Using non-tidal atmospheric loading model in space geodetic data processing: Preliminary results of the IERS analysis campaign

Transcription

1 Using non-tidal atmospheric loading model in space geodetic data processing: Preliminary results of the IERS analysis campaign Xavier Collilieux (1), Zuheir Altamimi (1), Laurent Métivier (1), Tonie van Dam (2), Graham Appleby (3), Johannes Boehm (4), Rolf Dach (5), Mathias Fritsche (6), Ramesh Govind (7), Rolf Koenig (8), Hana Krásná (4), Magda Kuzmicz-Cieslak (9,10), Sébastien Lambert (11), Frank G. Lemoine (10), Cinzia Luceri (12), Dan MacMillan (10), Maria Mareyen (13), Erricos Pavlis (9,10), and Daniela Thaller (5) EGU Vienna Tue, 09 Apr, 08:30 12:00 / Room R13 1

2 Previous studies IERS campaign Application of NT-atmospheric loading models at the observation level has been suggested or tested for GPS (Tregoning & van Dam, 2005), VLBI (Petrov & Boy, 2004; Boehm et al., 2009), SLR (unpublished) and DORIS (unpublished). Recommended by attendees of the GGOS Unified Analysis Workshop: Use NT-ATML models > Test NT-ATML model in the data processing Objectives of the IERS campaign: Evaluate the impact of applying non-tidal atmospheric loading model on the ITRF Evaluate the different ways to apply such corrections for the 4 geodetic techniques Time period: Standard solution Standard solution Standard solution Standard solution Standard solution Corrected solution Corrected solution Corrected solution Corrected solution Corrected t solution Set of Station position series (+ Earth Orientation Parameters and other parameters) Daily: GPS, VLBI Weekly: SLR, DORIS EGU Vienna 2

3 Loading model Loading model to be tested: Based on NCEP (6h) surface pressure data provided at 2.5 x 2.5 degree spacing over the Earth. Station displacements in the CM frame Sub-daily atmospheric tides have been filtered (Tregoning and Watson, 2009) Spherical coefficients of the gravitational potential changes up to degree and order 50 (satellite techniques) (Tregoning et Watson, 2009) Raw model Model: filter applied Model: filter applied + atmospheric tides EGU Vienna 3

4 Submitted datasets Technique Software AC Loading model SLR EPOSOC GFZ GGFC & GGFC for gravity (Both solutions) SLR GeodynII/Solve ASI GGFC & GGFC for gravity SLR GeodynII/Solve JCET GGFC & GGFC for gravity SLR GeodynII/Solve GA GGFC & NCEP (GSFC) for gravity SLR RGO SATAN NSGF GGFC & GGFC for gravity SLR Bernese BKG GGFC & GRACE AOD for gravity VLBI CALC/SOLVE GSFC GGFC VLBI CALC/SOLVE OPA GGFC VLBI VieVS TU Wien GGFC DORIS Geodyn/Solve GSFC GGFC & GGFC for gravity (Both solutions) GPS Bernese COD GGFC GPS Bernese TU Dresden GRACE dealiasing (+ NT ocean loading) EGU Vienna 4

5 Weekly series: DORIS, SLR Changes in Terrestrial frame origins Daily series: GPS degree-1 > 3 of the SLR solutions (not shown) show no origin change: need to be investigated degree-1 Significant origin change only for DORIS Z translation! (-1.4+/-0.3mm in ; 0.5+/- 0.1 mm/yr ) EGU Vienna 5

6 Tech. SLR VLBI DORIS GPS AC GFZ GSFC GSFC CODE Combination Selection of one solution per technique for combination tests Include local ties at co-location site ITRF2008 type combination (NB: ) Position difference can be as large as 5.0 mm but generally smaller than 1.0 mm! Red: positive difference Blue: negative difference Vertical velocity differences from two multi-technique combinations. With or without NT-ATML applied EGU Vienna 6

7 Tech. SLR VLBI DORIS GPS AC GFZ GSFC GSFC CODE Combination Selection of one solution per technique for combination tests Include local ties at co-location site ITRF2008 type combination (NB: ) > 3 years! Vertical velocity differences from two multi-technique combinations. With or without NT-ATML applied EGU Vienna 7

8 mm A priori vs post corrections (1/5) Alternative way to apply a NT-loading model: apply mean daily/weekly correction to standard solutions Post corrections at the Terrestrial Frame level Use epoch block information (min. and max. epoch) Ex: Height post corrections at one VLBI station Post correction Raw model Corrections made at the normal equation level EGU Vienna 8

9 A priori vs post corrections (2/5) Boxplot of the wrms of station position series differences per solution doris gps slr vlbi Boxplot: Only well performing stations are shown for SLR (notation -ex ). Both are shown for GFZ solutions Color boxplot: all points used to compute the WRMS of the series Black boxplot: 10% of «outlier»rejected per series before computing the WRMS. EGU Vienna 9

10 A priori vs post corrections (3/5) Boxplot of the wrms of station position series differences per solution After 6-parameter Helmert transformation doris Only well gps performing stations slr used on shown on these 3 plots Boxplot: Only well performing stations are shown for SLR (notation -ex ). Both are shown for GFZ solutions Color boxplot: all points used to compute the WRMS of the series Black boxplot: 10% of «outlier»rejected per series before computing the WRMS. EGU Vienna 10

11 Conclusion A priori vs post corrections (4/5) small differences in origin for satellite techniques (along Z only for DORIS but the three components for GPS/SLR) Position differences less than 0.2 mm WRMS for the 3 techniques except SLR. Larger differences due to the irregular weather dependent sampling of data acquisition. GFZ show smaller differences but is the only solution that model gravity field variations for both solutions. Limitation of the post corrections due to the averaging Other reasons for differences. Aliasing of signal in other parameters Empirical accelerations (Dach et al., 2011) Analysis of other parameters shows that aliasing is real but limited : Difference in tropospheric delays VLBI - TU-WIEN + Range biases in SLR. No obvious correlation btw range bias differences and height model time series + Troposphere wet delay in VLBI EGU Vienna 11

12 A priori vs post corrections (5/5) Vertical velocity differences between combined a priori and a posteriori solutions (all sites) Vertical velocity differences from two multi-technique combinations. With NT-ATML applied a priori or with NT-ATML applied as post corrections EGU Vienna 12

13 A priori vs post corrections (5/5) Vertical velocity differences between combined a priori and a posteriori solutions (> 3 years) Vertical velocity differences from two multi-technique combinations. With NT-ATML applied a priori or with NT-ATML applied as post corrections EGU Vienna 13

14 Conclusions First multi-technique combination of solutions with non-tidal atmospheric loading applied a priori First comparison of a priori and a posteriori corrections for DORIS and for SLR using different a priori modeling of the gravity field As expected : generally negligible changes in station positions and velocities for stations having more than 3 years of observations. Only DORIS solutions show frame origin change at the level of 1.4 mm in Z and 0.5 mm/yr over EOPs affected, but only slightly compared to their known accuracy (polar motion median WRMS for xp and yp: ~ 20 μas for VLBI; 20 μas for DORIS; 40 μas for SLR; 30 μas for GPS including NT-ocean loading) Applying corrections a priori slightly change the origin of the quasi-instantaneous frame (compared to expectation) as illustrated by differences between a priori and a posteriori corrections but no long-term effect EGU Vienna 14