[EAS] National Weather Service Message Flooding

[Mike McCarthy]

John,

WEA addresses the Geo-positining problem of the receiver fairly well 
already. The polygon's already have geo-coordinated corners and already 
outline the area of greatest threat. The cell folks who wisely implement 
the system have the "location of the device" issue figured out. IIRC, 
many of the the larger weather events to which you refer already have 
reference coordinates (eg., hurricane eye) in the reports and the 
targeted polygons convey the area of greatest hazard and threat in those 
cases as well.

The challenge is extending this system to the broadcast world and is a 
complex one to be certain....  And I appreciate the aspect of being in 
unfamilar areas when traveling during hazardous conditions.  I'm a 
trained spotter of 30+ years and NWS amateur station operator besides 
being a member of the local AMS.

Your SVR example is another why total county only triggering mechanism 
is real challenge.  Especially for exceptionally large counties like San 
Bernadino, CA. and Cook, Mn. to name but a couple.  However, the NWS can 
not simply base issuance of warnings on population centers or ignore 
issuing a warning because the event is outside a population center. It's 
still a ***populated*** area no matter how sparse per sq. mile.  IOW, 
meaning tax paying citizens live there just like we might live is a big 
metro area. They gotta issue them...

In tornado land, the best location tool is the radar, my prior comments 
not withstanding. Problem is the radar is always a few minutes behind 
and ground verification isn't often available. Particularly if a tornado 
occurs at night and/or makes a sharp turn (think Oklahoma this past 
spring). Then there is the accuracy at a distance issue.

So anything sent will be immediately perishable, create more message 
flooding and resultant confusion.  Right now the NWS is sending a new 
semi-official product "Tornado Emergency" providing updates as you 
describe.  But again, there is a component to message flooding where 
there are multiple TOR's in effect with accordingly similar numbers of 
"Emergency" messages coming out in a major outbreak.

For larger events, such as tropical and coastal storms, the dynamics of 
storm life and sample are such that any real time messaging other than 
radar based would again be immediately perishable. Squall dynamics can 
easily change between the 3 hour satellite or buoy samples. And as 
Katrina quite capably showed us, the best forcasting and real time data 
still can not provide pin point accuracy to the storm's movement or 
where it's strongest forces will be felt when they come ashore.

Flood hazards in areas prone to flash flooding are such that even near 
real time data may not be enough to save lives.  It is these areas which 
need the greatest study of hydrological features so as to define what 
level of rainfail in a given time frame at a given location and area 
triggers a warning.  Some places, like the desert, need only an inch of 
rain in a small area to create life hazard conditions within minutes of 
the event's start. Other areas can tollerate many inches before flood 
conditions develop over time. This is there the NWS needs to break down 
it's one size fits all silo (thanks Bill Ruck) and allow input from 
local EMS folks on what constitutes a flash flood.

However, in so far as the breaking down the boarders aspect, I still 
believe subdividing of counties offers the fastest approach as EAS 
already has that built into it's system.  How that's done is a subject 
of much needed discussion across all partners and silo's.

MM

On 9/28/2013 11:24 PM, John Willkie wrote:
> It seems to me that identifying weather emergencies by county or county subdivision (of any type) is a VERY bad idea.  While the NWS may respect county boundaries, their measures appear to be antiquated and extremely coarse.
>   
> Weather emergencies are dynamic and simply they do not respect geographic, toplogical or political boundaries.  A system that alerts based on such arbitrary boundaries will always result in false triggers (which will tend to be ignored) and viewer/listener fatigue, such as in my area when there are severe thunderstorm warnings (recently I saw two in the same newscast) for areas of my county in the desert, two ecosystems and almost 100 miles away from the bulk of the county population.
>   
> I see two ideal approaches.  The first is dynamic polygons.  For severe thunderstorms, the first area to identify is the actual area where the rain is falling.  A second set of polygons could identify the actual affected watersheds, since water tends to flow downhill into watersheds.  Perhaps even a third set of polygons identifying areas where fires have eradicated some or all of the underbrush "mud warnings."
>