How to handle location code?

[emthompson-usgs]

Currently, we basically ignore location code. Presumably it should be useful for distinguishing between different locations on a downhole or structural array but this doesn't seem to be reliable. For example, network NP station 8043 looks like this:

Channel Location
HNZ     37
HNN     38
HNE     39

If we actually treated these as separate locations then we would not be able to process this record for any metrics that require two horizontal components (which is most of them). If we ignore location code, then we have the potential to group channels together and treat them as though they are colocated even though they are not. The latter problem can be avoiding by screening out data that has been grouped together by network/station/instrument and still has more than 3 channels. I think for now, I'm going to use this approach but I'm not really happy with it.

Does anyone (especially @baagaard-usgs) have thoughts on the optimal approach here?

[baagaard-usgs]

With metadata from StationXML, the longitude, latitude, elevation, and depth are given on a channel by channel basis. If this information is available it could be used to determine which channels are collocated. A conservative approach would be to assume they are not unless the channel information indicates they are. I doubt we will drop many stations with this requirement.