GPS Accuracy Using Post Processed Data with CORS Stations
GPS Accuracy Using Post Processed Data with CORS Stations Gerald L. Mader, Knute A. Berstis, NOAA, NOS, National Geodetic Survey Silver Spring, MD Kevin Donnalley US Census Bureau Washington, DC
Presentation Topics • • • • • •
2010 Census Overview NGS Post Processing Software SIRF Chip Set GPS Evaluation Data Melbourne, FL National Geodetic survey (NGS) Testing Corbin, VA Next Steps to Achieve Post Processing Goals Summary
2010 Census • The Census Bureau will be using a Hand Held Computer (HHC) to capture a GPS structure point for every housing unit during Address Canvassing operations Address Canvassing timeframe is November 2008 thru July 2009 The HHC is a Field Data Collection Automation program (FDCA) contract requirement The contractor supplied GPS HHCs have Wide Area Augmentation System (WAAS) capability providing a 3 meter or less radial accuracy (95%) in an unobstructed environment
Census Bureau / NGS Partnership • The Census Bureau is partnering with NOAA’s National Geodetic survey (NGS) to extend the 3 meter accuracy coverage by post processing HHC data with Continuously Operating Reference Stations (CORS). • Census Bureau / NGS Memorandum Of Understanding (MOU) in progress. NGS has developed post processing software for Census Bureau. Beta version is available for testing. NGS has conducted independent tests using a hand held GPS receiver near structures to test the processing software.
[1] Wilson, David L. GPS WAAS Accuracy. http://users.erols.com/dlwilson/gpswaas.htm
NGS / CENSUS Post Processing Block Diagram GPS Input from HHC file
rovr.rnx rdsirf.inp
rdsirf
rover_file start_stop_times
nav_files start_stop_times
Get NAV files
brdc1230.07n …
cat start_stop_times nav_files rover_file > ptpos_1.inp ptpos_1.inp
rovr.kin ptpos_1
rovr.rnx
rovr.sum brdc1230.07n ref_files rovr.sum start_stop_times
xyz
Locate nearest 2-3 CORS and ftp rinex files
abcd1230.07o efgh1230.07o igkl1230.07o
cat start_stop_times xyzfile antfile nav_files rover_file ref_files > sdpos_c.inp sdpos_c.inp xyzfile antfile brdc1230.07n rovr.rnx abcd1230.07o
rovrabcd.kin sdpos_c rovrabcd.sum
rdsirf.inp:
start_stop_times:
f304_no_waas.gps
Sample Data
2007 01 18 12 01 00 2007 01 18 19 22 17
nav_files:
brdc1230.07n brdc1240.07n
rovr.sum:
XYZ: RMS: LLH: RMS: # OBS: # EPOCHS:
ref_files:
abcd1230.07o efgh1230.07o ijkl1230.07o
922985.086 10.431 28.0917557 2.500 28555 2072
-5554802.812 22.707 279.4340651 7.762
2985492.043 14.228 26.408 27.574
Bohne, Paul F. & Nobile, Marc P. MITRE Technical Report MTR070200, “Average GPS signal Availability Estimation”, pp. 1-1, September 2007.
CORS Network in CONUS 2007
HHC Evaluation Unit Testing in Melbourne, FL 10/17/2006
Post Processed Position Errors at NGS mark F304 (Melbourne, FL) re CORS COKO Ref. Station @ 50 Km. 4.0
North Error (Meters)
3.0 2.0 1.0 Post Processed Position Errors (1952 data points)
0.0 -1.0
Post Processed 95% Radial Error (1.67 m.)
-2.0
Census 95% Radial Error (3 m.) NGS Mark F304 ITRF 00 Coords
-3.0 -4.0 -4.0
-2.0
0.0
East Error (Meters)
2.0
4.0
Post Processed Position Errors at NGS MARK F304 (Melbourne, FL) re CORS BRTW Ref. Station @ 120 Km. 4.00
North Error (Meters)
3.00 2.00 1.00
Post Processed Position Errors (1953 data points)
0.00
Post Processed 95% Radial Data Error (1.55 m.)
-1.00
Census 95% Radial Error (3 m.)
-2.00
NGS Ref. Mark (F304 ITRF 00 Coords)
-3.00 -4.00 -4.00
-2.00
0.00
2.00
East Error (Meters)
4.00
HHC WAAS Position Errors re NGS F304 (Melbourne, FL) Ref. Mark
North Error (Meters)
6.00
4.00
2.00 HHC WAAS Position Error (2998 data points)
0.00
HHC WAAS 95% Radial Error (1.69 m.) Census 95% Radial Error (3 m.)
