Getting the Convective Boundary Layer Right: How it impacts Forecasts of Deep Convection

1
Getting the Convective Boundary Layer Right How
it impacts Forecasts of Deep Convection
Jack Kain ASP Summer Colloquium July 2006
2
Getting the Convective Boundary Layer Right in
models How it impacts Human Forecasts of Deep
Convection
Jack Kain ASP Summer Colloquium July 2006
3
My Background

• I am a numerical modeler by training, but I have
  spent the last several years working at the
  interface of research and daily operations with
  forecasters from the NCEP/Storm Prediction Center
  (SPC) and researchers from the NOAA/National
  Severe Storms Laboratory (NSSL)

4
Why the Boundary layer?
5
Why the Boundary layer?
CAPE
6
Small changes in Boundary Layer Temperature and
Moisture can have a Big Impact on CAPE and
CINAlso, the Temperature Profile just above the
top of the BL strongly affects CIN
But do forecasters actually care about
small-scale details in forecast-sounding
structure? Yes! Consider a sampling of DAY 2
OUTLOOKs, a product that SPC forecasters use to
provide guidance to local forecast offices
7
DAY 2 CONVECTIVE OUTLOOK NWS STORM
PREDICTION CENTER NORMAN OK 1111 AM CST SAT
APR 01 2006 VALID 021200Z - 031200Z
...THERE IS A MDT RISK OF SVR TSTMS OVER PARTS OF
IA...MO AND MUCH OF IL... FORECAST SOUNDINGS
INDICATE A POTENT COMBINATION OF INSTABILITY AND
VERTICAL SHEAR ACROSS MODERATE RISK AREA...WHICH
SHOULD SUPPORT INTENSE SUPERCELL DEVELOPMENT WITH
THE POTENTIAL FOR A FEW STRONG TORNADOES IN
ADDITION TO VERY LARGE...DESTRUCTIVE HAIL AND
DAMAGING WINDS. WITH TIME...STORMS MAY EVOLVE
INTO BOWING LINE SEGMENTS OR A MESOSCALE LEWP
STRUCTURE WITH CORRIDORS OF DAMAGING WINDS
POSSIBLE EWD ACROSS THE OH VALLEY SUNDAY NIGHT
FORECAST SOUNDINGS INDICATE A POTENT COMBINATON
OF INSTABILITY AND VERTICAL SHEAR ACROSS MODERATE
RISK
8
DAY 2 CONVECTIVE OUTLOOK NWS STORM PREDICTION
CENTER NORMAN OK 1239 PM CDT MON APR 03 2006
VALID 041200Z - 051200Z ...NO SVR
TSTM AREAS FORECAST... ...EARLY MORNING --
TX/SERN NM... MRGL/CONDITIONAL HAIL
PROBABILITIES EXIST OVER THIS REGION FOR FIRST
FEW HOURS OF PERIOD...AMIDST EXTENSION OF LATE
DAY-1 WAA/MOISTURE RETURN REGIME. ELEVATED
BUOYANCY IS FCST TO BE SMALL GENERALLY AOB 500
J/KG MOST AREAS. HOWEVER...ANY SUSTAINED
CONVECTION WILL HAVE POTENTIAL FOR HAIL GIVEN
STRONG VEERING AND SPEED SHEAR WITH HEIGHT
THROUGH CONVECTIVE COLUMN...WHICH SHOULD BE
ROOTED IN 800-850 MB LAYER BASED ON FCST
SOUNDINGS. ACTIVITY SHOULD DIMINISH DURING LATE
MORNING AS LLJ WEAKENS
ELEVATED BUOYANCY
CONVECTION
SHOULD BE ROOTED IN
800-850 MB LAYER BASED ON FCST SOUNDINGS.
9
DAY 2 CONVECTIVE OUTLOOK NWS STORM PREDICTION
CENTER NORMAN OK 1237 PM CDT TUE APR 04 2006
VALID 051200Z - 061200Z ...THERE IS A
SLGT RISK OF SVR TSTMS PORTIONS NRN PLAINS...
...NRN/CENTRAL PLAINS... WIDELY SCATTERED
TSTMS SHOULD DEVELOP FROM MID/LATE AFTERNOON INTO
EARLY EVENING ACROSS NRN HIGH PLAINS...FROM SRN
MT POTENTIALLY AS FAR E AS INVOF BLACK HILLS.
ACTIVITY INITIALLY SHOULD BE HIGH-BASED...SUPPORTE
