1. 1 On the use of Numerical Weather Models to Predict Neutral-Atmosphere Delays Felipe G. Nievinski

2. 2 The curse of tropospheric delay in GPS positioning One of main error sources in medium- to long- range kinematic applications. Estimation more challenging than in static app. –Due to, e.g., time-varying height. A better a priori prediction would be valuable. –Whether estimating or only correcting for tropospheric delay.

3. 3 Numerical Weather Models: a better picture of today’s weather? Typical tropospheric delay prediction models: (i) Climatological models (ii) Surface-measured pressure, temperature, humidity NWM aims at representing (i) The daily weather (ii) The entire 3-dimensional weather field. Northern half of 3D refractivity (unitless) field on Aug 16, 2004, 22:45 UTC (vertical scale 100x)

4. 4 Methods 1. How to predict delays with NWM 2. How to test if the delays are not wrong 3. How to assess whether the delays improve GPS applications

5. 5 Numerical integration: Coordinate conversion: Interpolation: Refractivity calculation: Predicting delays with NWM

6. 6 Tropospheric corrections: NWM–radiosonde discrepancy Total (cm) Hydro- static (cm) Non- hydro- static (cm)

7. 7 GPS positioning Two scenarios for kinematic processing: –Moving rover: on board ferry boat –Stationary rover: one of two base stations In each scenario, test and reference solutions: Moving roverStationary rover Test Solutions Individual baselines from each base station to ferry boat Baselines from one fixed base station to another base acting as rover Reference Solution Multi-base station to ferry boat 5 day-long static PPP at each base station

8. 8 Impact assessment 3 tropospheric prediction models assessed: –NWM –UNB3m –Saastamoinen with standard weather parameters –uncorrected observations (no model) Criteria: –Discrepancy in rover position between test and reference solutions –RMS of observation residuals

9. 9 Stationary rover: test–reference discrepancy

10. 10 Moving rover: discarding unreliable epochs Saint John Digby

11. 11 Moving rover (1): test–reference discrepancy

12. 12 Moving rover (2): test–reference discrepancy

13. 13 Conclusions and future work NWM has only marginal improvement on that particular 70 km baseline. Validation: vertical coordinates in NWM Study how it is handled in data assimilation Test it with GPS-equipped radiosondes Impact assessment: varying-length baselines Kinematic processing at stationary rover From 100 to 1,000 km

14. 14 Publications Paper at ION AM 2005 in Boston, MA Poster at AGU Joint Assembly in Washington, DC Paper at ION GNSS 2006 in Fort Worth, TX

15. 15 Thanks! Questions? Felipe G. Nievinski

16. 16 GPS signals are refracted in the Earth’s neutral atmosphere. –Hence timings (rangings) are delayed (increased). –~ 2.5 m at zenith direction, ~ 25 m at 5º elevation angle (for a station on the geoid) satellite receiver Neutral-Atmosphere Delays

17. 17 Tropospheric corrections: NWM self-discrepancy NWM Ray-trace vs Radiosonde ray-trace NWM Ray-trace vs NWM Saastamoinen Hydro- static (cm) Hydro- static (cm)