-2.00
NGS Mark F304 ITRF 00 Coords -4.00 -4.00
-2.00
0.00
2.00
East Error (Meters)
4.00
HHC WAAS Post Processed Position Errors at NGS mark F304 (Melbourne, FL) re CORS BRTW Ref. Station @ 120 Km. 4.00
North Error (Meters)
3.00 2.00 1.00 0.00
HHC WAAS Position Errors (1965 data points)
-1.00
HHC Post Processed WAAS 95% Radial Error (1.7 m.) Census 95% Radial Error (3 m.)
-2.00 NGS Mark F304 ITRF 00 coords -3.00 -4.00 -4.00 -3.00 -2.00 -1.00 0.00
1.00
2.00
East Error (Meters)
3.00
4.00
HHC Post Processed WAAS Position Errors at NGS mark F304 (Melbourne, FL) re CORS COKO Ref. Station @ 50 Km. 4.0
North Error (Meters)
3.0 2.0 HHC WAAS Position Errors (1967) data points) HHC Post Processed WAAS 95% Radial error (1.1 m.) Census 95% Radial Error (3 m.)
1.0 0.0 -1.0
NGS Ref. Mark (F304)
-2.0 -3.0 -4.0 -4.0
-3.0
-2.0
-1.0
0.0
1.0
2.0
East Error (Meters)
3.0
4.0
GPS Post Processed & WAAS Accuracy at Melbourne, FL NGS Reference Marks Post Processed with 50 Km Ref. Station
1.67 m. (95%) @ F304
Post Processed with 120 Km Ref. Station
1.55 m. (95%) @ F304
WAAS @ F304
1.69 m. (95%)
WAAS @ 5062 (THRON)
1.50 m. (95%)
WAAS Post Processed with 120 Km Ref. Station
1.70 m. (95%) @ F304
WAAS Post Processed with 50 Km Ref. Station
1.10 m. (95%) @ F304
NGS Corbin, VA Testing November – December 2007 • NGS conducted tests with a hand held Garmin 12XL GPS receiver • Objective of tests was to verify the NGS post processing software. • GPS post processed coordinates acquired at 12 building entrances of different sizes and at 2 to 2.5 meters offset. • One minute data sets taken with a one second sampling rate. • Third party software used to acquire data via a laptop and to convert Garmin binary data to RINEX format needed by NGS post processing software.
35
Garmin Post Processed GPS Errors re NGS Mark at House #1 Corbin, VA 12/27/2007
30
North Error (Meters)
25 20 15 10 5
Garmin GPS Errors re NGS Mark at House #1 - 11/60 (18%) Missing Data Census 95% Radial error (3 m.) - 5% of Data Within Census Accuracy NGS Mark at House #1
0 -5 -10 -15 -20 -25
-20
-15
-10
-5
0
5
10
East Error (Meters)
15
20
25
30
Tap & Go - First Point
Garmin Post Processed GPS Errors re NGS Mark at House #1 2 Meter Offset Corbin, VA 12/27/2007 4 2
North Error (Meters)
0 -2 -4 -6
Garmin GPS Errors re NGS Mark at House #1-Offset by 2m. - 4/60 (7%) Missing Data Census 95% Radial Error (3 m.) - 22% of Data Within Census Accuracy NGS 2 m. Offset Mark
-8 -10 -12 -14 -12
-10
-8
-6
-4
-2
0
2
East Error (Meters)
4
6
8
Tap & Go - First Point
Garmin Post Processed GPS Errors re NGS Mark @ GPS Bldg. #11 Corbin, VA 12/27/2007 60 50
North Error (Meters)
40 30 20 10 0 -10
Garmin GPS Errors re NGS mark at Bldg. #11 - 9/60 (15%) Missing Data Census 95% Radial error (3 m.) - No Data Within Census Accuracy
-20 -30
NGS Mark @ Building # 11 -40 -10
-5
0
5
10
15
20
East Error (Meters)
25
30
35
Tap & Go - First Data Point
Garmin Post Processed GPS Errors re NGS Mark @ GPS Bldg. 11 - 2 Meter Offset Corbin, VA 12/27/2007 10
North Error (Meters)
5
0
-5
Garmin GPS Errors re NGS Mark @ Bldg. 11 12/60 (20%) Missing Data Census 95% Radial Error (3 m.) 17% of Data Within Census Accuracy Limit NGS Mark @ Buiding 11
-10
-15
-20 -10
-5
0
5
10
East Error (Meters)
15
20
Tap & go - First Data Point
Summary - Garmin Post Processed Data 12/27/2007 Building ID
% Data Within 3 Meters
% Missing Data (60 Data Points)
5%
18 %
House #1 2 .5 m. Offset
22 %
7%
House #2-Door
1.7 %
25 %
5%
25 %
DEQ Storage-Door
18 %
15 %
DEQ Storage 2.5 m. Offset
17%
0%
GPS Bldg.
