COSMIC IROWG 2017 The ROM SAF multi-mission reprocessing: Bending angle, refractivity, and dry temperature Stig Syndergaard Hans Gleisner Kent B. Lauritsen Johannes K. Nielsen Danish Meteorological Institute, Copenhagen, Denmark Special thanks to: Lars Ørum Rasmussen (IT-department at DMI) Santi Oliveras (IEEC) Joint COSMIC Tenth Data Users Workshop and IROWG-6 Meeting, Estes Park, Colorado, USA, September 21-27, 2017 1
Outline ROM SAF reprocessing overview O-B statistics and multi-mission time-series Non-optimized vs. optimized bending angle etc Metop 4-ways bending angle comparisons Summary and conclusions Joint COSMIC Tenth Data Users Workshop and IROWG-6 Meeting, Estes Park, Colorado, USA, September 21-27, 2017 2
ROM SAF reprocessing overview Datasets: GRM-28-R1: Multi-mission (2001-2016) GRM-29-R1: Metop-A and B (2006/2012-/2016) GRM-30-R1: COSMIC (2006-2016) GRM-32-R1: CHAMP (2001-2008) GRM-33-R1: GRACE (2007-2016) Contents: - Bending angles (Level 1b) - Profiles of refractivity and dry temperature (Level 2a) - Temperature, humidity, pressure (Level 2b) - Surface pressure, tropopause height (Level 2c) - Gridded data of all variables (Level 3) Joint COSMIC Tenth Data Users Workshop and IROWG-6 Meeting, Estes Park, Colorado, USA, September 21-27, 2017 3
Processing chain DATA PROCESSING STEPS ALGORITHMS Level 1a PHASES, AMPLITUDES, SATELLITE POS & VEL PROVIDED BY UCAR & EUMETSAT Latest available versions of data. Level 1b BENDING ANGLES (L1, L2, LC) Level 2a REFRACTIVITY DRY TEMPERATURE Level 2b TEMPERATURE SPECIFIC HUMIDITY GEOPOTENTIAL HEIGHT Level 3 BENDING ANGLE REFRACTIVITY DRY TEMPERATURE TEMPERATURE SPECIFIC HUMIDITY GEOPOTENTIAL HEIGHT GEOMETRIC OPTICS > 20 km WAVE OPTICS < 25 km IONOSPHERIC CORRECTION STATISTICAL OPTIMIZATION ABEL TRANSFORM HYDROSTATIC INTEGRAL 1D-VAR RETRIEVAL QUALITY CONTROL BINNING & AVERAGING SAMPLING ERROR CORRECTION Canonical Transform (CT2) below 25 km. Transition between 20 and 25 km. Linear Combination of L1 and L2 BAs. L1-L2 extrapolation at low altitudes. 2-parameter fit of observed BA to a best fitting BA profile found by global search in BAROCLIM. Downward integration of hydrostatic integral from 150 km. Upper boundary conditions from d(log(n))/dh. Using ERA-Interim short-term forecasts as a priori. A number of checks at bending angle, refractivity, dry temperature, and 1D-var processing steps. Using ERA-Interim shortterm forecasts as reference. Weighted averaging into 5 o latitude grids. Sampling error estimation based on 2.5 o x2.5 o ERA-Interim fields. Joint COSMIC Tenth Data Users Workshop and IROWG-6 Meeting, Estes Park, Colorado, USA, September 21-27, 2017 4
ROM SAF reprocessing overview Datasets: GRM-28-R1: Multi-mission (2001-2016) GRM-29-R1: Metop-A and B (2006/2012-/2016) GRM-30-R1: COSMIC (2006-2016) GRM-32-R1: CHAMP (2001-2008) GRM-33-R1: GRACE (2007-2016) Contents: - Bending angles (Level 1b) This presentation:{ - Profiles of refractivity and dry temperature (Level 2a) - Temperature, humidity, pressure (Level 2b) - Surface pressure, tropopause height (Level 2c) - Gridded data of all variables (Level 3) Joint COSMIC Tenth Data Users Workshop and IROWG-6 Meeting, Estes Park, Colorado, USA, September 21-27, 2017 5
O-B statistics dry temperature Metop Oct 2007 COSMIC Oct 2007 Joint COSMIC Tenth Data Users Workshop and IROWG-6 Meeting, Estes Park, Colorado, USA, September 21-27, 2017 6
O-B statistics dry temperature Metop Oct 2016 COSMIC Oct 2016 Joint COSMIC Tenth Data Users Workshop and IROWG-6 Meeting, Estes Park, Colorado, USA, September 21-27, 2017 7
O-B statistics dry temperature GRACE Oct 2007 CHAMP Oct 2007 Joint COSMIC Tenth Data Users Workshop and IROWG-6 Meeting, Estes Park, Colorado, USA, September 21-27, 2017 8
O-B statistics dry temperature GRACE Oct 2016 Multi-mission Oct 2016 Joint COSMIC Tenth Data Users Workshop and IROWG-6 Meeting, Estes Park, Colorado, USA, September 21-27, 2017 9
O-B statistics refractivity Metop Oct 2016 COSMIC Oct 2016 Note: N N = p p T T Joint COSMIC Tenth Data Users Workshop and IROWG-6 Meeting, Estes Park, Colorado, USA, September 21-27, 2017 10
