Caster Locations vs Base Station Locations

Users have recently asked about the differences between the Lat-Long location found in the Caster Table and that found in the message streams, the RTCM 1005,1006,1007 messages types.  This is perhaps driven by the the simple map feature that was first added in SNIP Revision 1.0.0 which is modeled on similar graphical displays found in many NTRIP Clients.

This article serves to describe how each of these location values are used in the typical RTK system, and by SNIP.

First and foremost, the location shown in the caster table for each mountPt is only advisory – it is not actually used in RTK or PPP navigation.  Bad data here will not harm a rover’s ability to position in any way.  Still, one should set it correctly.  Its intended use is only to assist the NTRIP Client (the rover device) to determine what stream it wants.  Quite often there are errors found in the data published by others.  That is why some places in North America get the sign wrong (Hint to my fellow Americans: we live to the East of the prime meridian, so all our Longitude values are negative). And many systems with virtual reference stations (VRS), or even simple nearest site connection systems, will leave the value in the caster table entry blank. I.e. 0,0 – and that is why you often see reference stations claiming to be in the ocean south of Ghana. [See this article about zero, zero locations in caster tables for further details]

By definition, the caster table entry has only two digits of precision and is in degrees (the form: DD.dd).  By contrast (and also by definition) the ECEF values used in the actual RTCM messages is accurate to 0.0001 meters (expressed in ECEF values in the form: XYZ in meters).  So if you plot both on the same map you will notice some differences,  such as the two locations shown below.

basestationlocation_caster_vs_ecef_positions-com

The site called “gisar23”  is sending messages in the RTCXM2.3 format (which SNIP does not decode at this time), while the site “gisar30” is sending messages in the RTCM3 formats (with SNIP has decoded and has extracted the more precise location).   These come from the same antenna and therefore the exact same location, one is plotted using the rough caster table data, the other the precise ECEF data of the message stream.  You can see from the base map that “gisar30” is located alongside the edge of a building wall.  This is a site local to our offices (~50km) and is run by the GIS firm ESRI.  [We often use it for longer range baseline L1-only testing.]  In this image the rounding caused by the less precise data in the caster table is clear to see.

A simple hint to detect which source was used in a map (this is often true for similar maps created by other tools as well) is to look closely the height above the ellipsoid (if shown).  A caster table entry does not provide this value (so it is always shown as zero), but any EFEC value converted the LLH will.  The rounding is also a clue.

 

Further details about SNIP.  The SNIP process of determining where a reference station is located is as follows.  First, start the caster stream up and enable data servicing as quickly as possible (with a dummy placeholder caster entry).  Second, use the caster data provided by the sender/source once a caster table has been loaded.  Third, any time SNIP can decode a more accurate position from the data stream itself  (and it will also validate what the caster table entries say), replace the original data in the caster with the verified data and use it in further map displays.   This process is automatic, but takes about 120 seconds to complete based on how fast the caster table is sent and how often a reference station position message is sent by the stream.

 

Was this article helpful?

Related Articles

Help us make this topic better...