search

wmo-im / WSI

WIGOS Station Identifier
2 stars 1 forks source link

Secretariat vs. national WSI schemas for GOS stations #6

Open efucile opened 4 years ago

efucile commented 4 years ago

In 2016 the Secretariat assigned WSI to the stations present at that time in Vol. A. The schema used by the Secretariat is the following 0-20000-0-????? for synop stations with ????? = block + station number 0-20001-0-????? for raob stations with ?????= block + station number This was a one-off exercise to retire Vol. A and make operational OSCAR/surface. Members can assign WSIs using the following pattern 0-ISO country code - any one single digit- up to 16 alphanumeric characters and they have the flexibility to choose their own schema to do so. The fact that two different schemas (Secretariat and national) are going to be present at the same time can be seen as a problem and lead to some Members to create a double identifier for the same station. This is permitted, but not recommended. This is what is recommended in the Guide to WIGOS.

An observing facility may have several WIGOS station identifiers. Using OSCAR, it is possible to discover all the WIGOS station identifiers associated with that facility. In theory, this allows any of the possible WIGOS identifiers to be used in a report of an observation, but in practice, doing so would result in a lot of additional work for all users of the observation. A disciplined approach to using WIGOS station identifiers in a report will reduce the work for end-users.

An analysis of the issues related to having two identifiers for the same stations has to be carried out.

josusky commented 4 years ago

I am not sure if it is possible to add into the OSCAR surface catalogue a station with two identifiers. I have not seen such case so far. @efucile do you have an example that I could check? We already know that some countries prefer to keep using the "0-2000-0-IIiii" WSIs that they already have, while others prefer to have one numbering schema for all stations (thus prefer to use the "0-ISO country code" style). I do not think there is a chance for a wide consensus. I know that Israel is distributing their surface observations with WSIs, because for certain stations they do not have TSI. So I have checked some of the Israeli stations that historically had TSIs. For example there is record for "0-376-0-511" and "0-376-0-40170" that seem to refer to the same place (name and coordinates) but there are some subtle differences that suggests that theses are in fact two separated records in the OSCAR. That is not to blame Israel. In this case, they have the usual disadvantage of the pioneers. I think, we should urge WMO members to select and keep using one and only identifier for each station "forever". And possibly change the aforementioned formulation in the Guide to forbid duplicates.

golfvert commented 4 years ago

@josusky in issue https://github.com/wmo-im/WSI/issues/4 I am mentioning the "diversity" of options/solutions with regard to the implementation of WSI. When a standard/norm is too vague, it creates divergence and a messy situation. So, we need to reduce options. So, solutions that goes into better and stricter specifications have my strong preference.

efucile commented 3 years ago

Example of station with two WSIs Oran-Senia (Algeria) WSI 0-20000-0-60490 (primary) WSI 0-12-20000-60490 This is very creative as with the country code 12 for Algeria, in the non-primary WSI we have memory of the 20000.

jitsukoh commented 3 years ago

I think this issue should be solved before going to the “reporting WSI in BUFR” phase. If we change policy on how to assign WSI after Members start reporting observations with WSI, it would be a mess.

In my opinion, the problem is not having two identifiers for one station, but having messages from the station with two identifiers at the same time (parallel and/or mixed over time). The most we can expect Members to do is that Members choose either scheme (a 20000/200001 number or a number with the 2-digit country code) for each station once and stick with it in reporting observation. From the maintenance point of view of OSCAR/Surface, it is desirable if OSCAR/Surface has a functionality to "deprecate" WSIs that are not in use and to indicate which WSI should be referred instead.