CAP message ID to de-dup. Only those alert 
channels needing backwards capability to de-dup EAS/WRSAME ZCZC strings 
would need separate URLs and local CAP message queues.