flurries or feet? an analysis of qpf errors in an eastern north carolina snowstorm december 2-4,...
TRANSCRIPT
FLURRIES OR FEET?
An Analysis of QPF Errors in an Eastern North Carolina Snowstorm
December 2-4, 2000
Jason CaldwellSouth Carolina State Climate Office
WINTER WEATHER FORECASTING
CSTAR ProjectCSTAR Project
National Weather Service Offices (NC, SC, VA)
Focus on improving forecast skill in CAD, CF
Evaluation of the MM5 Real-Time Forecast Evaluation of the MM5 Real-Time Forecast SystemSystem
NCSU, NC SCO, NCSC, Capitol Broadcasting
Focus on sources of model error on the synoptic and mesoscale related to CAD, CF
Examine model performance relative to Eta
EDAS Surface Analysis
EDAS 500mb Analysis
EDAS 850mb Analysis
EDAS 250mb Analysis
SYNOPTIC SET-UP
VALID 12Z 2 DECEMBER 2000
Insert MM5 24h Forecast here
MM5 vs. ETA Quantitative Precipitation Forecasts(24-30 hour forecasts valid 12-18 Z on 3 December 2000)
TOTAL ACCUMULATED SNOWFALL2-4 DECEMBER 2000
(graphics courtesy NWS Raleigh)
What we know:What we know:
• Model forecasts missed a precursor vorticity maximum at 500 mb
• Models predicted coastal front too close to coast (site for surface cyclone formation)
• Models under-predicted intensity of cold-air damming
• Models produced heavy precipitation with a thermodynamic profile that was characterized by shallow cloud
• Operational sea-surface temperature analyses were too warm according to NCEP, could have been tied to coastal front errors
Lead vorticity disturbance at 500 mb(under-forecast by models) pushed the coastal front seaward, strengthened CAD, dried atmosphere over
central NC before main vorticity maximum arrived through evaporative processes and subsidence
HYPOTHESIS A
•EVALUATE MODEL ERRORS IN 500mb VORTICITY
•COMPARE VERTICAL CHARACTERISTICS BEFORE & AFTER VORT PASSAGE
•TEST FOR EVIDENCE OF SURFACE BASED RESPONSE
•HOW DOES THIS AFFECT QPF?
ETA 500mb Height/Vorticity 12h Forecast
MM5 500mb Height/Vorticity 12h Forecast
valid 00Z 3 December 2000
valid 00Z 3 December 2000
EDAS 500mb Height/Vorticity Analysis
valid 00Z 3 December 2000
VERTICAL CROSS-SECTION DIAGRAM
Yellow section indicates Central NC region; Black dot indicates location at max QPF gradient
Drying aloft due to Subsidence
Low-level drying as CAD strengthens
CAA increasing w/time
CAA increasing w/time
UVV as response to DiffThermAdv
Temperatures rapidly decreasing
w/height
EDAS Vertical Analysis (Omega,RH)
EDAS Vertical Analysis (Omega,RH)
EDAS Vertical Analysis (TAdv,Theta)
EDAS Vertical Analysis (TAdv,Theta)
valid 00Z 3 Dec 2000 valid 00Z 3 Dec 2000
valid 06Z 3 Dec 2000 valid 06Z 3 Dec 2000
valid 00Z 3 Dec 2000 valid 12Z 3 Dec 2000
valid 00Z 3 Dec 2000 valid 06Z 3 Dec 2000
ED
AS
FR
ON
TO
GE
NE
SIS
AN
D S
LP
AN
AL
YS
ES
(PR
E-
& P
OS
T-
500m
b V
OR
T M
AX
)
PRE-VORT POST-VORT
The strength and southern extent of the cold air damming was under-predicted by model forecasts and led to errors in QPF.
HOW DOES THIS AFFECT THE QPF?
The decreased magnitude of CAD shifts the location of the coastal front and affects low-level thermal and moisture profiles inland
- thermal gradient (baroclinic zone) is positioned farther west due to restricted areal coverage of cold dome
- moisture availability is higher due to less airmass intrusion from parent high
- lower atmosphere is less stable and more conducive to vertical motion
- coastal front more likely to propagate inland as WAA overtakes weaker CAD
Low pressure forms closer to the coast which extends the western boundary of precipitation into the Triad region
- enhanced upward vertical motion inland due to surface convergence at CF
- increased moisture transport with stronger and more easterly winds at low-levels
- prolonged duration of precipitation as a result of proximity
- prolific isentropic lift as warm air advection associated with cyclogenesis moves into NC over CAD at mid-levels
HOW DOES THIS AFFECT THE QPF?