0%
15 %
17 %
20 %
House #1- Door
House #2 2.5 m. Offset
GPS Bldg. 2 m. Offset
Future Tests • FDCA HHC handheld units with post processing capability available for testing in Oct. 2008. • Conduct additional testing with Census Bureau in GEO’s (Geography Division) established GPS Test Course. Tests may not be in Anne Arundel County but a similar scenario in Prince Georges County, MD. • Continue to fine tune NGS post processing software
Corbin, VA Test Area
Unobstructed GPS Coordinate at Street Intersection
Aerial Imagery Accuracy Building ID
Google Earth Pro
ESRI ArcGIS Accuracy Desktop 9.1 Within 3 Meters
Bldg. 8
5.76 m.
4.72 m.
Bldg. 10A
5.28 m.
2.99 m
X
Pump House
3.69 m.
0.6 m
X
House #1
4.49 m.
6.26 m.
Main Bldg. Annex
4.76 m.
5.03 m.
GPS Bldg.
5.41 m.
4.33 m.
Bldg. 7
6.29 m.
2.57 m.
X
Bldg. 5
5.68 m.
1.51 m.
X
Bldg. 6
5.26 m.
4.54 m.
DEQ Storage
5.26 m.
6.06 m.
Summary • NGS Post Processing accuracy using CORS Reference stations was less than 3 meters (95%) in unobstructed environments. • 3 meter GPS accuracy may not be achievable at residential entrances due to multipath & blocked signals; Census bureau has procedures in place to obtain the best possible GPS coordinate under these conditions. • Field data available to NGS for post processing (up to 140 million data points) 30 days after Address Canvassing is complete. • NGS is ready to assist Census Bureau in data post processing; additional NGS resources available to establish accurate coordinates in challenging environments. NGS has 37 State Geodetic Advisors and Coordinators that could assist the Census Bureau as consultants in the Address Canvassing effort.
Presentation Topics • • • • • •
2010 Census Overview NGS Post Processing Software SIRF Chip Set GPS Evaluation Data Melbourne, FL National Geodetic survey (NGS) Testing Corbin, VA Next Steps to Achieve Post Processing Goals Summary
2010 Census • The Census Bureau will be using a Hand Held Computer (HHC) to capture a GPS structure point for every housing unit during Address Canvassing operations Address Canvassing timeframe is November 2008 thru July 2009 The HHC is a Field Data Collection Automation program (FDCA) contract requirement The contractor supplied GPS HHCs have Wide Area Augmentation System (WAAS) capability providing a 3 meter or less radial accuracy (95%) in an unobstructed environment
Census Bureau / NGS Partnership • The Census Bureau is partnering with NOAA’s National Geodetic survey (NGS) to extend the 3 meter accuracy coverage by post processing HHC data with Continuously Operating Reference Stations (CORS). • Census Bureau / NGS Memorandum Of Understanding (MOU) in progress. NGS has developed post processing software for Census Bureau. Beta version is available for testing. NGS has conducted independent tests using a hand held GPS receiver near structures to test the processing software.
[1] Wilson, David L. GPS WAAS Accuracy. http://users.erols.com/dlwilson/gpswaas.htm
NGS / CENSUS Post Processing Block Diagram GPS Input from HHC file
rovr.rnx rdsirf.inp
rdsirf
rover_file start_stop_times
nav_files start_stop_times
Get NAV files
brdc1230.07n …
cat start_stop_times nav_files rover_file > ptpos_1.inp ptpos_1.inp
rovr.kin ptpos_1
rovr.rnx
rovr.sum brdc1230.07n ref_files rovr.sum start_stop_times
xyz
Locate nearest 2-3 CORS and ftp rinex files
abcd1230.07o efgh1230.07o igkl1230.07o
cat start_stop_times xyzfile antfile nav_files rover_file ref_files > sdpos_c.inp sdpos_c.inp xyzfile antfile brdc1230.07n rovr.rnx abcd1230.07o
rovrabcd.kin sdpos_c rovrabcd.sum
rdsirf.inp:
start_stop_times:
f304_no_waas.gps
Sample Data
2007 01 18 12 01 00 2007 01 18 19 22 17
nav_files:
brdc1230.07n brdc1240.07n
rovr.sum:
XYZ: RMS: LLH: RMS: # OBS: # EPOCHS:
ref_files:
abcd1230.07o efgh1230.07o ijkl1230.07o
922985.086 10.431 28.0917557 2.500 28555 2072
-5554802.812 22.707 279.4340651 7.762
2985492.043 14.228 26.408 27.574
Bohne, Paul F. & Nobile, Marc P. MITRE Technical Report MTR070200, “Average GPS signal Availability Estimation”, pp. 1-1, September 2007.
