GPS for crustal deformation studies. May 7, 2009

Transcription

1 GPS for crustal deformation studies May 7, 2009

2 High precision GPS for Geodesy Use precise orbit products (e.g., IGS or JPL) Use specialized modeling software GAMIT/GLOBK GIPSY OASIS BERNESE These software packages will Estimate integer ambiguities Reduces rms of East component significantly Model physical processes that effect precise positioning, such as those discussed so far plus Solid Earth Tides Polar Motion, Length of Day Ocean loading Relativistic effects Antenna phase center variations

3 High precision GPS for Geodesy Produce daily station positions with 2 3 mm horizontal repeatability, 10 mm vertical. Can improve these stats by removing common mode error.

4 Ambiguity Resolution: why

5 Ambiguity Resolution: why Average east repeatability bias free = 3.3 mm bias fixed = 2.7 mm 78 stations distributed around the globe

6 Ambiguity resolution Resolving integer ambiguity converts carrier phase biases into ultraprecise psuedorange L 1 = φ 1 λ 1 = R + c(δt u δt s ) + Z + I φ1 + M φ1 + B 1 + ε φ1 Blewitt, G., Carrier Phase Ambiguity Resolution for the Global Positioning System Applied to Geodetic Baselines up to 2000 km, J. Geophysical Research, 1989

7 Double Differencing j ΔL AB j = Δρ AB j + cδτ AB + ΔZ AB j ΔI AB j + ΔB AB k ΔL AB k = Δρ AB k + cδτ AB + ΔZ AB k ΔI AB k + ΔB AB jk ΔL AB = Δρ jk AB + ΔZ jk AB ΔI jk jk AB + ΔB AB Satellite and clock errors are gone Random errors are increased (e.g., multipath, measurement noise) Double difference phase ambiguity is a true integer: removes uncalibrated components of phase delay for receiver and satellite

8 Ambiguity Resolution jk ΔL AB = Δρ jk AB + ΔZ jk AB ΔI jk jk AB + λ ΔN AB How do we go about solving for N? What we end up doing is solving for widelane and narrowlane biases. First, form widelane linear combination of phase observables: L δ k i L δ k i i = c(φ 1k φ i 2k ) /( f 1 f 2 ) i = ρ k + I i k f 1 f 2 ( f 2 1 f 2 i 2 ) + λ δ B δ k + Δρ k i f 2 ( f 1 2 f 2 2 ) λ δ = c /( f 1 f 2 ) 86.2 cm

9 Ambiguity resolution Use pseudorange to calibrate widelane, solve for b B δ k i = 1 i i L δ λ k P δ k 2Δρ i k f 1 f 2 /( f 2 1 f 2 2 ) δ [ ] Computed at each data point, time averaged real value is taken Form double differenced widelane ij N δ kl = B δ k i B δ l i B δ k j + B δ l j Independent of knowledge of orbits, station locations Dependent on common visibility of satellites.

10 Ambiguity resolution Then solve for narrowlane ambiguities Narrowlane is ionospheric free combination: L c k i = ( f 1 2 L 1k i f 2 2 L 2k i ) /( f 1 2 f 2 2 ) i B c k = ( f 2 i 1 λ 1 B 1 k f 2 2 λ 2 B i 2 k) /( f 2 1 f 2 2 ) λ c 10.7 cm

11 Analysis Software What are some differences between GIPSY and GAMIT? GIPSY JPL Need a license Single difference Tightly constrains satellite parameters Data weighting independent of elevation angle More flexible: e.g., can be used for low earth orbiters GAMIT MIT Open source Double difference Loosely constraints satellite parameters Elevation angle dependent data weighting More focused on tools for solid earth science applications

12 How to get started? Don t need to process data yourself? Auto gipsy: PBO Analysis centers Acquire software GIPSY: oasis.jpl.nasa.gov/ GAMIT: gpsg.mit.edu/~simon/gtgk/ Learn software basics: UNAVCO short course on GAMIT : No regular courses offered on GIPSY currently read documentation & work with JPL ers

13 Processing Overview 1. Get Daily dual frequency GPS observations from a network of stations (could be 1 station) RINEX files 1. Edit the data for outliers, losses of lock by the receiver (cycle slips) 2. Model the observations Fixing orbits and clocks of the satellites, polar motions, etc. Modeling tidal effects, propagation delays, etc. 3. Estimate station coordinates and other station parameters 4. Solutions Station coordinates and covariances Tropospheres as a functions of time Phase biases as functions of satellite station pairs and time Station Clocks 5. Repeat for next day of data

14 Processing Steps Collect Data (campaign or continuous) Raw data is converted from binary to RINEX format Keep track of metadata Equipment used Antenna heights Get any additional data and metadata from archive ftp and web services Get precise orbits, clocks, and EOP from an archive/provider E.g, ftp://sideshow.jpl.nasa.gov/pub/gipsy/products Decide on Solution rate Data rates are usually 30 sec Solution rates are usually daily Network vs. Point Positioning More CPU, memory Faster, independent Do I need those biases fixed? How big is the signal? What is the solution rate? How long is the time series?

