[Rda-users] Station Location coordinate systems and projection

[Steve Schroeder]

Brian and whoever else is interested,

I have been attempting to locate upper air stations worldwide accurately, and have found that all catalogs are inaccurate, generally due to low accuracy of data originally provided to people who manage the catalog, and not updating the locations when the stations move.  For example, the primary WMO catalog for global synoptic and upper air (radiosonde) stations can be downloaded at

http://www.wmo.int/pages/prog/www/ois/ois-home.htm

The catalog of stations with synoptic or upper air observations is called Publication 9A and is updated every week, but WMO won't know of a station location unless a country supplies that information.

This catalog does not list climatological or cooperative stations, which you are presumably more interested in, but some of the problems I have found are as follows:
     1.  Until about 2 years ago, the locations were specified only to the nearest minute of latitude and longitude.  Now, locations can be listed to the nearest second (about 30 m accuracy), but over half of the stations still have not been updated.
     2.  Many locations were inaccurate, even to the stated accuracy, and some of them have not been updated for decades, meaning that the location was determined before modern geodesy existed.
     3.  It presumably would be less of a problem in New Mexico, but many weather stations are in remote locations where the latitude and longitude (and elevation) were not accurately determined.  For weather forecasting purposes, accuracy to 0.1 degree (roughly 10 km) was more than sufficient until recently.
     4.  Some catalogs are to the nearest 0.01 degree (or a greater number of decimal places), and some erroneous conversions have been found.
     5.  Some errors are simply typographical errors.  A few catalog entries have been found with the wrong hemisphere, or are a few degrees off.
     6.  Some "accurate" locations are the "official" airfield location (usually the middle of runways) or apparently some other location than the site of the surface instruments.

I am not an expert in geodesy, but from what I have read, station locations in the USA were specified using NAD29 before NAD83 (or WGS84), but the differences between these systems in New Mexico should be well under 100 meters.

The way I locate weather stations is to use online satellite photos at www.wikimapia.org, which uses the same photos as Google Earth.  "Traditional" surface weather stations are often distinctive with a Stevenson screen, rain gauge, and possibly other instruments such as an evaporation pan, connected with walkways.  At airports (generally), the Automated Surface Observing System (ASOS) is distinctive with an array of instruments in a row, and the online satellite photos are often clear enough to see this equipment.  However, in urban areas, looking for the station is like searching for a needle in a haystack, and in some places, the photos are not clear enough.  The photo at any location is usually replaced every few years, and the photos have become more clear in most cases, but globally I have found that around 10% of the locations have only low-resolution photos that are not clear enough to locate a station.  Modern automated weather stations are usually too small to be detectable on the photos, but if you have a site photo, you may be able to look for the same surrounding building and road configuration to locate the station.  Of course, if the station has moved recently, the photo may not be new enough to show the current location.

By the way, for a radiosonde station, I can check the station elevation because a radiosonde sounding reports heights (in "geopotential meters," which differ from geometric meters by less than 1 m except at fairly high elevations) at various levels in the sounding.  From the density of the air, it is possible to compute the "thickness" of the layer from the first above-surface height down to the surface pressure, so the surface elevation is the reported above-surface height minus the thickness of the air layer down to the surface pressure.  I detect station moves that way, since if a station moves horizontally, the new location is unlikely to be exactly at the same elevation as the old location, and the date when the computed elevation changes is obviously the date when the station moved, even if no one reports the move in a catalog.  It is possible to approximately compute the surface elevation for "synoptic" stations that report the surface pressure and sea level pressure (I compute the thickness of the imaginary column of air below the surface down to sea level), but for other weather observation types, it is not possible to make that calculation.

In general, the errors I have found regardless of catalog have been random, not systematic.  Older geodetic systems compared to WGS84 differ by about 700 m around Japan and 500 m around South America, but even those areas have large enough errors in the minutes of latitude and longitude that the errors are random, not systematic.

So, my questions are as follows:
     1.  What kind of stations are you trying to locate?  Cooperative, or some other type?
     2.  What is the source of the station locations?
     3.  What is the precision of the location data in the catalog?  To the nearest minute, or nearest second, or some other precision?  If the station location is stated to the nearest minute, that explains your problem even if the minutes are accurate, since a minute of latitude and longitude is a rectangle nearly 2 km on a side.

Thanks,

Steve Schroeder
Dept. of Atmospheric Sciences
Texas A&M University 

----- Original Message -----
> From: "Brian Woods, DOH" <Brian.Woods at state.nm.us>
> To: rda-users at mailman.ucar.edu
> Sent: Thursday, November 8, 2012 11:49:27 AM
> Subject: [Rda-users] Station Location coordinate systems and projection
> 
> 
> 
> 
> 
> Greetings. I recently attempted spatial joins of meteorological
> stations in New Mexico to zip and census polygons and found the raw
> longitude and latitude problematic to work with. Even after
> converting to decimal values it appeared station values are
> displaced from NM features in NAD 83 projections. Does anyone know
> the coordinate system used for station data or where that is
> documented and can be found?
> 
> 
> 
> Thank you.
> 
> 
> 
> Brian.
> 
> 
> 
> Brian Woods
> 
> Environmental Public Health Epidemiologist
> 
> Environmental Public Health Tracking
> 
> Environmental Health Epidemiology Bureau
> 
> Epidemiology and Response Division
> 
> New Mexico Department of Health
> 
> 1190 St. Francis Drive, N1309
> 
> 505-827-2868 (Office)
> 
> 505-476-1743 (EHEB Office)
> 
> 505-827-2110 (Fax)
> 
> http://nmhealth.org/eheb/
> 
> https://nmtracking.unm.edu/
> 
> 
> 
> 
> 
> 
> _______________________________________________
> Rda-users mailing list
> Rda-users at mailman.ucar.edu
> http://mailman.ucar.edu/mailman/listinfo/rda-users
>