Observation Sensitivity Experiments with a High Altitude - PowerPoint PPT Presentation

Observation Sensitivity Experiments with a High Altitude Meteorological Analysis System John McCormack Space Science Division, Washington DC David Kuhl Remote Sensing Division, Washington DC SPARC Data Assimilation Workshop 27 October 2017 Reading, UK

Observation Sensitivity: High-altitude DA Earlier studies (e.g., Xu et al., JGR 2011) have shown that DA using standard meteorological obs in the troposphere and limited stratospheric obs is enough to capture mean state of winds/temperatures in the mesosphere and lower thermosphere (MLT). To what extent are middle atmospheric (10-90 km) obs. needed to accurately specify MLT conditions, particularly during periods when middle atmosphere dynamics are rapidly evolving (e.g., sudden warmings)? To address this question, we compare two sets of meteorological analyses from the HA-NAVGEM system with and without middle atmospheric obs., and validate against independent ground-based data. We also present comparisons with independent satellite data from SOFIE U.S. Naval Research Laboratory SPARC DA Workshop 2017

High-Altitude NAVGEM NWP System: DA component Hybrid-4DVAR algorithm (Kuhl at al., 2013) uses linear combination of static and flow- dependent background error covariances Investigating E4D-Var Physics Short forecast: 9 hours Over 3M obs in 6-hour DA window Background Data ( x b ) Forecast Assimilation Long forecast: Analysis Radiosondes, Pibals, Dropsondes, Driftsonde Observations Model (DA) System (x a ) 10 days Land and Ship Surface Obs, Fixed and Drifting (y) NAVGEM E E4D-Var Buoys, Aircraft Obs Surface Winds: WindSat, Observation Error (R) ASCAT, SSMIS Ensemble Error Feature Tracked Winds: AVHRR, MODIS, VIIRS 9-hr Ensemble Covariance Ensemble Total Water Vapor: SSMI/SSMIS i ) Forecast (x b Testing (B) Generation GPS Bending Angle E4D-Var IR Sounding: IASI and AIRS (NPP CrIS) MW Sounding: AMSU-A (Ch 4-14), SSMIS (Ch 2-7, 22-24), SSMIS/MHS, NPP ATMS Middle Atmosphere: SABER T, MLS T, MLS O3, MLS H2O, SSMIS/UAS HYBRID α=0.5 4DVAR α=0 U.S. Naval Research Laboratory Presentation Title | 3

HA-NAVGEM: Middle Atmosphere Obs. Data coverage for SABER (blue), Aura MLS (red) and SSMIS/UAS on F16, F17, F18 DMSP platforms within a 6-hour assimilation window centered on 12 UTC 30 Jan 2010. after Hoppel et al. Mon. Wea. Rev., 2013. U.S. Naval Research Laboratory SPARC DA Workshop 2017

NAVGEM High Altitude NWP System: Global forecast model component Model Grid Spacing NAVGEM T119 L74 SSMIS UAS Channels nav_L74 NOGAPS-ALPHA: lm = 74, nprlev = 38 120 L74 NAVGEM 0.0001 Sponge layer 100 0.001 0.01 pressure altitude (km) 80 pressure (hPa) 0.1 60 1 40 10 20 100 0 1000 -180 -90 0 90 180 longitude ( o E) Forecast Model Setup • Semi-Lagrangian/semi-implicit dynamical core; T119 spectral truncation (grid spacing ~110 km) • 74 hybrid s -p levels, top at 6 x 10 -5 hPa or ~116 km ( for now ); Time step = 360 sec. Middle Atmosphere Physics: • Parameterized non-orographic gravity wave drag, O 3 & H 2 O photochemistry, exothermic chemical heating For more information see McCormack et al., JASTP 2017 and references therein. U.S. Naval Research Laboratory

HA-NAVGEM Geopotential Height Jan 2013 3-hourly global synoptic analyses With of u, v, T, Z, O3, middle H2O etc. atm. obs. U.S. Naval Research Laboratory SPARC DA Workshop 2017

HA-NAVGEM Geopotential Height Jan 2013 3-hourly global synoptic analyses No middle of u, v, T, Z, O3, atm. obs. H2O etc. U.S. Naval Research Laboratory SPARC DA Workshop 2017

