Download presentation
Presentation is loading. Please wait.
1
The Influence of Convective Parameterization on Model Forecasts of an East Coast Cyclone Photo courtesy of NWS Raleigh Kelly M. Mahoney and Gary M. Lackmann AMS WAF/NWP Conference 05 August 2005
2
Motivation How can forecasters know when they can best add value to NWP prediction? NWP forecasts can be highly sensitive to model physics [e.g., convective parameterization (CP) schemes] How can forecasters identify model forecast features characterized by less certainty, such as those driven by a CP scheme?
3
Motivation: Case Study 17 February 2004: challenging coastal front/cyclone, winter weather event Large upper trough approached East Coast, CAD event underway Where/when would cyclone form along coastal front? How far inland would wintry precipitation spread? AREA FORECAST DISCUSSION NATIONAL WEATHER SERVICE RALEIGH NC 305 PM EST MON FEB 16 2004 …WITH MODELS VERY DIVERGENT ACCUMULATIONS AND POSITION OF MAXIMUM AMOUNTS VERY VERY UNCERTAIN AND WILL NOT ISSUE ANY WATCHES OR ADVISORIES. AT BEST BELIEVE ADVISORIES MIGHT BE NEEDED BUT WITH SO MUCH UNCERTAINTY PREFER TO LET LATER SHIFT NAIL DOWN THE EXACT AREA OF CONCERN. A SPECIAL WEATHER STATEMENT WILL BE ISSUED. Eta QPF
4
12 UTC 17 February 2004 Manual surface analysis 500 hPa height, vorticity EDAS Analysis Cold-air damming Coastal front
5
00 UTC 18 February 2004 500 hPa height, vorticity Manual surface analysis EDAS Analysis Cyclone has formed along coastal front
6
Operational Eta model forecast 12Z 16 Feb Run, Valid at 06Z 18 Feb: 6-hourly precip (inches) Forecast temperatures < 0°C; frozen precipitation a concern 0.3+ 0.2 0. 1
7
Operational Eta 30-h forecast MSLP and 6-hourly convective precipitation (tenths of inches) 18Z 16 Feb operational run – Valid at 00Z 18 Feb SLP minima co-located with CP precip maxima
8
Workstation Eta model sensitivity experiment Coastal cyclones in Eta forecast co-located with convective precipitation maxima… Coincidence, or should forecasters treat these features as being less reliable? Test: Re-run case using workstation Eta: 1.) Betts-Miller-Janjic (BMJ) run (as operational) 2.) Kain-Fritsch (KF) run 3.) BMJ but with shallow convection disabled
9
Results – Total Precipitation BMJ: SLP, 3h QPFKF: SLP, 3h QPF BMJ – KF SLP diff BMJ – KF precip diff
10
Coastal Front Representation BMJ less-defined farther offshore winds mostly defined by two surface lows KF better-defined larger temperature gradient more convergence closer to coast brings precip farther inland BMJ 21Z 17 Feb KF 21Z 17 Feb Temp (shaded), 10-m wind streamlines
11
BMJ – No Shallow 27-hour forecasts from workstation Eta: Sea-level pressure, 3-h convective precipitation (mm) BMJKF 15 UTC 17 Feb - BMJ15 UTC 17 Feb - KF Character of precip, SLP more similar
12
BMJ: Shallow mixing scheme Why were there such significant differences? Shallow mixing eroded cap more rapidly; produced convection earlier in forecast. Inclusion of gridscale vertical motion produced convection in areas of large scale ascent. KF: Gridscale vertical motion F06 F12 F18 Black: Control BMJ Red: No Shallow Mixing Impact of CP scheme design: (29.7, -80.9)
13
Potential Vorticity (PV) Tools PV tendency equation: Vertical diabatic PV redistribution Horizontal diabatic redistribution Vertical diabatic term proportional to: vertical heating gradient background vorticity Friction Enhanced cyclonic convective response in cyclonic flow
14
BMJ 39-h forecast Workstation Eta forecast of 900–700 hPa potential vorticity (PVU), shaded as in legend at left, SLP (hPa, dashed contours), 3-h convective precipitation forecast (mm, solid contours) valid 03 UTC 18 Feb. Convective precip maxima in coastal-front zone: Coastal front Larger background vorticity More prone to diabatic “spinup” of lower PV maxima, cyclones (relative to convection in areas with lower background vorticity) BMJ Run 3-h convective precip (solid) 900-700 PV (shaded) SLP (dashed) L L Potential Vorticity (PV) Tools
15
So, what really happened? Why is this plow scraping slush around on a bare parking lot??