D BY MRGL BOUNDARY LAYER MOISTURE...STEEPENING
MIDLEVEL LAPSE RATES ASSOCIATED WITH APCHG
TROUGH...STRENGTHENING DEEP-LAYER SHEAR...AND
DIABATIC SFC HEATING. DEEPLY MIXED SUBCLOUD
LAYERS IN FCST SOUNDINGS SUGGEST POTENTIAL FOR
BOTH DAMAGING DOWNDRAFTS AND LARGE HAIL TO REACH
SFC.
DEEPLY MIXED SUBCLOUD LAYERS IN FCST
SOUNDINGS SUGGEST POTENTIAL FOR BOTH DAMAGING
DOWNDRAFTS AND LARGE HAIL TO REACH SFC
10
DAY 2 CONVECTIVE OUTLOOK NWS STORM PREDICTION
CENTER NORMAN OK 1235 PM CDT THU APR 06 2006
VALID 071200Z - 081200Z ...THERE IS A HIGH
RISK OF SVR TSTMS ACROSS NE MS...NRN AL...SCNTRL
TN AND FAR NW GA... INITIATE FRIDAY
AFTERNOON...SUPERCELLS WILL BE LIKELY DUE TO THE
STRONG VERTICAL SHEAR ASSOCIATED WITH THE
MID-LEVEL JET. NAMKF FORECAST SOUNDINGS FOR
TUPELO MS...FLORENCE ALABAMA AND HUNTSVILLE
ALABAMA FRIDAY AFTERNOON SEEM REASONABLE WITH
MODERATE TO STRONG INSTABILITY (MLCAPE VALUES
2500-3000 J/KG)...0-6 KM SHEAR VALUES AT/ABV 60
KT AND IMPRESSIVE LOW-LEVEL SHEAR PROFILES (0-3
KM SRH AT/ABV 350 M2/S2). IN THIS
ENVIRONMENT...STORMS SHOULD REMAIN DISCRETE FOR
THE LATE AFTERNOON AND EARLY EVENING HOURS. IN
ADDITION...LCL HEIGHTS ARE FORECAST TO REMAIN LOW
WHICH COMBINED
NAMKF FORECAST SOUNDINGS
SEEM REASONABLE
WITH MODERATE TO STRONG INSTABILITY
11
DAY 2 CONVECTIVE OUTLOOK NWS STORM PREDICTION
CENTER NORMAN OK 1235 PM CDT THU APR 06 2006
VALID 071200Z - 081200Z ...THERE IS A HIGH
RISK OF SVR TSTMS ACROSS NE MS...NRN AL...SCNTRL
TN AND FAR NW GA... FAVORABLE FOR TORNADOES.
LONG-TRACK STRONG OR EVEN VIOLENT TORNADOES WILL
BE POSSIBLE ESPECIALLY WITH SUPERCELLS THAT TRACK
ALONG ANY PRE-EXISTING BOUNDARIES OR CONVERGENCE
ZONES. THE INSTABILITY...SHEAR AND STEEP
MID-LEVEL LAPSE RATES NEAR 8.0 C/KM SHOWN ON THE
FORECAST SOUNDINGS WILL ALSO BE FAVORABLE FOR
LARGE HAIL AND WIND DAMAGE WITH THE SUPERCELLS.
VERY LARGE HAIL WILL ALSO BE POSSIBLE WITH THE
STRONGER SUPERCELLS
SHEAR AND STEEP
MID-LEVEL LAPSE RATES NEAR 8.0 C/KM SHOWN ON THE
FORECAST
SOUNDINGS WILL ALSO BE FAVORABLE FOR LARGE HAIL
AND WIND
12
DAY 2 CONVECTIVE OUTLOOK NWS STORM PREDICTION
CENTER NORMAN OK 1235 PM CDT THU APR 06 2006
VALID 071200Z - 081200Z ...THERE IS A HIGH
RISK OF SVR TSTMS ACROSS NE MS...NRN AL...SCNTRL
TN AND FAR NW GA... INSTABILITY IN THE
AFTERNOON HOURS. THE NAM AND GFS INITIATE
POST-FRONTAL CONVECTION EARLY IN THE AFTERNOON
NORTH OF A FRONTAL BOUNDARY EXTENDING WSWD TO
ENEWD ACROSS THE OH VALLEY. FORECAST SOUNDINGS
FOR THIS AREA FRIDAY AFTERNOON SHOW SUFFICIENT
INSTABILITY/SHEAR PROFILES AND STEEP LAPSE RATES
THAT SHOULD BE FAVORABLE FOR SUPERCELLS WITH
LARGE HAIL. FURTHER SOUTH ACROSS KY...SFC
DEWPOINTS IN THE UPPER 50S TO NEAR 60 F SHOULD
ALLOW MODERATE INSTABILITY TO DEVELOP. ALTHOUGH
CONVECTIVE INITIATION MAY...
FORECAST SOUNDINGS FOR THIS AREA FRIDAY AFTERNOON
SHOW SUFFICIENT INSTABILITY/SHEAR PROFILES AND
STEEP LAPSE
RATES THAT SHOULD BE FAVORABLE FOR SUPERCELLS
13
What components of a modeling system most
strongly affect (pre-convective) CAPE/CIN and BL
structures?