Comparisons with meteor radar winds NH winter 2012-2013 HA-NAVGEM T119L74 Analyses - Global 1 o lat/lon 3 hourly output - Nov 2012 – March 2013 (SSW early Jan 2013) - Compare analyses with and without MLS/SABER/UAS middle atmosphere obs. Meteor radar observation sites Location Periods Analyzed 1-18 Dec. 2009, 1-26 Jan 2010, 12-28 Feb 2010 Andenes 69.3 o N , 16.0 o E 1-20 Dec. 2012, 1-28 Jan 2012, 1-24 Feb. 2013 Trondheim 63.4 o N , 10.5 o E 1 Dec. 2012 – 28 Feb. 2013 1 Dec. 2009 – 28 Feb. 2010 Juliusruh 54.6 o N , 13.4 o E 1 Dec. 2012 – 28 Feb. 2013 1 Dec. 2009 – 28 Feb. 2010 Collm 51.3 o N , 13.0 o E 1 Dec. 2012 – 28 Feb. 2013 1 Dec. 2009 – 28 Feb. 2010 CMOR 43.3 o N , 80.0 o W 1 Dec. 2012 – 26 Feb. 2013 1 Dec. 2009 – 28 Feb. 2010 Bear Lake 41.9 o N , 111.4 o W 1 Dec. 2012 – 28 Feb. 2013 Ascension Is. 8.0 o S , 14.4 o W 1 Jan. 2010 – 31Mar. 2010 Tierra del U.S. Naval Research Laboratory 53.7 o S , 67.7 o W 1-31 Dec. 2012, 1 Feb. – 31 Mar 2013 Fuego Rothera 67.5 o S, 68.0 o W 15 Jan 2013 – 28 Feb 2013 McCormack et al., JASTP (2017)

NAVGEM vs. Meteor radar V at 88 km HA-NAVGEM 3-hourly analyzed winds Radar 1-hourly observations Day 201301 U.S. Naval Research Laboratory SPARC DA Workshop 2017

NAVGEM vs. Meteor radar V at 88 km HA-NAVGEM 3-hourly analyzed winds Radar 1-hourly observations Day 201301 U.S. Naval Research Laboratory SPARC DA Workshop 2017

NAVGEM vs. Meteor radar V at 88 km Radar 1-hourly observations HA-NAVGEM 3-hourly analyzed winds with middle atm. obs. Histogram of meridional winds for DJG 2012-2013 U.S. Naval Research Laboratory SPARC DA Workshop 2017

NAVGEM vs. Meteor radar V at 88 km Radar 1-hourly observations HA-NAVGEM 3-hourly analyzed winds, no middle atm. obs. Histogram of meridional winds for DJG 2012-2013 U.S. Naval Research Laboratory SPARC DA Workshop 2017

Composite V profiles at Juliusruh Jan 2013 SPARC DA Workshop 2017

Composite V profiles at Juliusruh Jan 2013 SPARC DA Workshop 2017

Composite V time series at Juliusruh Jan 2013 SPARC DA Workshop 2017

Mean Temperature Bias w.r.t. SOFIE 1. Interpolated NAVGEM reanalyses to precise latitudes, longitudes and times of each of the 1624 SOFIE limb occultations. 2. Interpolated SOFIE temperature profile onto Eckermann et al. (in preparation) NAVGEM full model layers

Summary • For Jan 2013 case studied here, impact of middle atm. obs. on HA-NAVGEM analyses is most pronounced above 0.1 hPa level. . • Comparisons with independent radar winds at Juliusruh show HA-NAVGEM without middle atm. obs. overestimates amplitude of semi-diurnal tide. HA-NAVGEM with middle atm. obs. has trouble getting phase of semi-diurnal tide correct above ~90 km. • Hybrid 4DVAR produces lower temperature bias compared to standard 4DVAR in SOFIE comparisons at/below 90 km. • Some questions going forward: Ø What other independent observations are out there for validating stratosphere/mesosphere reanalyses? Ø What are metrics we can use to validate high-altitude reanalyses? Ø How good are 1, 5, 10, 30-day forecasts using these reanalyses as initial conditions . U.S. Naval Research Laboratory

Acknowledgments Many thanks to the MLS and SABER instrument teams, to the meteor radar PIs, and to NRL-MRY NAVGEM development team. This work was supported by the Chief of Naval Research. U.S. Naval Research Laboratory SPARC DA Workshop 2017

Obs. Sensitivity Tests Do We Need Middle Atmosphere Obs.? U.S. Naval Research Laboratory Latitude