September 29, 2009 1

QPEQPE

Bob KuligowskiBob KuligowskiNOAA/NESDIS/STARNOAA/NESDIS/STAR

Walt PetersenWalt PetersenNASA-MSFCNASA-MSFC

GOES-R Science Meeting

September 29, 2009 2

OutlineOutline

SCaMPR Review and PlansSCaMPR Review and Plans GOES-RRR Proposal: Improved MW RetrievalsGOES-RRR Proposal: Improved MW Retrievals MSFC QPE WorkMSFC QPE Work Questions / Discussion (all)Questions / Discussion (all)

September 29, 2009 3

SCaMPR ReviewSCaMPR Review

SSelf-elf-CaCalibrating librating MMultivariate ultivariate PPrecipitation recipitation RRetrievaletrieval Calibrates IR predictors to MW rain ratesCalibrates IR predictors to MW rain rates Updates calibration whenever new target data (MW Updates calibration whenever new target data (MW

rain rates) become availablerain rates) become available Selects both best predictors and calibration Selects both best predictors and calibration

coefficients; thus, predictors used can change in coefficients; thus, predictors used can change in time or spacetime or space

Flexible; can use any inputs or target dataFlexible; can use any inputs or target data

September 29, 2009 4

SCaMPR Overview SCaMPR Overview (cont.)(cont.)

Two calibration steps:Two calibration steps:» Rain / no-rain discrimination using discriminant Rain / no-rain discrimination using discriminant

analysisanalysis» Rain rate using stepwise multiple linear regressionRain rate using stepwise multiple linear regression

Classification scheme based on BTD’s (ABI only):Classification scheme based on BTD’s (ABI only):» Type 1 (water cloud): TType 1 (water cloud): T7.347.34<T<T11.211.2 and T and T8.58.5-T-T11.211.2<-0.3<-0.3

» Type 2 (ice cloud): TType 2 (ice cloud): T7.347.34<T<T11.211.2 and T and T8.58.5-T-T11.211.2>-0.3>-0.3

» Type 3 (cold-top convective cloud): TType 3 (cold-top convective cloud): T7.347.34>T>T11.211.2

September 29, 2009 5

Predictors from ABI / GLM

Target rain ratesAggregate to spatial resolution of target data

Match raining target pixels with corresponding predictors in

space and time

Match non-raining raining target pixels with corresponding

predictors in space and time

Calibrate rain detection algorithm using discriminant

analysis

Calibrate rain intensity algorithm using regression

Create nonlinear transforms of predictor values

Apply calibration coefficients to independent data

Subsequent ABI / GLM data

Matched non-raining predictors

and targets

Matched raining predictors and

targets

START

END

Ass

embl

e tr

ain

ing

data

set

Cal

ibra

te r

ain

rate

re

trie

val

Out

put

September 29, 2009 6

SCaMPR PlansSCaMPR Plans

Incorporate numerical model moisture fields to Incorporate numerical model moisture fields to handle hydrometeor evaporationhandle hydrometeor evaporation

Correct for orographic effects on rainfallCorrect for orographic effects on rainfall Incorporate convective life cycle information (i.e., Incorporate convective life cycle information (i.e.,

changes in rain rate vs. cloud properties during changes in rain rate vs. cloud properties during convective cycle)convective cycle)

Incorporate Lagrangian (cloud-following) Tb changesIncorporate Lagrangian (cloud-following) Tb changes Incorporate retrieved cloud microphysics (i.e., cloud-Incorporate retrieved cloud microphysics (i.e., cloud-

top particle size, phase)top particle size, phase)

September 29, 2009 7

OutlineOutline

SCaMPR Review and PlansSCaMPR Review and Plans GOES-RRR Project: Improved MW RetrievalsGOES-RRR Project: Improved MW Retrievals MSFC QPE WorkMSFC QPE Work Questions / Discussion (all)Questions / Discussion (all)

September 29, 2009 8

GOES-RRR Project GOES-RRR Project SummarySummary

Proposal Team: N. Wang, E. Bruning, R. Albrecht, Proposal Team: N. Wang, E. Bruning, R. Albrecht, K. Gopalan, M. Sapiano (UMD/ESSIC); R. K. Gopalan, M. Sapiano (UMD/ESSIC); R. Kuligowski, R. Ferraro (NESDIS/STAR)Kuligowski, R. Ferraro (NESDIS/STAR)

Objectives: Objectives:

(1) To improve (1) To improve microwave (MW)-based precipitation estimates by connecting the ice-phased microphysics commonly observed by GOES-R lighting and GPM MW instruments.