Time-series refractivity Sep 2001 CHAMP Oct 2008 Apr 2006 COSMIC Dec 2016 Mar 2007 GRACE... Dec 2016 (RE1A Metop-A from Oct 2006) (RE1A Metop-B from Oct 2012) Oct 2007 Metop-A... Dec 2016 Feb 2013 Metop-B... Dec 2016 Joint COSMIC Tenth Data Users Workshop and IROWG-6 Meeting, Estes Park, Colorado, USA, September 21-27, 2017 11
Non-optimized vs. optimized bending angle Metop A+B, October 2016 Statistical optimization approach: Two-parameter fit of observed BA to a best fitting BA profile found by global search in BAROCLIM Background error fixed at 50%; dynamical estimated observation error QC assuring no more than 10% influence from background below 40 km Joint COSMIC Tenth Data Users Workshop and IROWG-6 Meeting, Estes Park, Colorado, USA, September 21-27, 2017 12
Mean difference consistency Metop A+B, October 2016 Against ECMWF forecasts Against ECMWF forecasts Against ECMWF forecasts Against ECMWF forecasts Approach minimizes influence from climatology in derived products preserves the mean Joint COSMIC Tenth Data Users Workshop and IROWG-6 Meeting, Estes Park, Colorado, USA, September 21-27, 2017 13
Mean difference consistency Metop A+B, October 2016 Differences due to different forecasts data are the same Joint COSMIC Tenth Data Users Workshop and IROWG-6 Meeting, Estes Park, Colorado, USA, September 21-27, 2017 14
Metop 4-ways bending angle comparisons EUMETSAT PHASES, AMPLITUDES & ORBITS Metop A+B October 2016 UCAR PHASES, AMPLITUDES & ORBITS EUMETSAT BENDING ANGLE RE1A BENDING ANGLE RE1B BENDING ANGLE UCAR BENDING ANGLE Version 1.4 hi-res (removed 178 outliers) Official ROM SAF CDR Alternative ROM SAF CDR Version 2016_0120 37471 matches 37047 matches 36231 matches same level 1a processing; different level 1b processing; different level 1a processing; same level 1b processing; same level 1a processing; different level 1b processing; Joint COSMIC Tenth Data Users Workshop and IROWG-6 Meeting, Estes Park, Colorado, USA, September 21-27, 2017 15
Metop 4-ways bending angle comparisons GO WO Very good consistency In the mean Note different scale except here WO Note: EUM data are processed with wave optics (WO) at all altitudes; standard deviation is smaller for the thinned (and smoothed) bending angle as shown in presentation by Axel von Engeln Joint COSMIC Tenth Data Users Workshop and IROWG-6 Meeting, Estes Park, Colorado, USA, September 21-27, 2017 16
Metop 4-ways bending angle comparisons GO WO Very good consistency In the mean except here WO Note: EUM data are processed with wave optics (WO) at all altitudes; standard deviation is smaller for the thinned (and smoothed) bending angle as shown in presentation by Axel von Engeln Joint COSMIC Tenth Data Users Workshop and IROWG-6 Meeting, Estes Park, Colorado, USA, September 21-27, 2017 17
Metop 4-ways bending angle comparisons GO WO Very good consistency In the mean except here WO Note: EUM data are processed with wave optics (WO) at all altitudes; standard deviation is smaller for the thinned (and smoothed) bending angle as shown in presentation by Axel von Engeln Joint COSMIC Tenth Data Users Workshop and IROWG-6 Meeting, Estes Park, Colorado, USA, September 21-27, 2017 18
Summary and Conclusions ROM SAF reprocessing #1 is done validation ongoing Comprises four missions (CHAMP, GRACE, COSMIC, Metop) and 15 years of data Focus here on validation of bending angle, refractivity, and dry temperature Validation indicate generally good quality and consistent global statistics over time and between different missions Minor issue to be looked into: larger bias for CHAMP & GRACE at low altitudes Metop 4-ways comparisons reveal consistent mean above the troposphere, but otherwise quite different error characteristics between UCAR, EUMETSAT Secr., and ROM SAF reprocessed bending angles We are grateful to UCAR/CDAAC and the EUMETSAT Secretariat for providing the low level data (phases, amplitudes, orbits) on which the ROM SAF reprocessing is based Joint COSMIC Tenth Data Users Workshop and IROWG-6 Meeting, Estes Park, Colorado, USA, September 21-27, 2017 19