CAD Erosion: Patterns, Processes, and Prognoses

1
CAD Erosion Patterns, Processes, and Prognoses

• NWS/NCSU Collaboration
• CSTAR Regional Workshop
• October 16, 2003

2
Outline

• 1.) CAD Erosion Mechanisms
• 2.) Synoptic Settings for CAD Erosion
• 3.) CAD Erosion Forecasting Exercise
• 4.) CAD Erosion Science Case Study
• 5.) Model Performance
• 6.) Future Directions

3
Cold Air Damming Erosion

• Problem Models frequently underestimate duration
  of CAD sensible weather
• Questions
• Is this always the case?
• What patterns are most prone to model error?
• How can I tell ahead of time when the model is
  going to be wrong, and by how much?

4
CAD Erosion Objectives

• Understand physical processes leading to CAD
  erosion (not studied previously in detail)
• Elucidate different CAD erosion scenarios,
  identify synoptic settings and processes
• Pinpoint model errors during CAD erosion
• Provide guidance for correction of model output
  in CAD erosion scenarios

5
Approach to problem

• Much work needed to understand CAD erosion
• What physical processes are active?
• Do different physical processes act in different
  synoptic settings?
• How are models representing erosion?
• Pinpoint model errors
• Forecaster recognition of observational erosion
  signs
• Improves prediction, regardless of model
  performance

6
CAD Erosion Processes

• Cold dome protected by capping inversion layer
• Processes weakening inversion facilitate mixing,
  CAD demise
• Richardson useful for interpreting CAD erosion
  (Stull 1988)

• Small Ri favors mixing across inversion, CAD
  erosion

7
CAD Erosion Processes

• 1.) Cold Advection Aloft (top-down erosion)
• Cold advection above inversion weakens it (lowers
  Ri) mixing promoted
• Scenarios Cold-Frontal Passage, CFA, Coastal
  Cyclone

8
1.) Cold Advection Aloft (top-down erosion)
1200 UTC
0000 UTC
9
CAD Erosion Processes

• 2.) Surface heating (bottom-up erosion)
• Surface heating reduces Ri, promotes mixing
• Near-surface warm advection (e.g., coastal front
  advance)
• Solar radiation
• Upward soil heat flux (cold air over warm ground)
• Clear-air, warm season CAD often ends in this
  manner

10
2.) Surface heating (bottom-up erosion)
11
2.) Surface heating (bottom-up erosion)
12
CAD Erosion Processes

• 3.) Shear-induced mixing at inversion
• Strong shear/weak inversion (small Ri) favor
  mixing
• Strong (S or SW) flow above inversion large
  shear
• Process may be more of a result than a cause

10
13
CAD Erosion Processes

• 4.) Divergence near surface
• Pressure falls to north or east
• cyclone to NW of CAD region or coastal low
• Stronger PGF, cold dome air moves out
• Depth of cold dome decreases via continuity

Cross-sectional view
14
CAD Erosion Processes

• 5.) Inland Coastal Front Movement
• Boundary of cold dome (CF) moves inland
• May accompany pressure falls to N, surface
  divergence
• Most pronounced in eastern damming region

15
Synoptic Erosion Settings (Wendy Stanton)

• Goal Describe synoptic patterns which typically
  accompany CAD erosion
• Create Composites
• Representative of erosion processes
• What does the atmosphere look like at CAD demise?
• Pattern recognition for forecasters

16
Compositing Strategy

• Used 90 classical CAD cases, 1984-1995
• from Bailey et al. (2003) algorithm
• Examined patterns in MSLP and surface ?
• Locations of hi/low centers and fronts at erosion
  times
• Classified according to synoptic pattern at
  erosion time
• NW Low (23 cases)
• Cold Front Passage (14 cases)
• Coastal Low (25 cases)
• Residual Cold Pool (23 cases)

17
1.) Cold-Frontal Passage

• Cold advection above cold dome weakens inversion
• Mixing promoted as Richardson Number drops
• Model performance adequate
• provided accurate depiction of cold front
  passage
• may scour cold dome before forecast FROPA

18
2.) Residual Cold Pool

• Slow, gradual demise due to lack of synoptic
  forcing
• Reduction in cloud cover, solar heating in cold
  dome
• CAD effects dont always go away when pressure
  ridge does

19
3.) Coastal Low

• Cold advection aloft may weaken inversion
• Divergence, cold dome air advected toward coast
• Damming region may remain in cold air

20
4.) Northwestern Low

• Divergence, inland coastal front movement
• Mixing and entrainment at inversion, strong
  southerly shear
• Characterized by more significant model errors

21
Erosion Scenario Results

• Erosion can occur in distinct synoptic
  environments
• statistically significance differences
• CAD erosion setting not consistently linked to
  damming sub-type
• Synoptic patterns indicative of erosion processes
• Multiple processes may contribute to demise