(2) To improve the GOES-R rain-rate (QPE) algorithm by using these estimates as input.

September 29, 2009 9

Basic Problem and Basic Problem and ApproachApproach

Problem:Problem:» Since the GOES-R rain-rate algorithm uses MW-derived rain Since the GOES-R rain-rate algorithm uses MW-derived rain

rates for calibration, errors in the MW rain-rates are reflected in rates for calibration, errors in the MW rain-rates are reflected in the GOES-R rain rates.the GOES-R rain rates.

» The MW rain-rate algorithm over land has problems of properly The MW rain-rate algorithm over land has problems of properly identifying convective/stratiform rain type, which causes errors in identifying convective/stratiform rain type, which causes errors in rain-rate retrievals.rain-rate retrievals.

Planned approach: Planned approach: » Use GOES-R GLM lighting data to constrain the MW Use GOES-R GLM lighting data to constrain the MW

convective/stratiform classification and thus convective/stratiform classification and thus reduce the misclassification of rain type.

» Use the improved MW rain-rates to produce better calibration for Use the improved MW rain-rates to produce better calibration for the GOES-R rain rate algorithm. the GOES-R rain rate algorithm.

September 29, 2009 10

0 5 10 15 20 25 30 35 40 45 500

10

20

30

40

50

60

70

80

90

PR RR (mm/hr)

TM

I R

R (

mm

/hr)

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 10

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

PR Conv. ratioT

MI

CS

I

TRMM PR v.s. TMI rain-rates TRMM PR v.s. TMI convective ratio

Errors in Current TRMM Errors in Current TRMM AlgorithmAlgorithm

Improved convective/stratiform classification should lead to Improved convective/stratiform classification should lead to more accurate MW rain rate retrievals.more accurate MW rain rate retrievals.

September 29, 2009 11

VIRS 10.8 m + LIS flashes

An example of a MCSTRMM microwave and lighting sensors as the proxy for GOES-RPotential Improvement to Potential Improvement to Current MW AlgorithmCurrent MW Algorithm

TMI 85GHz PCT+ LIS flashes

September 29, 2009 12

PR rain rates TMI rain rates

Potential Improvement to Potential Improvement to Current MW AlgorithmCurrent MW Algorithm

September 29, 2009 13

OutlineOutline

SCaMPR Review and PlansSCaMPR Review and Plans GOES-RRR Proposal: Improved MW RetrievalsGOES-RRR Proposal: Improved MW Retrievals MSFC QPE WorkMSFC QPE Work Questions / Discussion (all)Questions / Discussion (all)

Rainfall Rate Start

Aggregate ABI data to microwave footprints

Input ABI data

New microwave rain rates available?

Input microwave rain rates

Match ABI predictors with microwave footprints

Compute ABI derived parameters (e.g., temperature

differences)

Use discriminant analysis to find rain detection predictors

and coefficients

Use regression to find rain rate predictors and

coefficients

Rain detection and rate predictors and

coefficients

Yes

No

Input necessary ABI data

Determine raining areas

Compute rain rates in raining areas

Rainfall Rate End

Output

Errors In

Error magnified

Current algorithm depends strongly on presence of a “calibrating” PMW estimate-OK at O(1x1o) areas over warm oceans for emission based methods when integrated over longer time intervals-Questionable accuracy over land (much work to be done prior to/during GPM)-Questionable application to diagnosing “rate” at GOES-R pixel level.

Overarching focus: How to use Lightning when it occurs.

Lightning Input to SCaMPR Flow

Lightning Guidance?

Lightning Guidance?

