WRF-EnKF Lightning Assimilation Real-Observation Experiments Overview

11

WRF-EnKF Lightning Assimilation WRF-EnKF Lightning Assimilation Real-Observation ExperimentsReal-Observation Experiments

OverviewOverview

Cliff Mass, Greg Hakim, Phil Regulski, Ryan TornCliff Mass, Greg Hakim, Phil Regulski, Ryan Torn

Department of Atmospheric SciencesDepartment of Atmospheric SciencesUniversity of WashingtonUniversity of Washington

February 15, 2008February 15, 2008

22

OverviewOverview

WRF-EnKF OverviewWRF-EnKF Overview Lightning Assimilation Technique ReviewLightning Assimilation Technique Review Project ReviewProject Review Model Estimate ExperimentsModel Estimate Experiments Case StudiesCase Studies

December 2002December 2002 October 2004October 2004 November 2006November 2006

ConclusionsConclusions In-progress/Future work In-progress/Future work

33

The Use of EnKF for Lightning The Use of EnKF for Lightning AssimilationAssimilation

In this work the ensemble Kalman filter is applied In this work the ensemble Kalman filter is applied using the WRF model (WRF-EnKF). using the WRF model (WRF-EnKF).

The covariance statistics from EnKF that spread the The covariance statistics from EnKF that spread the impact of observations are flow-dependentimpact of observations are flow-dependent

Lightning observations are ideal to provide WRF-Lightning observations are ideal to provide WRF-EnKF with more observational information in areas EnKF with more observational information in areas with few in situ observations (e.g. Pacific Ocean)with few in situ observations (e.g. Pacific Ocean)

Impact of lightning observations will propagate into Impact of lightning observations will propagate into areas of high forecast impact (West coast and beyond)areas of high forecast impact (West coast and beyond)

44

Lightning Assimilation TechniquesLightning Assimilation Techniques

Original Lightning Experiment (LTNG2)Original Lightning Experiment (LTNG2) NLDN/LR lightning strike is detectedNLDN/LR lightning strike is detected Lightning strike is converted into lightning rate from Lightning strike is converted into lightning rate from

nearby LTNG observationsnearby LTNG observations Lightning rate converted into convective rainfall rate using Lightning rate converted into convective rainfall rate using

Pessi/Businger convective rain rate/lightning rate Pessi/Businger convective rain rate/lightning rate relationshiprelationship

Convective rainfall (mm) is assimilated into WRF-EnKFConvective rainfall (mm) is assimilated into WRF-EnKF Successful test on Dec. 2002 case Successful test on Dec. 2002 case

55

Lightning Assimilation TechniquesLightning Assimilation Techniques

Minimal Counting Technique Lightning Experiment (LTNG4)Minimal Counting Technique Lightning Experiment (LTNG4) NLDN/LR lightning strike is detectedNLDN/LR lightning strike is detected Lightning strike is converted into lightning rate from nearby Lightning strike is converted into lightning rate from nearby

LTNG observationsLTNG observations Once any nearby lightning strikes are used to calculate a Once any nearby lightning strikes are used to calculate a

lightning density they are no longer available as an lightning density they are no longer available as an assimilation point (although they are still used to calculate assimilation point (although they are still used to calculate LTNG densities)LTNG densities)

Lightning rate converted into convective rainfall rate using Lightning rate converted into convective rainfall rate using Pessi/Businger convective rain rate/lightning rate relationshipPessi/Businger convective rain rate/lightning rate relationship

Convective rainfall (mm) is assimilated into WRF-EnKFConvective rainfall (mm) is assimilated into WRF-EnKF Successful testing on Dec. 2002 caseSuccessful testing on Dec. 2002 case

66

Project ReviewProject Review

Research CompletedResearch Completed Dec. 2002 Test Case - Analysis and 12-hr ForecastsDec. 2002 Test Case - Analysis and 12-hr Forecasts Oct. 2004 Test Case - AnalysisOct. 2004 Test Case - Analysis Nov. 2006 Test Case - AnalysisNov. 2006 Test Case - Analysis Modifications to LTNG2 and LTNG4 models to improve analysis and Modifications to LTNG2 and LTNG4 models to improve analysis and

forecastsforecasts In-progressIn-progress

Forecast analysis of Oct. 2004 Test CaseForecast analysis of Oct. 2004 Test Case Forecast analysis of Nov. 2006 Test CaseForecast analysis of Nov. 2006 Test Case 1-hr assimilation of cumulative convective rain rate data (previously 1-hr assimilation of cumulative convective rain rate data (previously

using 6-hr cumulative totals) for LTNG2 and LTNG4 modelsusing 6-hr cumulative totals) for LTNG2 and LTNG4 models

77

Recent ExperimentsRecent ExperimentsCan we further improve Dec 2002 performance?Can we further improve Dec 2002 performance?