22
4.) An Example CAD Erosion Forecasting

• An example of coastal low erosion scenario
• Example to illustrate challenge of CAD erosion
  forecasting
• This time, more forecast information provided
  (still not enough, I know)

23
Eta Analysis, 12 UTC 28 October 2002
Oct. 28th, TMAX/TMIN/PRE RDU 62/50/.93
Eta Forecast Sounding for RDU
24
Eta 6-h Forecast, 12 UTC 28 October 2002
Oct. 28th, TMAX/TMIN/PRE RDU 62/50/.93
Eta Forecast Sounding for RDU
25
Eta 12-h Forecast, 12 UTC 28 October 2002
Oct. 28th, TMAX/TMIN/PRE RDU 62/50/.93
Eta Forecast Sounding for RDU
26
Eta 18-h Forecast, 12 UTC 28 October 2002
Eta Forecast Sounding for RDU
27
Eta 24-h Forecast, 12 UTC 28 October 2002
Oct. 29th, TMAX/TMIN/PRE RDU 50/44/.44
Eta Forecast Sounding for RDU
28
Eta 30-h Forecast, 12 UTC 28 October 2002
Oct. 29th, TMAX/TMIN/PRE RDU 50/44/.44
Eta Forecast Sounding for RDU
29
Eta 36-h Forecast, 12 UTC 28 October 2002
Oct. 29th, TMAX/TMIN/PRE RDU 50/44/.44
Eta Forecast Sounding for RDU
30
Eta 42-h Forecast, 12 UTC 28 October 2002
Eta Forecast Sounding for RDU
31
Eta 48-h Forecast, 12 UTC 28 October 2002
Oct. 30th, TMAX/TMIN/PRE RDU ?/?/.32
Eta Forecast Sounding for RDU
32
Eta 54-h Forecast, 12 UTC 28 October 2002
Oct. 30th, TMAX/TMIN/PRE RDU ?/?/.32
Eta Forecast Sounding for RDU
33
Eta 60-h Forecast, 12 UTC 28 October 2002
Oct. 30th, TMAX/TMIN/PRE RDU ?/?/.32
Eta Forecast Sounding for RDU
34

• Now, verify your predictions for RDU
• The next slide will show verification numbers
• If your max/min total error magnitude is
• lt2 degrees Take the afternoon off and go
  tailgate
• 2-12 degrees Stick around but feel free to
  doze off at times
• gt12 degrees Alright Lackmann, you didnt give
  us enough information! Whats your point??

35

• Verification

Oct. 28th, TMAX/TMIN RDU Max 62 Min 50 Pre.
0.93
Oct. 29th, TMAX/TMIN RDU Max 50 Min 44 Pre.
0.44
Oct. 30th, TMAX/TMIN RDU Max 48 Min 42 Pre.
0.32
Wet CAD Bite worse than SLP bark! Must look
beyond the wedge (in SLP) to assess CADs
sensible weather impact Where did model go
wrong?? Case study investigation (Stanton)
36
RDU MOS Comparisons for Coastal Low CAD Erosion
Case
12Z (AVN?) MOS from
37
Case 2 Coastal Low CAD Erosion
Wed, 30 October 2002, 12Z obs, Eta SLP
38
Case 2 Coastal Low CAD Erosion
Thu, 31 October 2002, 00Z obs, Eta SLP
39
Visible Satellite and Surface Observations
18 UTC 30 Oct 2002
40
Coastal Low CAD Erosion 10/30/02

• 10/30/02 Eta 2-m temperature forecast with
  surface observations, 18 UTC 30 October, 2002

41
2-m Temp, Eta Analysis, 00 UTC 31 Oct
2-m Temp, Eta 36-h Fcst, valid at 00 UTC 31 Oct
Eta Bias, Forecast Analysis
42
Motivation
Eta Model 36-h Fcst Vs. Observations
GSO Obs Solid Model Dashed, Bold Winds
00Z 31 Oct 2002
43
Analysis of October Event

• Case Study
• Performed observational analyses
• Investigated errors in Eta forecasts
• Hypothesis 1 Cloud-radiation interaction source
  of error
• Hypothesis 2 Model PBL processes related to
  early erosion
• Evaluation of model ability to predict thermal
  advection, coastal low
• Conducted MM5 simulations
• Model PBL scheme and cloud albedo experiments
• Goal pinpoint source of model error

44
12 UTC 29 Oct 2002
45
00 UTC 30 Oct 2002
46
12 UTC 30 Oct 2002
47
00 UTC 31 Oct 2002
48
12 UTC 31 Oct 2002
49
Time Series of Inversion Strength and
Differential Thermal Advection (GSO)
Red Columns 850-mb Thermal Advection Blue
Columns 1000-mb Thermal Advection
50
Time Series of Inversion Strength and Solar
Radiation (GSO, High Point)
51
1000/900mb Divergence (10-5 s-1) Red Lines
SLP (mb) Blue Lines
Eta Analysis, 00 UTC 31 Oct 2002
52
MM5 Control Run

