12
Warn-on-Forecast and High-Impact Weather Workshop, February 6-7, 2013, National Weather Center, Norman, OK Utility of GOES-R geostationary lightning mapper (GLM) using hybrid variational- ensemble data assimilation in regional applications Milija Zupanski, Karina Apodaca, and Man Zhang Cooperative Institute for Research in the Atmosphere Colorado State University Fort Collins, Colorado, U. S. A. [ http://www.cira.colostate.edu/projects/ensemble/ ] Acknowledgements: - Louis Grasso, John Knaff, Mark DeMaria, Steve Lord - JCSDA - NOAA NESDIS/GOES-R

Acknowledgements: Louis Grasso, John Knaff , Mark DeMaria , Steve Lord JCSDA NOAA NESDIS/GOES-R

  • Upload
    etta

  • View
    39

  • Download
    0

Embed Size (px)

DESCRIPTION

- PowerPoint PPT Presentation

Citation preview

Page 1: Acknowledgements: Louis Grasso, John  Knaff , Mark  DeMaria , Steve Lord JCSDA  NOAA NESDIS/GOES-R

Warn-on-Forecast and High-Impact Weather Workshop,February 6-7, 2013, National Weather Center, Norman, OK

Utility of GOES-R geostationary lightning mapper (GLM) using hybrid variational-ensemble

data assimilation in regional applications

Milija Zupanski, Karina Apodaca, and Man Zhang

Cooperative Institute for Research in the AtmosphereColorado State University

Fort Collins, Colorado, U. S. A.[ http://www.cira.colostate.edu/projects/ensemble/ ]

Acknowledgements:- Louis Grasso, John Knaff, Mark DeMaria, Steve Lord- JCSDA - NOAA NESDIS/GOES-R

Page 2: Acknowledgements: Louis Grasso, John  Knaff , Mark  DeMaria , Steve Lord JCSDA  NOAA NESDIS/GOES-R

Goals of the project(1) Develop capability to use GOES-R Geostationary Lightning Mapper (GLM) observations in

prototype hybrid variational-ensemble data assimilation system (HVEDAS) (2) Evaluate its impact in regional data assimilation (DA) applications to severe weather (3) If there is a NOAA interest in further investigation/implementation of GLM observations in data

assimilation, support such an effort in collaboration with EMC.

• Benchmark system incorporates- WRF-NMM- Vertical updraft lightning observation operator- WWLLN lightning flash rate (proxy for GOES-R GLM)- Prototype hybrid variational-ensemble DA system (Maximum Likelihood Ensemble

Filter)

• Enhanced system additionally incorporates- GSI+CRTM forward (nonlinear) operators

- All-sky SEVIRI IR radiances (proxy for GOES-R ABI)- All-sky MW radiances (AMSU-A)- Vertical profiles of T and Q (AIRS, IASI)- NOAA HWRF- Hydrometeor-based lightning observation operator (e.g., McCaul et al. 2009)

Page 3: Acknowledgements: Louis Grasso, John  Knaff , Mark  DeMaria , Steve Lord JCSDA  NOAA NESDIS/GOES-R

All-sky radiance assimilationControl runAMSU-A NOAA-16 retrieved cloud liquid water

obs

IR: SEVIRI cloudy radiance assimilation (Total cloud condensate)

Enhanced system with MLEF-HWRF: Assimilation of cloudy radiances(M. Zhang et al. 2013a,b)

- Assimilation is able to improve clouds in TC - Improved TC intensity- Marginal (but positive) impact on TC track

MW: AMSU-A cloudy radiance assimilation (Total cloud condensate)Radar obs Clear-sky radiance assimilation All-sky radiance assimilation

Page 4: Acknowledgements: Louis Grasso, John  Knaff , Mark  DeMaria , Steve Lord JCSDA  NOAA NESDIS/GOES-R

Benchmark system: experimental setup- NOAA WRF-NMM model at 27km / 9km resolution- Use MLEF as a prototype HVEDAS- 32 ensembles- 6-hour assimilation interval- World Wide Lightning Location Network (WWLLN) observations- Control variables: PD, T, Q, U, V, CWM

Surface weather mapValid 04/27/2011 at 00UTC

SPC storm reportsValid 04/27/2011

Focus on 9 km inner domain

Tornado outbreak of April 27-28, 2011, southeastern U.S.