• Parameterizations of
• Boundary Layer/Turbulence
• Shallow moist convection (do we need this with
• 1-4 km grid spacing?)
• Surface processes
• Atmospheric radiation
• Advection

14
Consider two very different shallow Convective
Parameterizations (both used in the NCEP SREF)

• Kain-Fritsch scheme a mass-flux convective
  adjustment scheme
• Betts-Miller-Janjic (BMJ) scheme a static
  profile convective adjustment scheme

15
KF scheme Checking for the possibility of deep
(precipitating) convectioncan a parcel reach its
LFC and form a deep cloud?
16
KF scheme Can a parcel reach its LFC and form a
deep (precipitating) cloud?
17
KF scheme Can a parcel reach its LFC and form a
deep (precipitating) cloud?
18
KF scheme Can a parcel reach its LFC and form a
deep (precipitating) cloud?
19
KF scheme Can a parcel reach its LFC and form a
deep (precipitating) cloud?
20
KF scheme determining shallow-convection
adjustment profiles
Cloud-base mass flux At what rate is mass is
processed by moist shallow convection? Currently
scaled as a function of TKE in the BL
21
KF shallow convection parameterizationrequires
many assumptions

• We think we have a good handle on
• Cloud-base thermodynamic properties
• Cloud base, cloud top
• Adjustment time scale

• We have less confidence in assumptions related
  to
• Cloud-base mass flux how much of the BL mass
  participates in moist convection?
• Vertical profiles of entrainment, detrainment and
  vertical mass flux, which determine profiles of
  heating and drying.

22
Its cloud illusions that I recallI really
dont know clouds at all
- Joni Mitchell
23
KF scheme adjustment profiles
24
The Betts-Miller-Janjic (BMJ) parameterization
based on the philosophy of Alan Betts (Alan knows
clouds)

• The fundamental characteristics of updrafts and
  downdrafts depend on internal cloud dynamics and
  microphysics, which cannot be represented
  adequately by simple sub-models used in mass-flux
  parameterizations
• Observations and theoretical considerations tell
  us that convecting atmospheres never stray too
  far from definable profiles.
• Convective sources of heat and moisture can be
  defined in terms of these profiles.

25
Like the KF scheme, the BMJ scheme first checks
for the possibility of deep (precipitating)
convection

• As with all convective parameterizations, CAPE is
  a necessary condition for activation of deep
  convection, but the BMJ scheme reverts to shallow
  (non-precipitating) convection if
• The cloud (unstable) layer is too dry
• The cloud layer is too shallow