• Provides a higher-resolution dataset
• Further examine the role of erosion mechanisms
• Facilitates examination of model errors
• Designed to resemble operational models

53
MM5 Configuration

• Non-hydrostatic
• Sigma coordinate
• 36-km hor. grid spacing
• 45 vert. sigma levels
• Initialized EDAS analyses
• Interpolated to 40-km
• 28 vert. pressure levels
• Boundary conditions EDAS analyses
• updated every 6-h

• Betts-Miller Convective
• MRF-PBL Scheme
• OSU-LSM
• Dudhia Simple Ice (Dudhia, 1989)
• Cloud-radiation scheme (Dudhia, 1989)

54
MM5 Domain
55
Time Series of Inversion Strength and
Differential Thermal Advection (GSO)
Red Columns 850-mb Thermal Advection Blue
Columns 1000-mb Thermal Advection
Control run over-weakens
56
Time Series of Inversion Strength and Solar
Radiation (GSO)
Over-estimation of CA aloft, radiation
contributing to over-dramatic weakening?
57
MM5 Sensitivity Tests

• Cloud albedo
• Betts et al.,1997 Yucel et al.,1998 NCEP
  Communication
• Hypothesis
• Excessive solar radiation through clouds can lead
  to premature erosion
• Increasing (decreasing) cloud albedo decreases
  (increases) sfc temps
• Edited albedo array in swrad.f source code
• Exp. 1 raised albedo to near 99
• Exp. 2 lowered albedo by 30

58
MM5 Control Run Red Low Albedo Run Green High
Albedo Run Blue
All runs eroded CAD too early
30-h simulation, 18 UTC 30 Oct 2002
59
Eta Model Orange
No saturated layer
30-h forecast, 18 UTC 30 Oct 2002
60
MM5 Sensitivity Tests

• PBL scheme
• Hypothesis
• surface temp forecast sensitive to PBL scheme
• Control MRF-PBL
• Non-local formulation
• Overactive mixing (Vogelezang and Holtslag,1996
  Braun and Tao,2000)
• Blackadar (no OSU-LSM)
• Stable regime local
• Unstable non-local

61
Control Run (no LSM) Red Blackadar Run (no
LSM) Green
30-h simulation, 18 UTC 30 Oct 2002
62
Eta Model Orange
30-h forecast, 18 UTC 30 Oct 2002
63
(No Transcript)
64
Oct 2002 Case Study Results

• Observations/Analyses
• Two erosion mechanisms, on 31st
• Cold advection aloft
• Solar radiation
• Cold advection aloft dominant?
• Most noticeable changes in vertical profile seen
  aloft
• Top-down erosion
• Erosion in model atmosphere was bottom-up (solar)

65
Oct 2002 Case Study Results

• MM5 Simulations
• Model weakened inversion too dramatically on 30th
  
• Inversion weakening correlated with
• Overestimated shortwave radiation
• Overestimated cold advection aloft
• Cloud albedo, PBL scheme sensitivity
• Raising albedo, using Blackadar an improvement
• All simulations still eroded the cold dome
  prematurely
• MM5 more accurate than Eta

66
Implications

• Synoptic erosion scenarios
• Pattern recognition
• Infer erosion process influence
• Model performance related to erosion process
• Which scenarios are most problematic?
• NW low worst?
• Coastal low, residual cold pool when CAD dome
  saturated
• Cold frontal passage well handled, but model may
  exhibit NW low
• Which physical processes are most problematic?
• Cloud/radiation interactions with shallow stratus
• Cases with intense, shallow inversion
• Differential thermal advection should be
  relatively well handled

67
Modeling

• Eta model previously let sun shine through
  shallow clouds
• Fixed on 8 July 2003
• New radiation scheme due in fall Eta bundle
• Case study results (with MM5) suggest this isnt
  whole problem
• MM5 and Workstation Eta
• Could configure MM5 with Blackadar scheme
• Do CAD configuration runs
• Turn down Eta diffusion, run with K-F CP scheme

68
Future Work

• Development of forecaster tools based on results
• Diagnose exact location of inversion
• Calculate Ri directly AWIPS?
• Examine cases representative of other erosion
  scenarios
• Explore model cloud/radiation interaction
  sensitivity

69
Thank You!

• Wendy Stanton (NWS MRY)
• NOAA CSTAR Program
• NWS RAH and others
• Esp. Gail Hartfield, Scott Sharp, Doug Schneider,
  Jonathan Blaes, Kermit Keeter
• Mike Brennan
• Unidata Program NSF/NCAR/PSU (MM5)
• NCEP, Brad Ferrier
• NWS CSTAR offices (esp. Kermit Keeter, Scott
  Sharp, Gail Hartfield, Neil Stuart, Larry Lee)
  and many others
• Jeff Waldstreicher (Profiler and RASS images,
  encouragement)

70
(No Transcript)
71
Control Run Red Control Run (no LSM) Green
18 UTC 30 Oct 2002