Page 5: Acknowledgements: Louis Grasso, John  Knaff , Mark  DeMaria , Steve Lord JCSDA  NOAA NESDIS/GOES-R

Lightning flash rate observation operatorCurrent version:

- maximum vertical velocity- works with any microphysics, but less accurate

Next version:- cloud hydrometeor based (graupel flux, cloud ice – McCaul et

al. 2009) - requires more advanced microphysics, but more realistic

Evaluation steps for new observation type:

1. Observation bias/pdf- check skewness of probability density function

2. Single observation experiment- analysis response, impact on model initial conditions

3. Observation information measure- quantify impact of observations in assimilation

4. Physical interpretation of data assimilation- check whether analysis correction appears physically consistent

Page 6: Acknowledgements: Louis Grasso, John  Knaff , Mark  DeMaria , Steve Lord JCSDA  NOAA NESDIS/GOES-R

Lightning observation bias / pdf

Original formulation

Corrected formulation

histogram/pdf

Normalized innovation vector:

Since the original pdf is skewed, need to correct observation operator Introduce multiplication parameter a and minimize cost function

histogram/pdf

Page 7: Acknowledgements: Louis Grasso, John  Knaff , Mark  DeMaria , Steve Lord JCSDA  NOAA NESDIS/GOES-R

Assimilation of WWLLN lightning observations: single observation experiment

Relevant for data assimilation: lightning observations impact initial conditions of model dynamical variables

Impact of a single lightning observation on the analysis:

Q increment at 700 hPaValid 04/27 at 12UTC

T increment at 700 hPaValid 04/27 at 12UTC

Wind increment at 700 hPaValid 04/27 at 12UTC

Page 8: Acknowledgements: Louis Grasso, John  Knaff , Mark  DeMaria , Steve Lord JCSDA  NOAA NESDIS/GOES-R

Observation information content using Shannon information measures

Since eigenvalues of the matrix ZTZ are known and the matrix inversion is defined in ensemble space, the flow-dependent DFS can be computed

In ensemble DA methods DFS can be computed exactly in ensemble subspace:

Change of entropy / degrees of freedom for signal (DFS)

• Gaussian pdf greatly reduce the complexity since entropy is related to covariance

Change of entropy due to observations

Use Shannon information (e.g. entropy) as an objective, pdf-based quantification of information (Rodgers 2000; Zupanski et al. 2007)

Entropy

Page 9: Acknowledgements: Louis Grasso, John  Knaff , Mark  DeMaria , Steve Lord JCSDA  NOAA NESDIS/GOES-R

Assimilation of WWLLN lightning observations:Degrees of Freedom for Signal

- Time-dependent information content - Shows the actual use of observations in each data assimilation cycle- Pixels correspond to error covariance localization used in DA

Cycle 104/27/11 at 00UTC

Cycle 304/27/11 at 12UTC

Cycle 504/28/11 at 00UTC

Page 10: Acknowledgements: Louis Grasso, John  Knaff , Mark  DeMaria , Steve Lord JCSDA  NOAA NESDIS/GOES-R

Assimilation of WWLLN lightning observations: Local impact on storm environment

Analysis increments:

Lightning data assimilation increases the advection of low-level vorticity into the region of large CAPE

Wind increment at 850 hPaValid 04/28 at 00UTC

Vorticity increment at 850 hPaValid 04/28 at 00UTC

Background CAPEValid 04/28 at 00UTC

Page 11: Acknowledgements: Louis Grasso, John  Knaff , Mark  DeMaria , Steve Lord JCSDA  NOAA NESDIS/GOES-R

Apodaca, K., M. Zupanski, M. Zhang, M. DeMaria, L. D. Grasso, J. A. Knaff, and G. DeMaria, 2013: Evaluating the potential impact of assimilating GOES-R GLM satellite lightning observations. To be submitted to Tellus. (Feb 2013)

Zhang, M., M. Zupanski, M.-J. Kim, and J. Knaff, 2013a: Direct Assimilation of all-sky AMSU-A Radiances in TC inner core: Hurricane Danielle (2010). Mon. Wea. Rev., accepted with minor revisions.

Zhang, M., M. Zupanski, and J. Knaff, 2013b: Impact assessment of SEVIRI data assimilation for Hurricane model initialization. To be submitted to Q. J. Roy. Meteorol. Soc. (April 2013)

Zupanski M., 2013: All-sky satellite radiance data assimilation: Methodology and Challenges. Data Assimilation for Atmospheric, Oceanic, and Hydrologic Applications, S.-K. Park and L. Xu, Eds, Springer-Verlag Berlin, in print.

Related publications

Page 12: Acknowledgements: Louis Grasso, John  Knaff , Mark  DeMaria , Steve Lord JCSDA  NOAA NESDIS/GOES-R

WRF-NMM, MLEF, and WWLLN observations combined in a prototype regional HVEDAS

Maximum updraft-based lightning observation operator requires on-line correction Preliminary results encouraging

Summary

Future Work

Use of more advanced lightning observation operator (McCaul et al. 2009) Combined assimilation of WWLLN and NCEP observations (e.g., GSI+CRTM) Combined assimilation of all-sky MW, IR (ABI) radiances and lightning (GLM) Conduct a thorough evaluation of the value-added impact of lightning data in

regional data assimilation applications to:- Tropical cyclones- Severe weather- Focus on forecast evaluation