Presentation is loading. Please wait.

Presentation is loading. Please wait.

The GSI Capability to Assimilate TRMM and GPM Hydrometeor Retrievals in HWRF Ting-Chi Wu a, Milija Zupanski a, Louie Grasso a, Paula Brown b, Chris Kummerow.

Similar presentations


Presentation on theme: "The GSI Capability to Assimilate TRMM and GPM Hydrometeor Retrievals in HWRF Ting-Chi Wu a, Milija Zupanski a, Louie Grasso a, Paula Brown b, Chris Kummerow."— Presentation transcript:

1 The GSI Capability to Assimilate TRMM and GPM Hydrometeor Retrievals in HWRF Ting-Chi Wu a, Milija Zupanski a, Louie Grasso a, Paula Brown b, Chris Kummerow ab, and John Knaff c a CIRA/CSU b Department of Atmospheric Science/CSU c NOAA/Center for Satellite Applications and Research 14th Technical Review Meeting & Science Workshop on Satellite Data Assimilation May 31-June 2, Monterey, California 1

2 Some Facts More than 90% of all observations used in global NWP systems come from satellite. However, limitations in NWP systems and computing resources prevents 95% of collected satellite data from being assimilated… Satellite instruments do not observe physical quantities such as temperature, wind, humidity, etc. – Satellite instruments measure radiation emitted from the surface and atmosphere of the Earth. – Radiance is related to T, Q, cloud variables, etc via radiative transfer equation (e.g. CRTM) Satellite data assimilation: radiance vs. retrievals 2

3 Hurricane WRF (HWRF) System HWRF 2014 (v3.6a) Operational Implementation GSI: Gridpoint Statistical Interpolation HWRF employs the hybrid GSI (3DVar+Ens) static + ensemble background error covariance WRF NMM core 3

4 HWRF Domain (v3.6a)  The inner-most domain does not assimilate the majority of the satellite observations.  Information contained in the hurricane core is important  Conventional observations are sparse over ocean Domain (Δx)sizeDA (Hybrid 20% static + 80% ensemble Δx=50km) Conv. Obs assimilated Sat. Radiance assimilated d01 (27 km)216 x 432Sort of Clear sky only ghost d02 (9km)106 x 204Yes (H: 300km/V:600 hPa)YesClear sky only ghost d03 (3km)198 x 354Yes (H: 150km/V:600 hPa)YesNone 4

5 5

6 Hurricane GPROF Transformed into 2 vertically integrated quantities: integrated solid water content (SWC) in kg m -2 integrated liquid water content (LWC) in kg m -2 solid = ice + 1/2 of mixed-phase water liquid = cloud water + rain + 1/2 of mixed-phase water Hurricane GPROF incorporates HURDAT2 for TMI/GMI: instantaneous rainfall rate hydrometeor profiles : cloud water, rain, mixed-phase, and ice The latest Goddard PROFiling (GPROF v2014) retrieves precipitation rate and vertical profiles of hydrometeors from TMI and GMI, but contains little hurricane data. 6

7 How to Assimilate Them? Develop and implement new observation operators to assimilate integrated SWC and integrated LWC. Two possible approaches: 1. Add hydrometeor control variables (Future Work) 2. Extend the impact to the current set of control variables in operational HWRF-GSI that includes T (temperature), P s (surface pressure), and q (specific humidity). Assumption: All water vapor in excess of the saturation is immediately condensed out. The observation operators (h solid and h liquid ) are defined as a vertical integration of water vapor mixing ratio in excess of saturation with respect to ice and liquid. 7

8 Observation Operators : h solid and h liquid q v : water vapor mixing ratio q si : saturation mixing ratio wrt ice q sl : saturation mixing ratio wrt liquid ρ : air mass density; f(T,P,q v ) Δz: layer thickness k: model vertical index k 0 : level where T=T 0 =273.15K k mix : level where T=T mix =253.15K Saturation vapor pressure  h solid and h liquid = f(T, P, q v ) = f(T, P s, q) 8

9 Single Observation Experiment The analysis increment (analysis - background) from such an experiment experiment can be used to analysis background analysis increment examine the structure of the background error covariance P f Understand how the observation information is distributed spatially among different variables. Two experiments 1-OBSSOLID and 1-OBSLIQUID Place a single observation of integrated SWC/LWC with value of 0.5 kg m -2 to the east of storm center noHybrid DA (pure 3DVar) vs. Hybrid DA (Var+Ens) i.e. 100% static vs. 20% static + 80% flow-dependent 9

10 noHybrid (pure 3DVar) 1-OBSSOLID (top) 1-OBSLIQUID (bottom) 1-OBSICE: maximum increment at upper troposphere 1-OBSLIQUID: maximum increment near ground increase moisture and lower temperature —> reach saturation 10

11 Hybrid (3DVar+Ens) 1-OBSSOLID (top) 1-OBSLIQUID (bottom) larger increments comparing to the nohybrid experiments, and extend vertically through a significant portion of troposphere Both experiments act to intensify the storm, in different degrees 11

12 Real Case: Hurricane Leslie (2012) ExpsObs Assimilated in ghost d02 Obs Assimilated in ghost d03 ConventionalSatelliteConventionalSatellite CTLYesClear skyYesNone AddWCYesClear skyYesIntegrated SWC + LWC First cycle starts at 2012/08/30 18 UTC for both experiments Run two-consecutive cycles We will focus on analysis from the second cycle 12

13 13

14 Analysis @ 2 nd cycle 14

15 HWRF Forecast from 2 nd cycle 15

16 Forecast Rain Rates Synthetic Satellite Image TMI PR 16

17 Future Work Extend current work by assimilate ATMS all-sky radiances using the operational HWRF system. See item (1) on slide 7. 17

18 Analysis: In general, moister and cooler air in mid to lower tropospheric layers of the hurricane core are evident in AddWC analysis, suggesting a tendency toward reaching saturation by lowering temperature and increasing moisture. Forecast: no obvious improvement in track, intensity, and size forecast. Similar results Gonzalo Possible causes of the inconclusiveness : – the analysis used to initialize an HWRF forecast is unbalanced – the existence of model errors in HWRF that are not accounted for by GSI – the absence of hydrometeor variables in the list of operational GSI control variables. Thank You Summary 18


Download ppt "The GSI Capability to Assimilate TRMM and GPM Hydrometeor Retrievals in HWRF Ting-Chi Wu a, Milija Zupanski a, Louie Grasso a, Paula Brown b, Chris Kummerow."

Similar presentations


Ads by Google