26
SPC AC 011711 DAY 2 CONVECTIVE OUTLOOK
NWS STORM PREDICTION CENTER NORMAN OK 1111
AM CST SAT APR 01 2006 VALID 021200Z -
031200Z ...THERE IS A MDT RISK OF SVR
TSTMS OVER PARTS OF IA...MO AND MUCH OF IL...
...THERE IS A SLGT RISK OF SVR TSTMS FROM
THE GREAT LAKES SWD INTO THE MID SOUTH AND
ARKLATEX... --POTENTIALLY SIGNIFICANT
SEVERE WEATHER EPISODE IS BEING FORECAST
ACROSS THE MIDWEST AND MID MS VALLEY SUNDAY AND
SUNDAY NIGHT-- ...SYNOPSIS... MID
AND UPPER-LEVEL TROUGH NOW OVER THE PACIFIC NW
AND GREAT BASIN WILL TRANSLATE EWD INTO THE
GREAT PLAINS LATE SATURDAY NIGHT INTO SUNDAY
MORNING...PRIOR TO INTENSIFYING INTO A CLOSED
CIRCULATION OVER THE UPPER MIDWEST SUNDAY
NIGHT. AS THIS OCCURS...ACCOMPANYING MID AND
HIGH-LEVEL JET STREAKS WILL PROPAGATE THROUGH THE
TROUGH BASE INTO MO AND IL/IND SUNDAY
AFTERNOON AND EVENING. IN THE
LOW-LEVELS...PRIMARY SURFACE LOW INITIALLY OVER
ERN NEB WILL DEVELOP ENEWD INTO LOWER MI BY
MONDAY MORNING. ASSOCIATED WARM FRONT WILL
CONCURRENTLY LIFT NEWD THROUGH THE OH VALLEY AND
GREAT LAKES WHILE TRAILING COLD FRONT SURGES
EWD THROUGH THE MID/LOWER MO AND MS VALLEYS
AND INTO THE OH AND TN VALLEYS THROUGH THE
FORECAST PERIOD. ...MID/LOWER MO VALLEY
INTO THE GREAT LAKES...MID MS AND OH
VALLEYS... A NW-SE ORIENTED BAND OF MAINLY
ELEVATED TSTMS IS EXPECTED TO BE ONGOING
02/12Z FROM VICINITY OF THE SURFACE LOW OVER ERN
NEB/WRN IA INTO WRN KY AND MIDDLE TN...LARGELY
DRIVEN BY WAA ALONG BROAD SWLY LLJ. THIS
ACTIVITY WILL LIKELY CONTINUE THROUGH THE PERIOD
AHEAD OF RETREATING WARM FRONT INTO THE GREAT
LAKES WITH THE STRONGEST STORMS CAPABLE OF
MAINLY LARGE HAIL. IN THE WAKE OF THIS
INITIAL ACTIVITY...THE PRESENCE OF STEEP
MID-LEVEL LAPSE RATES / 7-8 C/KM / COUPLED WITH
BOUNDARY-LAYER DEWPOINTS IN THE LOW TO MID 60S
WILL CONTRIBUTE TO A MODERATELY UNSTABLE AIR
MASS ACROSS SYSTEM WARM SECTOR WITH MLCAPES
INCREASING TO 1500-2500 J/KG FROM SERN
SE/CNTRL MO SWWD INTO THE ARKLATEX AND LOWER
MS VALLEY. INCREASED FORCING FOR ASCENT
AHEAD OF MID-LEVEL TROUGH AND EXIT REGIONS OF
ACCOMPANYING JET STREAKS WILL CONTRIBUTE TO
RAPID SURFACE-BASED TSTM DEVELOPMENT BY EARLY
TO MID AFTERNOON FROM NEAR SURFACE LOW OVER
WRN OR CNTRL IA SWD ALONG COLD FRONT INTO MO.
ADDITIONAL SURFACE-BASED STORMS WILL ALSO BE
POSSIBLE AT THIS TIME ALONG WARM FRONT.
FORECAST SOUNDINGS INDICATE A POTENT COMBINATION
OF INSTABILITY AND VERTICAL SHEAR ACROSS
MODERATE RISK AREA...WHICH SHOULD SUPPORT
INTENSE SUPERCELL DEVELOPMENT WITH THE POTENTIAL
FOR A FEW STRONG TORNADOES IN ADDITION TO VERY
LARGE...DESTRUCTIVE HAIL AND DAMAGING WINDS.
WITH TIME...STORMS MAY EVOLVE INTO BOWING LINE
SEGMENTS OR A MESOSCALE LEWP STRUCTURE WITH
CORRIDORS OF DAMAGING WINDS POSSIBLE EWD
ACROSS THE OH VALLEY SUNDAY NIGHT. ...MID
SOUTH INTO ARKLATEX... AS MENTIONED
ABOVE...PRE-FRONTAL AIR MASS IS FORECAST TO
BECOME MODERATELY UNSTABLE BY AFTERNOON OWING
TO STEEPENING LAPSE RATES AND A MOIST
BOUNDARY-LAYER. GIVEN A STRONGER CAP AND THAT
LARGE-SCALE FORCING FOR ASCENT AND LOW-LEVEL
FRONTAL CONVERGENCE WILL TEND TO WEAKEN WITH
SWD EXTENT...SOME UNCERTAINTY EXISTS WITH REGARD
TO EVENTUAL CONVECTIVE COVERAGE AND INTENSITY
ACROSS REGION. LATEST SHORT TERM MODEL
GUIDANCE SUGGESTS THAT TSTMS SHOULD BEGIN TO
BUILD SWWD ALONG FRONT BY MID TO LATE
AFTERNOON FROM NEAR THE CONFLUENCE OF THE OH
AND MS RIVERS INTO PORTIONS OF ERN AR...WRN TN
AND PERHAPS SWD INTO NRN LA AND NRN MS.
THE COMBINATION OF MODERATE INSTABILITY AND 45-55
KT WLY MID-LEVEL FLOW SUGGEST THE THREAT FOR A
FEW SUPERCELLS AND BOWING LINE SEGMENTS CAPABLE
OF DAMAGING WINDS...LARGE HAIL AND A FEW
TORNADOES. THIS SEVERE THREAT WILL LIKELY
CONTINUE SUNDAY NIGHT ACROSS THE TN VALLEY AND
NRN PORTIONS OF MS AND AL. ..MEAD..
04/01/2006 CLICK TO GET WUUS02 PTSDY2
PRODUCT NOTE THE NEXT DAY 2 OUTLOOK IS
SCHEDULED BY 0800Z
Conceptual basis for BMJ shallow-convection
parameterization
- from Betts (1986 QJRMS)
27
BMJ Scheme determining shallow-convection
adjustment profiles
3) Anchor Tadj to the sounding at cloud base,
draw it parallel to mixing line
4) Slide Tadj so that net heating is zero
28
Consider the impact of parameterized BMJ shallow
convection in a normal diurnal cycle
Model Initial Condition Raob BMX 12 Z 11 May 2000
29
BMJ adjustment process (Eta model forecast)
Initial time
30
BMJ adjustment process (Eta model forecast)
3 h forecast
31
Compare Eta model forecast with raob at 00Z 12 h
forecast
Model forecast Raob BMX 00Z 12 May 2000
32
How about same 12h forecast using KF shallow
convection instead of BMJ (EtaKF)?
33
24 h Eta/EtaKF soundings vs. DFW raob
00Z RAOB Opnl Eta

