From brideout at haystack.mit.edu Wed Mar 3 10:20:13 2004 From: brideout at haystack.mit.edu (William Rideout) Date: Wed Jul 21 10:29:11 2004 Subject: [gps-developers] Update to GPS processing/graphing code Message-ID: <4045F7AD.7080202@haystack.mit.edu> John, The GPS processing toolkit on my webpage has been updated to include some new graphing capabilities. These new abilities are: 1. Matlab scripts make_gps_movie_cart and make_gps_movie_polar now have the option to plot in corrected geomagnetic coordinates or geodetic. 2. The new script make_gps_image_gsm will plot tec mapped into the GSM XY plane using field lines from the Tsyganenko model (2001 version). 3. All scripts will prompt you for inputs if you don't supply them in the function arguments. So, for example, you can just type "make_gps_image_gsm" in Matlab. The three scripts are called: make_gps_movie_cart make_gps_movie_polar make_gps_image_gsm These matlab scripts should run on Linux, Windows, or Solaris. Bill -- Bill Rideout MIT Haystack Observatory Email: brideout@haystack.mit.edu Phone: 781 981-5624 From brideout at haystack.mit.edu Tue Mar 16 13:23:46 2004 From: brideout at haystack.mit.edu (William Rideout) Date: Wed Jul 21 10:29:11 2004 Subject: [gps-developers] Modified tec file format Message-ID: <40574632.2080604@haystack.mit.edu> John, Just for the record... The modified format for tec files has the same first four columns as the original tec files, but the 5 new columns that described line-of-sight data: 1. time 2. pierce-point latitude 3. pierce-point longitude 4. vertical tec 5. station latitude 6. station longitude 7. az 8. el 9. los tec The naming format for these new files is the same as before, except its tec-all-DDD-HH-MM.dat instead of tec-DDD-HH-MM.dat Bill -- Bill Rideout MIT Haystack Observatory Email: brideout@haystack.mit.edu Phone: 781 981-5624 From brideout at haystack.mit.edu Fri Mar 19 14:35:22 2004 From: brideout at haystack.mit.edu (William Rideout) Date: Wed Jul 21 10:29:12 2004 Subject: [gps-developers] Status of Kevin Duh's Matlab receiver bias code Message-ID: <405B4B7A.9080004@haystack.mit.edu> Anthea, I've been testing Kevin Duh's code in Matlab before I port it to python to integrate into our production GPS data processing. The correlation between his receiver biases and those we get using the 0 tec method was very good, with typical scatter being just a few tec units. His code was having a problem in that about 25% of the time receiver biases were coming back as NAN even when they were physically close to other receivers. I traced this problem to his lower tec filter which was set to -5. Since the tec values he reads are not yet corrected for receiver bias, negitive biases can be quite common. When I modified that limit, the failure rate dropped to 2%. So here's my detailed plan for merging this code into our present process: 1. After the first pass through the data, python will create groups of receivers to be processed together. There will be a maximum size per group (say 50 receivers). There will also be a distance constant that determines if a receiver can join a group (say, 100 km). To create a group, the first receiver in the list that's not already in the list is added to a new group. Then the rest of the receivers are searched to see if they're close enough to add to the group. For each new receiver added, the rest of the group will be searched to see if any non-members can now be added. This continues until max size is reached or no new members can be added. 2. For each group, the python version of Kevin's code will be called. Since Kevin's code really only generates differential biases, I will use two separate methods to set the average bias: a. If the group contains any receivers that JPL determined the bias for, the dc offset will be set to match JPL biases as closely as possible. b. If not, then the dc offset will be set to match our 0 tec biases as closely as possible. 3. If Kevin's code fails (ie, there are no nearby receivers), we'll just use 0 tec values. This is really not an issue, since slight errors in isolated tec values will be very hard to notice. Bill -- Bill Rideout MIT Haystack Observatory Email: brideout@haystack.mit.edu Phone: 781 981-5624