CORS Network in CONUS 2007
HHC Evaluation Unit Testing in Melbourne, FL 10/17/2006
Post Processed Position Errors at NGS mark F304 (Melbourne, FL) re CORS COKO Ref. Station @ 50 Km. 4.0
North Error (Meters)
3.0 2.0 1.0 Post Processed Position Errors (1952 data points)
0.0 -1.0
Post Processed 95% Radial Error (1.67 m.)
-2.0
Census 95% Radial Error (3 m.) NGS Mark F304 ITRF 00 Coords
-3.0 -4.0 -4.0
-2.0
0.0
East Error (Meters)
2.0
4.0
Post Processed Position Errors at NGS MARK F304 (Melbourne, FL) re CORS BRTW Ref. Station @ 120 Km. 4.00
North Error (Meters)
3.00 2.00 1.00
Post Processed Position Errors (1953 data points)
0.00
Post Processed 95% Radial Data Error (1.55 m.)
-1.00
Census 95% Radial Error (3 m.)
-2.00
NGS Ref. Mark (F304 ITRF 00 Coords)
-3.00 -4.00 -4.00
-2.00
0.00
2.00
East Error (Meters)
4.00
HHC WAAS Position Errors re NGS F304 (Melbourne, FL) Ref. Mark
North Error (Meters)
6.00
4.00
2.00 HHC WAAS Position Error (2998 data points)
0.00
HHC WAAS 95% Radial Error (1.69 m.) Census 95% Radial Error (3 m.)
-2.00
NGS Mark F304 ITRF 00 Coords -4.00 -4.00
-2.00
0.00
2.00
East Error (Meters)
4.00
HHC WAAS Post Processed Position Errors at NGS mark F304 (Melbourne, FL) re CORS BRTW Ref. Station @ 120 Km. 4.00
North Error (Meters)
3.00 2.00 1.00 0.00
HHC WAAS Position Errors (1965 data points)
-1.00
HHC Post Processed WAAS 95% Radial Error (1.7 m.) Census 95% Radial Error (3 m.)
-2.00 NGS Mark F304 ITRF 00 coords -3.00 -4.00 -4.00 -3.00 -2.00 -1.00 0.00
1.00
2.00
East Error (Meters)
3.00
4.00
HHC Post Processed WAAS Position Errors at NGS mark F304 (Melbourne, FL) re CORS COKO Ref. Station @ 50 Km. 4.0
North Error (Meters)
3.0 2.0 HHC WAAS Position Errors (1967) data points) HHC Post Processed WAAS 95% Radial error (1.1 m.) Census 95% Radial Error (3 m.)
1.0 0.0 -1.0
NGS Ref. Mark (F304)
-2.0 -3.0 -4.0 -4.0
-3.0
-2.0
-1.0
0.0
1.0
2.0
East Error (Meters)
3.0
4.0
GPS Post Processed & WAAS Accuracy at Melbourne, FL NGS Reference Marks Post Processed with 50 Km Ref. Station
1.67 m. (95%) @ F304
Post Processed with 120 Km Ref. Station
1.55 m. (95%) @ F304
WAAS @ F304
1.69 m. (95%)
WAAS @ 5062 (THRON)
1.50 m. (95%)
WAAS Post Processed with 120 Km Ref. Station
1.70 m. (95%) @ F304
WAAS Post Processed with 50 Km Ref. Station
1.10 m. (95%) @ F304
NGS Corbin, VA Testing November – December 2007 • NGS conducted tests with a hand held Garmin 12XL GPS receiver • Objective of tests was to verify the NGS post processing software. • GPS post processed coordinates acquired at 12 building entrances of different sizes and at 2 to 2.5 meters offset. • One minute data sets taken with a one second sampling rate. • Third party software used to acquire data via a laptop and to convert Garmin binary data to RINEX format needed by NGS post processing software.