00Z RAOB Etakf

34
Shallow convection mean moisture tendencies 21h
fcst
Opnl Eta
EtaKF
35
Performance characteristics like this that are
common knowledge to model developers are often a
mystery (but potentially very useful) to
forecasters
36
Interactions with forecasters convinced us that
knowledge of these behavior characteristics could
help forecasters interpret model forecast
soundings, motivating us to.

• Document characteristic biases during 2001
  Spring Program

• Publish guidance for forecasters (Baldwin et
  al. 2002, Wea. Forecasting)

- Provide formal SPC/WFO training on model
forecast sounding interpretation
- Introduce diagnostic versions of BMJ and KF
schemes into the SPCs NSHARP sounding analysis
program
37
48 h Eta CAPE forecast
Consider a real forecasting dilemma from several
years ago
Wind barbs 0-6 km shear vector
38
(No Transcript)
39
(No Transcript)
40
SPC forecasters identified the characteristic
bias of the Eta model, boldly suggested the 48 h
forecast sounding was in error
SPC day 2 convective outlook valid 28 Mar 2002
41
48 h later SHV observed sounding
42
Preliminary Summary PBL structures, CAPE, CIN
values, etc. are all quite sensitive to
parameterization of shallow moist convection

• How about the sensitivities to PBL/turbulence
  parameterizations? Compare CBL soundings from
  dx4km forecasts from
• WRF-ARW using the YSU (Yonsei University) PBL
• WRF-NMM using the MYJ (Mellor-Yamada-Janjic) PBL
• without parameterization of moist convection

43
Possible Approaches for Parameterization of the
Convective Boundary Layer
- From Stull (1988)
44
Sounding comparison 24h forecast valid 00Z 24
May at DDC
ARW with YSU PBL
Good forecast
Model
Raob
NMM with MYJ PBL
PBL too shallow, cold, moist clouds just broke
up!
Model
Raob
45
ARW with YSU PBL
Model
Raob
NMM with MYJ PBL
Raob
Model
46
ARW with YSU PBL
Raob
Model
NMM with MYJ PBL
Raob
Model
47
Quantifying errors in model-forecast
soundingssimple comparison of T, qv values at a
given pressure level is inadequate!
Is the forecast sounding any good? What aspects
of the sounding are predicted well?
48
How does one quantify these errors?
That one was way off!
49
How does one quantify these errors?
50
How does one quantify these errors?
That one was BS! (Bad Sounding)
S. Weiss
51
A simple approach to quantifying important
structure errors in convective-BL soundings

• Use criteria based on static stability to define
  top of CBL
• Compute depth of CBL (top of constant q layer)
• Compute mean q in CBL
• Compute mean qv in CBL

How about the CIN layer?
52
Sounding verification How well is the CIN layer
defined?
100 mb
- Compare this area in forecasts and observations
53
Sounding verification How well is the moisture
profile at the top of the BL defined?
100 mb
- Compare this area in forecasts and observations
54
Qualitative assessment of systematic biases in
Convective Boundary Layers
55
Summary

• Getting the boundary layer right is critically
  important for predicting moist convection.
• Currently available parameterizations of the BL
  and of shallow convection are deficient in many
  (hard to define) ways.
• We are counting on the next generation of
  meteorologists (thats you!) to solve these
  problems