Model-estimate and observations of cumulative convective precipitation calculated Model-estimate and observations of cumulative convective precipitation calculated by LTNG density occasionally have large innovations leading to large increments by LTNG density occasionally have large innovations leading to large increments in the model’s dynamical fields, possibly leading to locally unbalanced statesin the model’s dynamical fields, possibly leading to locally unbalanced states

Set a upper bound on assimilated cumulative convective precipitationSet a upper bound on assimilated cumulative convective precipitation

Minimal Counting Technique Lightning Experiment (LTNG4) w/ modifications Minimal Counting Technique Lightning Experiment (LTNG4) w/ modifications (LTNG5)(LTNG5)

Same techniques as LTNG4 but with upper bound (18mm)Same techniques as LTNG4 but with upper bound (18mm) Less improvement in analysis and forecastsLess improvement in analysis and forecasts Original LTNG4 still best performing model to test further on new regimesOriginal LTNG4 still best performing model to test further on new regimes

Original Lightning Experiment (LTNG2) w/ modifications (LTNG6)Original Lightning Experiment (LTNG2) w/ modifications (LTNG6) Same techniques as LTNG2 with upper bound (18mm)Same techniques as LTNG2 with upper bound (18mm) Less improvement in analysis and forecastsLess improvement in analysis and forecasts Original LTNG2 still best performing model to test further on new regimesOriginal LTNG2 still best performing model to test further on new regimes

Another solutionAnother solution Reduce cumulative value of convective rain assimilated from 6- to 1-hr block Reduce cumulative value of convective rain assimilated from 6- to 1-hr block (In-progress)(In-progress)

88

Case Study #1 – December 2002Case Study #1 – December 2002Minimum SLP recorded at extra-tropical cyclone’s centerMinimum SLP recorded at extra-tropical cyclone’s center

Limiting upper bound of cumulative convective

precipitation degrades analysis performance (LTNG5/6)NCEP (black dashed) is NCEP subjective analysis

99

Case StudiesCase Studies

Case #1: December 16-21, 2002Case #1: December 16-21, 2002 AnalysisAnalysis

LTNG Analysis Impact of SLP, H500 fieldsLTNG Analysis Impact of SLP, H500 fields 12-hr Forecasts12-hr Forecasts

GFS Analysis v. Fcst 12-hr ControlGFS Analysis v. Fcst 12-hr Control GFS Analysis v. Fcst 12-hr LTNG2GFS Analysis v. Fcst 12-hr LTNG2 GFS Analysis v. Fcst 12-hr LTNG4GFS Analysis v. Fcst 12-hr LTNG4

1010

Major FindingsMajor Findings

EnKF allows lightning to have a EnKF allows lightning to have a substantial influence on substantial influence on atmospheric analyses atmospheric analyses

The influence can be widespread The influence can be widespread over the domainover the domain

1111

Case Study #1 – December 2002Case Study #1 – December 2002Impact of Lightning Assimilation: LTNG2Impact of Lightning Assimilation: LTNG2

1212


1313


1414


1515


1616


1717


1818


1919


2020


2121


2222


2323


2424


2525

The Impact on Forecasts The Impact on Forecasts Were MixedWere Mixed

2626

Case Study #1 – December 2002Case Study #1 – December 200212-hr Forecast Comparison: Control12-hr Forecast Comparison: Control

2727

Case Study #1 – December 2002Case Study #1 – December 200212-hr Forecast Comparison: LTNG212-hr Forecast Comparison: LTNG2

2828


2929


3030


3131


3232


3333


3434


3535


3636


3737


3838


3939


4040


4141


4242


Case #2: October 4-7, 2004Case #2: October 4-7, 2004 AnalysisAnalysis

LTNG Analysis Impact of SLP, H500 fieldsLTNG Analysis Impact of SLP, H500 fields Number of LTNG strikes during test case is much Number of LTNG strikes during test case is much

smaller than Dec. 2002 casesmaller than Dec. 2002 case

4343

Case Study #2 – October 2004Case Study #2 – October 2004Impact of Lightning Assimilation: LTNG4Impact of Lightning Assimilation: LTNG4

4444


4545


4646


4747


4848


4949


5050


5151


5252


Case #3: November 8-12, 2006Case #3: November 8-12, 2006 AnalysisAnalysis

LTNG Analysis Impact of SLP, H500 fieldsLTNG Analysis Impact of SLP, H500 fields

5353

Case Study #3 – November 2006Case Study #3 – November 2006Impact of Lightning Assimilation: LTNG4Impact of Lightning Assimilation: LTNG4

5454


5555


5656


5757


5858


5959


6060


6161

ConclusionsConclusions

Assimilated lightning observations impact analysesAssimilated lightning observations impact analyses Case 1: Lightning observations improving location of front Case 1: Lightning observations improving location of front

associated with extratropical cycloneassociated with extratropical cyclone Cases 2 & 3 forecasts will be impacted by lightning Cases 2 & 3 forecasts will be impacted by lightning

observationsobservations Mixed forecast improvements seen in 12 hr forecasts Mixed forecast improvements seen in 12 hr forecasts

for Dec. 2002 Test Casefor Dec. 2002 Test Case

6262

In-progress/Future WorkIn-progress/Future Work

In-ProgressIn-Progress Reduce WRF cumulative convective precipitation assimilation from 6- to 1-hr Reduce WRF cumulative convective precipitation assimilation from 6- to 1-hr

to avoid unbalanced statesto avoid unbalanced states 24-hr Forecasts for Oct 2004 Case24-hr Forecasts for Oct 2004 Case 24-hr Forecasts for Nov 2006 Case24-hr Forecasts for Nov 2006 Case

Future WorkFuture Work Test Case #4 – December 2006Test Case #4 – December 2006 Investigate robustness of Pessi/Businger lightning rate/convective rain rate Investigate robustness of Pessi/Businger lightning rate/convective rain rate

relationshiprelationship Implement UW-ATMS lightning rate/convective rain rate relationshipImplement UW-ATMS lightning rate/convective rain rate relationship Investigate lightning/graupel relationshipInvestigate lightning/graupel relationship

Questions/Comments?Questions/Comments?

WRF-EnKF Lightning Assimilation Real-Observation Experiments Overview

Documents

lightning density

lightning assimilationin

convective rainfall

nearby lightning strikes

wrf model wrfenkf

test caseforecast analysis

ltng densitieslightning

test case analysisnov