G023-06
10-year global evaluation of tropospheric noise corrections derived from routinely leveraged weather models
Wednesday, 16 December 2020: 10:15
Virtual
Zhong Lu1, David P Bekaert2, Jeremy Maurer3, Yang Lei4, Simran Sangha5, Rohan Weeden6, Romain Jolivet7, Richard J Walters8, Falk Amelung9, Eric Jameson Fielding10, Prashant Kumar11, Zhenhong Li12 and Sean Buckley5, (1)Southern Methodist University, Dallas, TX, United States, (2)JPL/NASA/Caltech, Pasadena, CA, United States, (3)Missouri University of Science and Technology, Rolla, United States, (4)Caltech, Pasadena, CA, United States, (5)NASA Jet Propulsion Laboratory, Pasadena, CA, United States, (6)Alaska Satellite Facility, Fairbanks, AK, United States, (7)Ecole Normale Supérieure, PSL Research University, CNRS UMR 8538, Laboratoire de Géologie, Paris, France, (8)University of Durham, Durham, United Kingdom, (9)University of Miami, Department of Marine Geology and Geophysics, Miami, FL, United States, (10)Jet Propulsion Lab Caltech, Pasadena, CA, United States, (11)Atmospheric Sciences Division, AOSG, EPSA, Space Applications Centre, ISRO, Ahmedabad, India, (12)COMET, Newcastle University, School of Engineering, Newcastle Upon Tyne, United Kingdom
Abstract:
Tropospheric path noise severely deteriorates the retrieval accuracy of the surface displacement measurements from InSAR. It is a recurrent bottleneck among most of the common application areas including the studies of earthquake and volcano cycles, landslides, hydrogeology, cryosphere, and anthropocentric hazards among others. The upcoming NASA-ISRO SAR satellite mission has plans for including the tropospheric noise correction layer in its products, but has not settled on its definition.
Weather models are often used to reduce tropospheric noise in InSAR data, however no community consensus on which model is best suited for a given region or whether a temporal interpolation improves the delay estimate. We will present a global 10-year statistical analysis comparing GNSS with ECMWF-HRES, GMAO and ERA-5 weather models delays, in order to provide guidance on which model performs best over a given area and time-period. We will use variograms and seasonal amplitude metrics, evaluated over seasonal, monthly, annual, and decadal intervals to capture performance variability in time. Tropospheric delays are often linearly interpolated to the SAR acquisition UTC time. To evaluate the impact of such an interpolation method, we will perform a global 2-year analysis comparing tropospheric delays computed at various locations with GNSS delay time-series. We compute our zenith delays using our RAiDER (Raytracing-based Atmospheric Delay Estimator for RADAR) package, which we have released in the open-source domain, and which also includes the statistical framework leveraged in the presented work.
