University of Wisconsin Convective Initiation (UWCI) Developed by Justin Sieglaff, Lee Cronce, Wayne Feltz CIMSS UW-M ADISON, M ADISON, WI Kris Bedka SSAI, H AMPTON VA Mike Pavolonis and Andy Heidinger NOAA/STAR/ASPT M ADISON, WI

What does satellite data add to convective detection?

UWCI Objectives Provide a CI nowcast signal during day and night Minimize false alarm at the expense of some probability of detection Use alternative method for time trend computation (non-AMV) to minimize pixelation Provide coherent radar-like satellite-based CI signal as direct AWIPS/N-AWIPS satellite convective initiation decision support aid in field

Source of Confusion: More than one CI algorithm exists! CI from CIMSS – 24-hour coverage – Uses cloud-typing algorithm to detect phase changes – Uses multispectral GOES IR data – Not computational intensive – Being extended via object tracking CI from UAH (SatCast Ver2) – Daytime Only (improved) – Uses pixel-based mesoscale Atmospheric Motion Vectors – Uses multispectral GOES IR data – Being extended via object tracking

What is output from UWCI? Value of 0: No CI nowcast Value of 1: “Pre-CI Cloud Growth”: growing liquid water cloud Value of 2: “CI Likely”: associated with growing supercooled water or mixed phase cloud Value of 3: “CI Occurring”: Associated with a cloud that has recently transitioned to thick ice – that is, it has glaciated

What is output from UWCI? Value of 0: No CI nowcast Value of 1: “Pre-CI Cloud Growth”: growing liquid water cloud Value of 2: “CI Likely”: associated with growing supercooled water or mixed phase cloud Value of 3: “CI Occurring”: Associated with a cloud that has recently transitioned to thick ice – that is, it has glaciated GROWING MEANS THE CLOUD IS COOLING, BECOMING TALLER

What is output from UWCI? Value of 0: No CI nowcast Value of 1: “Pre-CI Cloud Growth”: growing liquid water cloud Value of 2: “CI Likely”: associated with growing supercooled water or mixed phase cloud Value of 3: “CI Occurring”: Associated with a cloud that has recently transitioned to thick ice – that is, it has glaciated GROWING MEANS THE CLOUD IS COOLING, BECOMING TALLER Once the cloud has glaciated, UWCI not needed

Characteristics of UWCI Nowcasts cloud development in regions not dominated by cirrus Uses only IR channels from GOES satellite Results available ~2 minutes after satellite scan (distribution to AWIPS takes an additional 5-10 minutes) Operates in regular and RSO mode Large spatial coverage: CONUS/Hawaii/Pacific Up to minutes of lead time before significant radar echoes/lightning strikes Low FAR, good POD, error sources understood

When should you use UWCI with caution? Algorithm does not work well where cirrus cloud cover exists Fast-moving clouds can cause false alarms Thirty minutes between images? False alarm rate increases CI is more a diagnostic tool – that is, you lose the good lead times – in true maritime Tropical airmasses Explosive development? UWCI may be more diagnostic

When should you use UWCI with caution? Algorithm does not work well where cirrus cloud cover exists Fast-moving clouds can cause false alarms Thirty minutes between images? False alarm rate increases CI is more a diagnostic tool – that is, you lose the good lead times – in true maritime Tropical airmasses Explosive development? UWCI may be more diagnostic (a function of time resolution)

UWCI: Algorithm logic – an overview in words 1.Box average 11 micron brightness temperature (BT) is calculated for current time and previous time, using specific classes from GOES Cloud Typing product 2.Unfiltered Cloud Top Cooling (CTC) Rate is calculated by differencing box average 11 micron BT for current time from previous time 3.Large/small box approach eliminates most of false CTC due to cloud motion (and additional checks reduce false cooling further) 4.CI Nowcast is assigned to remaining CTC pixels with cooling rates less than or equal to -4.0K/15 minutes by leveraging GOES Cloud Typing product

What is Box-Averaging? Newly developing convective clouds move short distances between successive images due to the high-temporal resolution of current GOES Clouds remain within a “boxed” region for 2 consecutive GOES scans. Mean IR temperatures and cloud microphysical properties of the clouds (i.e. “box-averaging”) can be differenced to identify new convective initiation This Eulerian framework means cloud motion is a significant issue: “new” non-convective clouds can enter the boxed region and influence the box-averaged properties, producing false cloud top cooling signal A set of tests is developed to preserve the signal associated with convective initiation while filtering out signal from rapidly moving and/or non-convective clouds

13 pixels What is the box-average T BB of the cloudy pixels (T1) in the small box at the current time? 7 pixels For the center (Yellow) pixel: This is done for every pixel in the image; T BB is the 11-  m brightness temperature At each pixel along the perimeter: box-average the cloudy pixels in a 7x7 box; find the minimum T BB for all points along the perimeter (T2)

