Combining GOES-R and GPM to improve GOES-R rainrate product Nai-Yu Wang, University of Maryland, ESSIC/CICS Kaushik Gopalan, ISRO, India Rachel Albrecht, CPTEC, Brazil Weixin Xu, University of Maryland, ESSIC/CICS Rober Adler, University of Maryland, ESSIC/CICS

Motivation GOES-R QPE algorithm uses microwave rain estimates as a calibration To enhance microwave-based convective and stratiform precipitation partition using lightning by connecting the ice-phased microphysics commonly observed by lightning sensors and microwave radiometers

Relating lightning to microwave convective properties TRMM October 10, 2004 over Brazil (TRMM #39346) LIS event rate (events/min) TMI 85 GHz V-pol TBs Ice, mixed -phase Microwave + Lightning Radar convective/stratiforn type Delineate Convective/Stratiform Rain Rate LIS flash rate (flashes/min)

TRMM LIS/TMI/PR Database (proxy for GOES-R and GPM) Four years (2002-2005) of TRMM radar/radiometer/lightning data in 0.1° grid resolution over land PR convective fraction estimates, near surface rain rates, reflectivity profiles TMI TBs, convective fraction estimate (from 10/37/85 GHz), rain-rates (from 85 GHz) LIS radiances, event rates, group rates and flash rates 14 millions raining pixels are used to investigate relationships between lightning frequency/occurrence and convective/statirom partition in the precipitation system observed by microwave

Interesting lightning-precipitation Statistics Four years, 14M TRMM TMI/PR/LIS over land precipitation data (in 0.1o grid resolution) reveal that 6% of rain data has lightning flash rate > 0 fl/min 13.5% of the lightning occur in stratiform, 86.5% in convective convective rain probability increases with increasing lightning frequency. For example, 34% of rainfall is convective for low flash rates (0-1 fl/min), whereas the convective probability increases to 99.7% for high flash rates (>= 2 fl/min).

Raining pixels from Jan 2002 – Dec 2004 Lightning relationships with radar reflectivity and passive microwave 85 GHz Tb Raining pixels from Jan 2002 – Dec 2004 Lightning occurrence and flash rates increase with decreasing 85 GHz TB and increasing reflectivity!

Lightning information in separating convection and straiform precipitation Lightning info is definitely very useful in discriminating radar features of convective and stratiform precipitation Colder Tb generally increases with increasing lightning rates

TRMM PR/TMI/LIS Database stratified by lighting flash rates Category Criterion Number of pixels (Percentage) Rain Type Distribution Convective/Stratiform/other (%) Category 0 (CAT0) Pixels where LIS detects no flashes ~13 million (94%) 6% / 61% / 33% Category 1 (CAT1) 0-1 flashes ~470,000 (3.4%) 34% / 21% / 45% Category 2 (CAT2) 1-2 flashes ~221,000 (1.6%) 47% / 11% / 42% Category 3 (CAT3) > 2 flashes ~89,000 (0.6%) 99.7% / 0.25% / 0%

Using Lightning flash rates in microwave convective and stratiform partitioning Lightning flash rates are used to classify convective areal fraction by pre-classifying microwave TBs into 4 groups of increasing convective probability, before rain-rate retrievals RR = RRconv P(C) +RRstrat (1 - P(C)) P(C) : convective fraction P(C) = a1 TB10V + a2 (TB37V+TB85V)/2 + a3 NPOL + a4 STDEV + a5 MINIMA +k

Effect of Lightning in the microwave convective-stratiform paritioning TMI P(C) TMI-LIS P(C) 5 PCTILE MEDIAN 95 PCTILE CAT0 no flash 0.10 0.51 0.09 0.47 CAT1 0-1 fl/min 0.11 0.41 0.71 0.29 0.54 0.75 CAT2 1-2 fl/min 0.16 0.50 0.76 0.45 0.68 0.84 CAT3 >2 fl/min 0.24 0.59 0.83 0.98 0.996 1

Improvement of Passive Microwave Retrievals (Used as Calibrator for IR Baseline Algorithm) An Example: Lightning Impact on Rain Rate Retrievals radar microwave microwave + lightning P(C) rain rate Overall impact of of lightning on rainrate is 5-10%, but focused on highest rainrates

IR Rain Estimation Issues and Motivation of incorporating Lightning Limitations of infrared-based rain estimates: -- Only “see” the top of precipitating cloud; (though cloud growth or structure can be considered) -- May treat cold cirrus clouds as intense convection; -- May misrepresent convective rain: location, area and rain intensity (especially under relatively uniform cold cloud shields in mature MCSs); But geostationary rain estimation still very important because of temporal resolution and rapid access How would lightning information help? -- Provide information associated with convection location and intensity (~ rainfall rate)

IR (and IR + Lightning) Rain Estimation Applied to TRMM Data  Initial IR technique is variation of Convective Stratiform Technique (CST, Adler and Negri, 1988)  Refine initial CST technique  Lightning information are used to define convective cores “unseen” by IR and eliminate IR cloud top minima incorrectly identified as “convective”

IR-based C/S Technique (CST) STEPS: 1.Find local minima (Tmin); 2. Slope test 3. Assign conv. area; 4. Define strat. Area (T mode) IR Lightning + Radar (Conv/Strat) CST (and most IR techniques) does a GOOD job in catching young convective cells. CST (from IR) Conv. Area by PR

IR-Lighting-Combined C/S Technique (CSTL) Conv. cores w/o lightning in mature systems are removed Conv. areas (with flash) missed by CST are added; Lightning + IR Radar (Conv/Strat) CST (from IR+L) CST (from IR) CST does a relatively POOR job for mature convective systems.

CST combined with Lighting (CSTL) 1. CST defined convective cores w/o any lightning flashes are removed; 2. Convective areas with flashes but missed by CST are added; 3. Convective rainrate is retrieved as a function of both IR Tb and lightning flash density. Rain Rate Assignment Stratiform area: 2.5 mm/hr; Conv. without lightning: as a function of Tmin; Conv. with lightning: as a function of lightning FD;

Values of Lightning to IR CST * CST and CSTL evaluated by PR on convective areas; * 2000 cases (> 20 lightning flashes) are selected; Lightning improves the convective detection (POD), lowers the false alarm (FAR); and improve CSI by 30%.

CST Issues and Use of Lightning Rainfall Rate (20 km resolution) IR Tb TMI CST CSTL

Summary and Future Work Relationships between total lightning and precipitation measurements from TRMM radar/radiometer/lightning are analyzed. A new algorithm to use lightning information to delineate microwave convective areal fraction is developed; Results reveal that lightning flash rates primarily improves the identification of deep convection and heavy convective rainfall ; Bias error reduction of 6% and RMS error increase of 4.5% (JGR paper in press) Initial work completed in developing framework for testing IR (and IR + lightning) rain estimation with TRMM data and for comparing results with GOES-R baseline algorithm Preliminary results indicate obvious value of lightning information to establish location of convective cores “unseen” by IR and eliminate incorrect core identification by IR. Preliminary statistics indicate significant improvement in rain estimation with use of lightning data Next steps include: full analysis of TRMM IR and IR+L rain estimation to carefully quantify lightning impact and its potential and limitations in comparison/ combination with Baseline algorithm use of CHUVA and other data sets to evaluate IR+L with Baseline and test with time resolution/evolution