|
The graph above shows the number of satellites above 10 deg. and the
HDOP and PDOP for the satellite geometry at Miami, FL on January 18, 2003.
This pattern will vary over time and depend on which satellites are working,
when they are replaced, and whether there are working spares, etc.
Note that any obstructions at a particular location can reduce the number
of satellites available and hence make DOP worse. |
|
Note that the horizontal Estimated Position Error (EPE) =
HDOP * UERE * 2. This is the 2DRMS value which is usually taken as
the 95% limit. The User Equivalent Range Error (UERE) is a value
that depends on other factors. See Sam
Wormley's GPS/GIS Resources (GPS Error Sources). |
|
|
The graph above shows the number of satellites above 10 deg. and the
HDOP and PDOP for the satellite geometry at London, England on January
18, 2003. Note that any obstructions at a particular location can
reduce the number of satellites available and hence make DOP worse. |
|
|
The graph above shows the satellite orbits overhead with a 10 deg.
cutoff at London on January 18, 2003. Note the "hole" from overhead
(90deg.) and to the north and also that any obstructions can put additional
"holes" in the view. |
|
|
The graph above shows the satellite orbits overhead during the day
(to reduce the clutter from the previous image). The center, 90 deg.,
is directly overhead (above you). The outer edge is the horizon.
The lack of satellites to the north will mean the accuracy in the north-south
direction will not be as good as the accuracy in the east-west direction. |
|
|
The graph above shows the satellite orbits overhead with a 10 deg.
cutoff at Miami, FL on January 18, 2003. Note the "hole" to the north
is smaller than the hole above London because Miami is closer to the equator.
Any obstructions can put additional "holes" in the view. |
|
|
The graph above shows the satellite view at Miami from 4:30 to 5:30
am. Note the lack of satellites overhead. There are only two
above 30 deg. during the period with PRN 2 just entering near the end and
PRN 30 on its way out at the beginning.
Three satellites are needed for a 2D fix and four satellites with a
good geometry are needed for a 3D fix. If enough of the satellites
"near" the horizon are blocked, you won't get a location fix at all! |
|
|
The graph above shows the satellite view at Colorado Springs on
April 11, 2006 at about 1:30 pm. There are seven to eight
satellites in a SW to NE row (since PRN 16 is getting close to the
horizon, it may get blocked).
If you are in a canyon such that satellites PRN 3, 6, 9, and 14 are
blocked, you might think that you still have a good position since
there are about eight satellites in view (seven if PRN 16 gets blocked
too). However, although you will probably get good accuracy along
that SW to NE axis, you don't have good geometry in the NW to SE
direction. So don't expect the accuracy in the NW to SE direction
to be very good.
|
|
|
Here's another example of a satellite pattern that can result in
higher than typical position errors. The graph above shows the
satellite view at Colorado Springs on
May 6, 2006 at about 6:30 pm. There are about seven satellites in
a row or less if those close to the horizon get blocked. If you
are in a canyon such that satellites PRN 4,11, 16, and 30 are
blocked (and maybe 13, 24, and/or 14), then you have another group of
satellites that are mostly in a line resulting in poor location values
until PRN's 20 and 25 separate enough.
The satellites are moving from the dot to the number.
Note that the pattern changes a little from year to year and with the
launch of the next satellite. Sometimes orbits of existing
satellites are changed too.
|
|