13 pixels What is the box-average T BB of the cloudy pixels (T1) in the small box at the current time? 7 pixels For the center (Yellow) pixel: This is done for every pixel in the image; T BB is the 11-  m brightness temperature At each pixel along the perimeter: box-average the cloudy pixels in a 7x7 box; find the minimum T BB for all points along the perimeter (T2)

13 pixels What is the box-average T BB of the cloudy pixels (T1) in the small box at the current time? 7 pixels If (T1-T2) < 0 K, then convective cooling is occurring ; More than 1 pixel must be cloudy in the 7x7 box For the center (Yellow) pixel: This is done for every pixel in the image At each pixel along the perimeter: box-average the cloudy pixels in a 7x7 box; find the minimum T BB for all points along the perimeter (T2)

Which of these cooling points reflects cloud growth (in the vertical) and which reflects cloud motion (in the horizontal)?

Which of these cooling points reflects cloud growth (in the vertical) and which reflects cloud motion (in the horizontal)? Also: Convective Initiation should ignore cooling cirrus clouds (where the initiation happened a while ago!)

Look at the phase change, as well

All liquid & warmer than 32 F Supercooled or mixed Glaciation has occurred

Convection develops where CI predicted it along warm front

Cloud Mask finds cloudy pixels Cloud typing determines the type of cloud Surface EmissivitySurface Temperature Clear-Sky Atmospheric Absorption/Emission information 10.7  m data 13.0  m data 3.9  m data 6.7  m data

Cloud Mask finds cloudy pixels Cloud typing determines the type of cloud Clear Fog Supercooled Water Water Mixed Phase Cirrus Thick Ice Unknown Cirrus overlap

Cloud Mask finds cloudy pixels Cloud typing determines the type of cloud Clear Fog Supercooled Water Water Mixed Phase Cirrus Thick Ice Unknown Cirrus overlap

Cloud Mask finds cloudy pixels Cloud typing determines the type of cloud Clear Fog Supercooled Water Water Mixed Phase Cirrus Thick Ice Unknown Cirrus overlap CI not computed for these pixels

Cloud Mask finds cloudy pixels Cloud typing determines the type of cloud Supercooled Water Water Mixed Phase Cirrus Thick Ice Cirrus overlap

Cloud Mask finds cloudy pixels Cloud typing determines the type of cloud Supercooled Water Water Mixed Phase Cirrus Thick Ice Cirrus overlap Determine percentages of each cloud type in 7x7 and 13x13 box Determine cooling rate of cloudy pixels (normalized to a 15-minute  t) Throw out pixels that show warming relative to the values around the perimeter of the larger box Throw out pixels if >50% of pixels in large box are ice clouds

UWCI tests to reduce FAR Throw out pixels that show warming relative to the values around the perimeter of the larger box Throw out pixels if >50% of pixels in large box are ice clouds Throw out pixels if the number of cirrus/overlap pixels exceeds the number of water cloud pixels Throw out pixels if the number of overlap pixels exceeds the number of water/supercooled/mixed phase pixels. Cloud-top cooling might mean thin cirrus are simply becoming thicker. Throw out pixels if the number of cirrus/overlap pixels exceed water/supercooled/mixed phase pixels at the earlier time and the number of water/supercooled/mixed phase pixels exceeds the number of cirrus/overlap pixels at the later time. CI means the relative number of ice compared to water pixels should be increasing.

I S THERE COOLING DUE TO C LOUD M OTION ?

IS THE COOLING OUTSIDE THE MAIN UPDRAF T?

ARE THERE TOO MANY ICE PIXELS?

I S THERE MOSTLY THIN C IRRUS WHEN MANY CLOUD TYPES ARE PRESENT ?

ARE THE MICRO- PHYSICAL CHANGES ‘WRONG’ ?

ARE THERE TOO MANY ‘OVERLAP’ PIXELS – CIRRUS PIXELS OVER LOW CLOUDS?

IS THE INFERRED THUNDER- STORM UPDRAFT IS TOO WEAK?

Examples!

