[Realtime-feeds] Feed JSON field question

Jeroen van Aart jeroen at mompl.net
Tue Apr 15 23:24:30 UTC 2014


On 04/15/2014 03:24 PM, Fee, Jeremy wrote:
> Hello,
>
> IDs are most likely to change within the first 15-20 minutes as networks
> contribute their information.
>
> Each seismic network that locates an event assigns a unique ID specific to
> that network.  Our software associates all of these contributions together
> into one event, which can be referenced using any ID from any contributor.
>   The ID that appears in feeds is ID assigned by the most preferred
> contributor at the time the feed is generated.
>
> We run redundant, geographically separated systems that are not guaranteed
> to process information in the same order.  This approach eliminates a
> potential single point of failure and is eventually consistent.

> On Tue, Apr 15, 2014 at 3:57 PM, Matt Keller <mattrk at mattrk.com> wrote:

>> For my application, i just need to track an event with some kind of unique
>> identifier. I would not want to duplicate the event if the ID suddenly
>> changed. I'm just curious how the process works and if i need to worry
>> about it. I might also mention that i'm only monitoring the all_hour feed.
>> So it may not be a huge deal for that particular feed.
>>
>> Thanks again for everyone's input. As a programmer (and not a scientist),
>> this information has helped a lot.

Please note, having developed and released applications using usgs 
feeds, that it is not easily avoidable to occasionally have duplicate 
events. I believe it is the side effect of the procedure outline above. 
The main difference between these duplicates is a different ID, 
sometimes an updated magnitude level, depth and/or location.

I think that in code it is too much of a hassle to filter those out 
since it requires some actual thinking to notice them, as opposed to 
simple boolean decision making. This I think would tread more into the 
realm of Bayesian statistics, an effort which I personally would leave 
to such things as programming spamfilters.

I found that the drawback is minimal and not worth the effort. And only 
really noticeable when you use alerts to alert a user about a strong 
earthquake it may pop up more than one alert about the same event (each 
one updated). One could see this as a feature...

Hope that helps,
Jeroen

-- 
Earthquake Magnitude: 4.6
Date: 2014-04-15  21:08:21.870 UTC
Date Local: 2014-04-15 14:08:21 PDT
Location: 149km SE of Kirakira, Solomon Islands
Latitude: -11.3951; Longitude: 162.9021
Depth: 35 km | e-quake.org


More information about the Realtime-feeds mailing list