35
Garmin Post Processed GPS Errors re NGS Mark at House #1 Corbin, VA 12/27/2007
30
North Error (Meters)
25 20 15 10 5
Garmin GPS Errors re NGS Mark at House #1 - 11/60 (18%) Missing Data Census 95% Radial error (3 m.) - 5% of Data Within Census Accuracy NGS Mark at House #1
0 -5 -10 -15 -20 -25
-20
-15
-10
-5
0
5
10
East Error (Meters)
15
20
25
30
Tap & Go - First Point
Garmin Post Processed GPS Errors re NGS Mark at House #1 2 Meter Offset Corbin, VA 12/27/2007 4 2
North Error (Meters)
0 -2 -4 -6
Garmin GPS Errors re NGS Mark at House #1-Offset by 2m. - 4/60 (7%) Missing Data Census 95% Radial Error (3 m.) - 22% of Data Within Census Accuracy NGS 2 m. Offset Mark
-8 -10 -12 -14 -12
-10
-8
-6
-4
-2
0
2
East Error (Meters)
4
6
8
Tap & Go - First Point
Garmin Post Processed GPS Errors re NGS Mark @ GPS Bldg. #11 Corbin, VA 12/27/2007 60 50
North Error (Meters)
40 30 20 10 0 -10
Garmin GPS Errors re NGS mark at Bldg. #11 - 9/60 (15%) Missing Data Census 95% Radial error (3 m.) - No Data Within Census Accuracy
-20 -30
NGS Mark @ Building # 11 -40 -10
-5
0
5
10
15
20
East Error (Meters)
25
30
35
Tap & Go - First Data Point
Garmin Post Processed GPS Errors re NGS Mark @ GPS Bldg. 11 - 2 Meter Offset Corbin, VA 12/27/2007 10
North Error (Meters)
5
0
-5
Garmin GPS Errors re NGS Mark @ Bldg. 11 12/60 (20%) Missing Data Census 95% Radial Error (3 m.) 17% of Data Within Census Accuracy Limit NGS Mark @ Buiding 11
-10
-15
-20 -10
-5
0
5
10
East Error (Meters)
15
20
Tap & go - First Data Point
Summary - Garmin Post Processed Data 12/27/2007 Building ID
% Data Within 3 Meters
% Missing Data (60 Data Points)
5%
18 %
House #1 2 .5 m. Offset
22 %
7%
House #2-Door
1.7 %
25 %
5%
25 %
DEQ Storage-Door
18 %
15 %
DEQ Storage 2.5 m. Offset
17%
0%
GPS Bldg.
0%
15 %
17 %
20 %
House #1- Door
House #2 2.5 m. Offset
GPS Bldg. 2 m. Offset
Future Tests • FDCA HHC handheld units with post processing capability available for testing in Oct. 2008. • Conduct additional testing with Census Bureau in GEO’s (Geography Division) established GPS Test Course. Tests may not be in Anne Arundel County but a similar scenario in Prince Georges County, MD. • Continue to fine tune NGS post processing software
Corbin, VA Test Area
Unobstructed GPS Coordinate at Street Intersection
Aerial Imagery Accuracy Building ID
Google Earth Pro
ESRI ArcGIS Accuracy Desktop 9.1 Within 3 Meters
Bldg. 8
5.76 m.
4.72 m.
Bldg. 10A
5.28 m.
2.99 m
X
Pump House
3.69 m.
0.6 m
X
House #1
4.49 m.
6.26 m.
Main Bldg. Annex
4.76 m.
5.03 m.
GPS Bldg.
5.41 m.
4.33 m.
Bldg. 7
6.29 m.
2.57 m.
X
Bldg. 5
5.68 m.
1.51 m.
X
Bldg. 6
5.26 m.
4.54 m.
DEQ Storage
5.26 m.
6.06 m.
Summary • NGS Post Processing accuracy using CORS Reference stations was less than 3 meters (95%) in unobstructed environments. • 3 meter GPS accuracy may not be achievable at residential entrances due to multipath & blocked signals; Census bureau has procedures in place to obtain the best possible GPS coordinate under these conditions. • Field data available to NGS for post processing (up to 140 million data points) 30 days after Address Canvassing is complete. • NGS is ready to assist Census Bureau in data post processing; additional NGS resources available to establish accurate coordinates in challenging environments. NGS has 37 State Geodetic Advisors and Coordinators that could assist the Census Bureau as consultants in the Address Canvassing effort.