Dryline CI Case

UTC Instantaneous CI Nowcast Possible CI Possible CI Occurring CI Likely

UTC Instantaneous CI Nowcast Possible (All ‘warm’ liquid) CI Possible (All ‘warm’ liquid) CI Occurring (Glaciation) CI Likely (Supercooled/Mixed Phase) (N-AWIPS Screengrab)

KAMA 2103 UTC Base Reflectivity KAMA 2018 UTC Base Reflectivity CI nowcast at 2015 UTC CI likely CI occurring Non-detection due to cirrus KAMA 2035 UTC Base Reflectivity First echo >= 35 dBZ, 17 minutes after nowcast 29 April 2009

SD/NE Border Warm Frontal 0545 UTC 0546 UTC 0615 UTC 0606 UTC UTC Nocturnal Instantaneous CI Nowcast

Nebraska : Nocturnal Case IR4 image and CTC at 0245Z Radar reflectivity at 0246Z IR4 image and CTC at 0315Z Radar reflectivity at 0311Z

51 AWIPS CI/CTC Interaction with Sullivan (MKE) NWS Office 04:30 UTC 06:30 UTC 05:02 UTC Forecaster generated screen captures from AWIPS at MKE CI likely CI occurring Lightning 4:31 Z 5:01 Z 6:31 Z

Where will development occur?

Cloud top cooling example in AWIPS

Satellite-based Convective Initiation Decision Support following to Milwaukee Weather Forecast Office, Storm Prediction Center, Aviation Weather Center, and Central Region NWS

Convective Initiation and Overshooting- top demonstration at NOAA HWT 2010 The following list of products offers opportunity for near-real time Warning Related utility. Now Available – Orange, Near Future - Green Baseline Products: Volcanic Ash: detection & Height - Alaska, Pacific, and AWC Option 2 Products: Aircraft Icing Threat – NASA LaRC, Bill Smith Jr Turbulence - AWC Convective Initiation (UWCI used where SATCAST not automated yet) Enhanced “V” / Overshooting Top Detection – HWT and HPC Low Cloud and Fog – AWC and Alaska SO 2 Detection – AWC and Alaska 70 University of Wisconsin Convective Initiation Proxy Overshooting TopUWCI Cloud Top Cooling Rate 8-km Psuedo-GLM 24 May 2008

Accomplishments and Activities for Successful Research

More information on UWCI Satellite blog: (check on the ‘Convective initiation’ category) Strengths and weaknesses are discussed at ground/GOES_CINowcast.htm ground/GOES_CINowcast.htm – Ice clouds and rapidly moving clouds cause problems – 30-minute step between imagery? Detection drops

Updates Since 2010 HWT WMS serve of UWCI/OTTC – svis.50,ci,wxwarn.35,nexrphase New Ice mask added to help forecasters determine where UWCI will not work (no cloud phase transitions) Added GOES-West and Hawaii domain – oks/ Linking GOES convective cloud top cooling rate to anticipated future radar signal

Use WDSS-II software to track the evolving 11 μm top of troposphere emissivity cloud field from cumulus infancy (3 sat. pixel minimum) to convective maturity (700 pixels). Simultaneously carry along UWCI cloud top cooling rate and CI nowcasts, as well as various unique radar fields (i.e. Base; composite; and reflectivity at 0C, -10C, and -20C isotherms; VIL; MESH; POSH; etc.) for each unique object ID. State of the art framework for automated UWCI validation and computation of any lead-time value provided by UWCI CTC/CI prior to the above observed radar signatures for a given object.

"If the UWCI CTC algorithm has a developing convective cloud with a cooling rate of -15 K / 15 min, how often will that developing storm achieve a VIL of 40 g/m 2 ?.. and just as importantly, with what lead time?" GOAL: To provide additional support and warning leadtime for forecasters by addressing the following type of question: Various radar fields and thresholds to be studied are those most frequently observed with deep convection producing severe hail, wind gusts and tornadoes.

CI Conclusions Algorithm considers both cloud-top temperature changes and cloud type changes Works day and night Lead time of up 45 minutes before 35 dbZ echoes is common Several checks to minimize false alarms: When CI says convection is initiating, it usually is (POD: 55.6%; FAR: 26.0%)

CI Conclusions Algorithm considers both cloud-top temperature changes and cloud type changes Works day and night Lead time of 30 minutes before 35 dbZ echoes is common Several checks to minimize false alarms: When CI says convection is initiating, it usually is (POD: 55.6%; FAR: 26.0%) Nowcast forecasts more than half the storms that develop Forecaster can trust that 3 out of 4 forecasts verifies

CI Conclusions Algorithm considers both cloud-top temperature changes and cloud type changes Works day and night Lead time of 30 minutes before 35 dbZ echoes is common Several checks to minimize false alarms: When CI says convection is initiating, it usually is (POD: 55.6%; FAR: 26.0%) Nowcast forecasts more than half the storms that develop Forecaster can trust that 3 out of 4 forecasts verifies Stats are from UWCI nowcasts and newly-developing CG-producing storms occurring within SLGT (or greater) RSK boxes for 23 convective afternoons (1800 – 0000 UTC) in Spring/Summer 2008 and CI nowcasts: 288; LI events: 260. Includes clear sky days when POD is very high and cirrus-filled days when POD is low.