![Variograms of delay 07/27/11 Netherlands Australia Distance [km] InSAR WRF](https://image.slidesharecdn.com/the-role-of-weather-models-in-mitigation-of-tropspheric-delay-for-sar-interferometry-110727152539-phpapp01/85/The-role-of-weather-models-in-mitigation-of-tropspheric-delay-for-SAR-Interferometry-ppt-20-320.jpg)
The role of weather models in mitigation of tropspheric delay for SAR Interferometry.ppt
- 1. The role of numerical weather models (NWM) in mitigation of tropospheric delay for SAR Interferometry July 27, 2011 Shizhuo Liu 1 , Agnes Mika 2 , Wenyu Gong 3 , Franz Meyer 3 , Ramon Hanssen 1 , Don Morton 3 and Peter Webley 3 1 Delft institute of Earth Observation and Space Systems (DEOS), the Netherlands 2 BMT AGROSS, the Netherlands 3 University of Alaska Fairbanks, United States Department of Earth Observation and Space Systems (DEOS), Aerospace engineering InSAR WRF
- 2. Delay observed by repeat-pass SAR Interferometry 07/27/11 Temporal difference: Spatial difference: p q t1 t2 observed delay: (spatio-temporal difference)
- 3. Spatial characteristics of delay in InSAR 07/27/11 8 ERS1/2 tandem interferograms over Groningen, the Netherlands a b c d e f g h trend: c, e, h local anomaly: a, g trend+anomaly: b, d, f Trend + Variation (water vapor) mm
- 4. Delay in mountainous regions 07/27/11 = Trend + Variation (random) p q Atmospheric-only interferogram Hawaii topography h trend + variation + vertical stratification m mm
- 5. Studies of regions with different climates 07/27/11 Netherlands Hawaii Mexico City Lake Moore, WA
- 6. Forecasting setup WRF ( ver 3.1 ): includes non-hydrostatic dynamics; Spatial domains: 27, 9, 3, 1 km ; Spin-up time: 12-16 hours ; Initial-boundary condition: FNL data (100 km, 6 hours); Land topography data: SRTM (90 m); Land-use data (MODIS 20-category); Microphysics: Morrison 2-moment Vertical levels: 28 (10 under 2km) 07/27/11
- 7. Hawaii (case No.1) InSAR (35-day) 07/27/11 WRF Foster JGRL, vol. 33, 2006 mm InSAR - WRF InSAR WRF InSAR - WRF
- 8. Hawaii (case No.2) 07/27/11 InSAR (35-day) WRF InSAR - WRF InSAR WRF InSAR - WRF
- 9. Topography of Mexico City 07/27/11 m
- 10. Mexico City (case No.1) 07/27/11 InSAR (35-day) WRF InSAR-WRF InSAR WRF InSAR-WRF mm
- 11. Mexico City (case No.2) 07/27/11 InSAR (35-day) WRF InSAR - WRF InSAR WRF InSAR - WRF
- 12. Inconsistency (case No.3) 07/27/11 InSAR (35-day) WRF InSAR - WRF
- 13. Cross-validation with MERIS 07/27/11 WRF InSAR MERIS mm
- 15. Netherlands (9 cases) 07/27/11 InSAR (35-day) WRF InSAR-WRF No.1 mm No.2 No.3
- 16. Netherlands 07/27/11 InSAR (35-day) WRF InSAR-WRF No.4 No.5 No.6
- 17. Netherlands 07/27/11 InSAR (35-day) WRF InSAR-WRF No.7 No.8 No.9
- 18. Southwest Australia (5 cases) 07/27/11 InSAR (35-day) WRF InSAR-WRF No.1 No.2 No.3
- 19. Southwest Australia 07/27/11 InSAR (35-day) WRF InSAR-WRF No.4 No.5
- 20. Variograms of delay 07/27/11 Netherlands Australia Distance [km] InSAR WRF
- 21. Results review In mountainous regions, topography dependent delay is well predicted by WRF in most cases. In these cases, 40% to 50% delay reduction can be achieved. However, its reliability is not 100% (80%) In flat regions, delay prediction by WRF is unrealistic and hardly bring significant delay reduction Moreover, the spatio-temporal delay variation predicted by WRF is underestimated at all spatial scales 07/27/11
- 22. Model tuning Initial boundary conditions: FNL -> ECMWF (50 km) ; Longer spin-up time: 12 hours -> 24 hours ; Vertical levels: 28 -> 40 (30 below ABL) ; 07/27/11
- 23. ECMWF versus FNL 07/27/11 Mexico City (case No.3) same model settings
- 24. Netherlands (case No.2) 07/27/11 InSAR ECMWF(WRF) InSAR-ECMWF FNL(WRF) InSAR-FNL
- 25. Netherlands (case No.9) 07/27/11 InSAR ECMWF(WRF) InSAR-ECMWF FNL(WRF) InSAR-FNL
- 26. Longer spin-up time and more vertical levels 07/27/11 Hawaii Mexico City Netherlands Australia InSAR WRF tuned WRF original
- 27. NWM (numerical weather models) work for topography-dependent delay when topography variation is significant (> 2000 km) - max 50% RMS reduction with ; - a reliability of 80% (improvement for 4 out of 5) ; NWM fail for lateral variation of water vapor at small scales (< 50 km) - always underestimation ; - max 30% reduction ; - a poor reliability (improvement for 2 out of 14) The low reliability of NWM for flat regions excludes it from operational tools for delay mitigation in SAR Interferometry. For mountainous regions, delay correction could go wrong as well, users should be careful and critical Conclusions 07/27/11
- 28. Thank you ! 07/27/11
- 29. Is the weather model generally bad for delay prediction ? 07/27/11 MERIS WRF Absolute delay:
- 31. Recommendations To improve the reliability of NWM it is necessary to include more meteorological observations with high spatial density Hindcasting using observations after satellite acquisitions would be also useful to constrain NWM aiming to increase its reliability 07/27/11
- 32. Tropospheric delay experienced by MW 07/27/11 hydrostatic (gas components) wet (water vapor) cloud droplets absolute delay due to troposphere: hydrostatic: long wavelength spatial gradient(pressure, temperature), i.e., trend wet/cloud: significant spatial variation, i.e., local variation
- 33. Numerical forecasting for delay mitigation 07/27/11 Earth’s surface NWM prediction dh z(h) (T, e, P) P: total air pressure e : water vapour pressure T: air temperature x y C onstants (Davis et al., 1985) Refractivity is obtained by taking temporal and spatial difference in sequence
- 34. Hawaii (case No.3) 07/27/11 InSAR (35-day) WRF InSAR - WRF InSAR WRF InSAR - WRF
- 35. Hawaii (case No.4) 07/27/11 InSAR (35-day) WRF InSAR-WRF InSAR WRF InSAR-WRF
Editor's Notes
- #7: FNL: final analysis (google) Microphyiscs: Morrison 2-moment is a good scheme for high-resolution forecasting (WRF manual)