Model forecasts predicted the placement of the coastal front west of the actual location off the coast of North Carolina and resulted in QPF errors
Model microphysics were overactive in producing precipitation with dry air aloft and a shallow cloud
HYPOTHESIS B
•EVALUATE PRECIPITATION MECHANISMS PRESENT
•DIAGNOSE VERTICAL TEMPERATURE/MOISTURE PROFILE OVER CENTRAL NC
•HOW DOES THIS AFFECT QPF?
INGREDIENTS FOR WINTER PRECIPITATION FORECASTS
AVAILABLE MOISTURE
*Skew-T profiles*Satellite Imagery
INSTABILITY
Temperatures decreasing w/height
Thermal and Moisture Advection
FORCING FOR ASCENT
Low/mid- level convergence
Upper-level divergence/diffluence
Differential Temperature Advection
EFFICIENCYBased on cloud Temperature and ice generation
*Max occurs at –15C w/strong forcing for ascent
TEMPERATUREDetermines p-type and snow/water ratio
MUCH DRIER THAN FORECAST
MUCH DRIER THAN FORECAST
Cross-section of Omega/RH EDAS Analysis
valid 12Z 3 December 2000
Cross-section of Omega/RH ETA 24h Forecast
valid 12Z 3 December 2000
24-hour MM5 and ETA Forecast Soundings and EDAS RAOB Raleigh-Durham International, NC
(valid 12Z 3 December 2000)
Models too moist below 850 mb
Cloud depth over-predicted by models
Less drying above 600 mb than in models
Cloud top T ~ -10C
Shallow Cloud Layer
The ETA model microphysics scheme allowed ice crystal growth at –10 degrees C which allowed precipitation generation to occur in shallow,
super-cooled water droplet clouds.
HOW DOES THIS AFFECT THE QPF?
Dendrite growth occurred in an non-conducive thermal environment generating spurious precipitation
Increased moisture availability in the models through an extended depth over-quantified precipitation totals, when in reality the
atmosphere was dry above 700mb
SUMMARY AND CONCLUSIONS
Classification of QPF Error Sources
(1) Synoptic-scale features (i.e. 500 mb vorticity)
(2) Circular feedbacks from multiple factors
(3) Model parameterizations
Model Inter-comparison
+ MM5 achieved better QPF forecasts than Eta yet MM5 also under-estimated the CAD and vort max
+ MM5 also out-performed the Eta in coastal front location, strength of CAD, and track of the cyclone
DIRECTIONS FOR FUTURE RESEARCH
( & OTHER CONSIDERATIONS)
Sensitivity experiments using:- High-resolution SST- PBL and Microphysics options
Examine low-level moisture influx in model forecasts- Relationship to latent heat release
Statistically quantify the relative weight of each proposed hypothesis in generating model QPF errors
Vertical Motion & Moisture Cross-Section
EDAS vs. ETA 12h Forecast valid 00Z 3 Dec 2000Low-level drier air as a result of subsidence in CAD region over central NC
RH values increase over eastern NC around 700mb in region of ascent where OMEGA= -7mb/s
Model indicates region of subsidence isolated west of RDU adjacent to the mountains
To the east, upward vertical motion (up to -11mb/s) predominates in the entire 850-500mb layer
RH too low over central NC at 500mb indicating drying and instability between 700-500mb
RH too high around 700mb level continues mid-level instability
Vertical Motion & Moisture Cross-Section
EDAS vs. ETA 18h Forecast valid 06Z 3 Dec 2000Positive values of OMEGA indicates subsidence at low-levels over central NC and offshore
Upward vertical motion (OMEGA=-9mb/s) strengthens near 700mb over eastern NC
RH values increased in the 850-700mb layer across all of NC
Substantial drying below 850mb and aloft above 700mb
Model indicates upward motion from SFC-500mb
Two OMEGA maxima (-12mb/s, -16mb/s) over central/eastern NC in the 700-500mb layer
RH too high at low-levels in CAD region (possibly due to erroneous precipitation)
Dry air at 500mb too far west over mountains; RH too high above 700mb east of mountains