1) Rainfall Detection and Convective and Stratiform (C/S) Precipitation

• C/S partitioning in most state of the art radar algorithms (problematic for satellite algorithms)

• “Low-hanging fruit”- Lightning is a “rain certain” parameter and by virtue of its origin, facilitates better C/S partitioning of rain area/volume (Grecu et al., 2000; Morales and Anagnostou, 2003)

• Focus: Presence/amount of lightning for identifying systematic differences (e.g., constraints) in cloud-system-wide C/S precipitation behavior

2) Application to Passive microwave (PMW) rainfall retrievals • Most lightning occurs over land- land focus for QPE application

• PMW rain estimates over land are “iffy”- algorithms empirically driven by assumed ice-scattering relationship to rain water content w/some inclusion of weak physics (e.g., GPROF

• Problem: As in the case of IR retrievals, even if the ice is there rainfall is often not coupled to the ice phase; not even a guarantee it will be useful in “big” rainfall events (e.g., spectacular failures of satellite QPE in shallow-convective environment floods).

• Focus: Identify specific situations where lightning occurs with systematic PMW bias or can “nudge” the answer- thus affecting error reduction in the SCaMPR calibration

What kind of guidance?

Identifying Systematic C/S behavior in TRMM Features

September 29, 2008 16

When lightning present, clear increase in convective area-fraction, feature rainrate, and convective rain volume regardless of “feature” area.

Stratiform behavior virtually identical between lightning/non-lightning cases

Convective

StratiformCon

vect

ive

Fra

ct.

Feature area

Feature area

Rai

n R

ae (

mm

/hr)

Vol

um

e R

ain.

Feature area

Ongoing Work

17

• Dual-pol rain map (done) + C/S partitioning with new GLM Proxy

• Relate C/S and flash behavior over time integrals and lifecycles, compared to TRMM “snapshots”.

• Match to GOES IR Tb areas (Morales and Anagnostou approach)

• 3-D ASCII features and pixels for the globe • Identify PMW biases sensitive to lightning

character• Extend C/S study to global domain

C/S in N. Alabama Testbed

TRMM Work

TRMM PMW: When TMI > (<) PR rain rate, deeper (shallow) convection, lightning flash count larger (smaller), 85 GHz colder (warmer)

Ground Validation Case Work (To be done in near future)

TMI< PR

TMI >PR

TMI< PR TMI >PR

JJA AfricaCFAD dBZ

JJA LIS FD (solid)85 GHz TBv (dash)

JJA PR RR (solid)TMI RR (dash)

TMI > 1.4 PR RRTMI = PRTMI < 0.6 PR RR

Lightning =

5 km

0oC

September 29, 2009 18

SCaMPR Plans: SCaMPR Plans: LightningLightning

Provided SCaMPR source code and documentation to R. Provided SCaMPR source code and documentation to R. Albrecht/Petersen for collaborative workAlbrecht/Petersen for collaborative work

Consider using lightning both as an aid to classification (i.e., Consider using lightning both as an aid to classification (i.e., convective vs. stratiform), rain rate prediction, PMW convective vs. stratiform), rain rate prediction, PMW “nudging”“nudging”

Current challenges:Current challenges:» Resolution of GLM is coarser than ABI—how to best use Resolution of GLM is coarser than ABI—how to best use

lightning data without creating an overly smooth output?lightning data without creating an overly smooth output?» Difficult to obtain total lightning over a large enough area Difficult to obtain total lightning over a large enough area

and long enough time for reliable SCaMPR calibration and long enough time for reliable SCaMPR calibration (GV?)(GV?)

» Where to insert lightning inputs?Where to insert lightning inputs? No final answers yet on several of these issuesNo final answers yet on several of these issues

FUTURE: A “STRAWMAN FRAMEWORK” FOR GOES-R GLM QPE ALGORITHM:

Petersen, 9/29/2009

SCaMPR Rain Type SCaMPR GForecast

Lightning +/- X minutes

Yes

Location, Cluster history, granularity/eccentricity, Flash extent, Flash rate [FR], D(FR)/Dt

Disaggregate (downscale)

Convective Area Fraction

Growing Mature Decaying

Stratiform Area Fraction

Growing Mature Decaying

GLM Pixel No SCaMPR

Pr(Rain volume/Rate Class |Character)** Includes assessment of possible PMW bias

Cluster w/Pixel information

September 29, 2009 20

Questions / Questions / Discussion?Discussion?