15 Processing Steps Process GPS observables Fixed orbits and clocks? Estimating Station positions Nuisance parameters (i. e. clocks, trops, etc.) Assess quality of estimates Are the formal errors reasonable? Are the values of the parameters reasonable? Are the residuals nominal? Were there a lot of data outliers and/or phase breaks? Post Process Time series of station positions for Scientific signals Geodetic studies

16 GIPSY OASIS: How it works

17 GIPSY: How its used /goa/bin/gd2p.pl -i chwk o -w_elmin 15 -d \ -n chwk -stop_before wash \ -tdp_in /sggs0/tdpfilein2 \ -orb_clk "flinn /tmp/orbits/ \ -add_ocnld -no_del_shad

18 Analysis of Very Large GPS Data Sets Challenge to analyze the large amount of data efficiently: - 10 years of data from station network. - Almost 2 years to process data set using single CPU. Solved problem by developing Network Processor: - Produces bias-fixed solutions using GIPSY on cluster computers. - Using 250 dual processor nodes, can re-analyze entire GEONET data set in under 2 days. May 7, 2009

19 Reference Frame The Terrestrial RF provides the stable coordinate system that allows us to link measurements over space and time. Cartesian Coordinate System Earth Centered Earth Fixed Defined by station coordinates and station velocities Velocities usually linear with time Not really true for any point on the Earth Some examples: International Terrestrial Reference Frame (ITRF) Stable North American Reference Frame (SNARF) workinggroups_projects/snarf/snarf.html

20 ITRF International Terrestrial Reference Frame Space geodesy based Satellite Laser Ranging for geocenter Very Long Baseline Interferometery (VLBI) for scale VLBI for celestial orientation GPS and DORIS (and VLBI and SLR) for coordinate positions and velocities DORIS is a French system for positioning of TOPEX/Poseidon & Jason satellites Identical to WGS84 at the meter level

21 Why does ITRF use multiple techniques? High precision geodesy is very challenging Accuracy of 1 part per billion Fundamentally different observations with unique capabilities Together provide cross validation and increased accuracy Technique Signal Source Obs. Type Celestial Frame UT1 VLBI Microwave Quasars Time difference SLR Optical Satellite Two-way range Yes No No Scale Yes Yes Yes Geocenter No Yes Yes Geographic Density GPS Microwave Satellites Range change No No Yes Real-time Yes No Yes Decadal Stability Yes Yes Yes

22 Reference Frame: ITRF05 Defined by station coordinates and velocities Coordinates at reference epoch: Jan 1, 2005.

23 Reference Frame: ITRF05 Defined by station coordinates and velocities Coordinates at reference epoch: Jan 1, 2005.

24 Reference Frame: SNARF Example of local reference frame definition Plate boundary velocities in ITRF05 aren t as easy to interpret:

25 Reference Frame: SNARF Need sites on stable NA. Some challenges: Glacial Isostatic Adjustment is significant signal in stable interior Historically, not as many sites in stable regions Shorter time series for defining velocities Sites are installed for navigation/surveying instead of tectonic purposes Similar to ITRF, combine time series from different analysis centers Different from ITRF, only uses GPS time series

26 SNARF and GIA Dominant vertical signal Need to model in order to determine translation between SNARF and ITRF

27 SNARF: Accounting for GIA Horizontal effects not very well constrained M. Tamisea

28 Plate Boundary Observatory Velocities From GPS Explorer website:

29 Combining different GPS velocities Velocity fields are in different realizations of frame Need to make sure solutions are loosely constrained Difficult to account for from different analysis software Need to put all solutions into a common reference frame Helmert s Transformation 7 Parameters Scale Translation Rotations

30 Evaluating GPS time series

31 If you see a signal in a GPS time series, what should you ask? Were precise techniques applied? Is the equipment and software recognized for high precision work? Were the orbits high precision orbits? Is the reference frame a high precision frame? Are there unmodeled and systematic errors? Are there correlations between the height and vertical estimates? Are the residuals, chi square, and other metrics of fit reasonable? Could this be some non tectonic phenomena? A strange tropospheric effect? A vertical and/or horizontal signal? Changes at the GPS site Is the signal regional or only seen on one station?

32 Other things to care about Was a quality source of orbits and clocks used? Was a fancy estimation strategy used? Was the solution regionally filtered? Are there regionally correlated signals? Tectonic source? Orbit error? Are the results consistent with expectations from best practices? Positions 3, 5, 6 mm (Bias free) 1, 1, 4 mm (Bias fixed) Velocities mm/yr over several years

33 Tree Growth/Trimming