16
Observed radar: 00Z 17 Feb – 12 Z 18 Feb (3h intervals)
17
Event Synopsis Favorable upper-level jet streak dynamics produced a quick burst of snow early across central NC…tapering off to mist for the remainder of the day Southern and eastern NC received a mix of sleet and freezing rain early, changing over to all rain by afternoon
18
SLP Comparison Red: KF Blue: BMJ Green: Edas Double-barreled coastal low (BMJ) verified quite well early in event Lower central pressure, concentrated center (KF) more accurate by 12 UTC 18 th 00Z 18 Feb 12Z 18 Feb
19
Conclusions Sensitivity tests confirm CP scheme strongly tied to: –location, character of coastal front, –coastal cyclogenesis: very different from BMJ to KF Differences consistent with hypothesis that CP-driven features should be regarded with less certainty Diabatic PV spin-up enhanced in regions of larger background vorticity, such as coastal front zone Plot convective precipitation, lower-tropospheric PV, SLP to pinpoint such features
20
Acknowledgements Forecasters, NWS – Raleigh NCEP – Workstation Eta NCEP – Matt Pyle, for Eta IC data NCEP – Brad Ferrier, for discussions on shallow mixing in BMJ scheme Drs. Mike Brennan, Al Riordan, and Sethu Raman
21
The End
23
Operational Eta 30-h forecast MSLP and 6-hourly convective precipitation (tenths of inches) 18Z 16 Feb operational run – Valid at 00Z 18 Feb Convective precipitation maxima in cyclonic coastal-front zone (larger background vorticity) therefore more prone to diabatic “spinup” of lower PV maxima, cyclones (relative to convection in areas with lower background vorticity)
24
Operational Eta model forecast 12Z 16 Feb Run, Valid at 06Z 18 Feb: MSLP(mb)
25
Why were there such significant differences? BMJ-KF Lifted Index (f03) KF omega (cm/s) and 3-hr convective precipitation (f27) BMJ: Shallow mixing schemeKF: Gridscale vertical motion
26
Other differences? 03Z 17 February BMJ 2-m dewpoint(°C),mslp KF 2-m dewpoint(°C),mslp 10m WS diff (kts) (BMJ – KF)2m Td diff (C) (BMJ – KF)2m LHF diff (BMJ – KF)10m WS diff(kts)(BMJ – KF) BMJ LHF (W/m2),mslpKF LHF (W/m2),mslp KF directly interacts with the gridscale environment via downdrafts and dry air entrainment. This likely contributes to the production of a greater LHFs due to cooler and drier surface temperatures in locations where the KF scheme has activated.
27
Comparison to Observations Purpose of study was to examine sensitivity Assessing accuracy important as well (to relate output to forecast considerations) EDAS analysis used
28
Precipitation comparison BMJ 3h precip (mm),mslp 12Z 17 FebKF 3h precip (mm),mslp 12Z 17 Feb Radar mosaic 09Z 17 Feb Radar mosaic 12Z 17 Feb
29
Precipitation comparison BMJ 3h precip (mm),mslp 00Z 18 FebKF 3h precip (mm),mslp 00Z 18 Feb Radar mosaic 21Z 17 FebRadar mosaic 00Z 18 Feb
30
Future Work Perform a more in-depth observational analysis Complete further modeling studies to assess the contribution of other processes to coastal cyclogenesis and frontogenesis. Acquire more accurate analyses to better compare model output. Share results with NWS forecasters
31
EDAS sea level pressure analysis with ship and buoy observations at 18Z 17 February
32
SST 17 Feb
33
Outline Motivation Event synopsis Forecast considerations Modeling study and results Summary Plans for future work
34
Results – CP Scheme Precipitation BMJKF BMJ – KF SLP diff BMJ – KF convective precip diff
35
Convective precip at 15Z (“X” marks skew-T location, (35;-75)) Skew-T at 09Z at (35;-75) Skew-T at 15Z at (35;-75) Warming due to LH release…
36
Motivation: Case Study 17 February 2004: challenging coastal front/cyclone, winter weather event Large upper trough approached East Coast, CAD event underway Where/when would cyclone form along coastal front? How far inland would wintry precipitation spread? AREA FORECAST DISCUSSION...UPDATED NATIONAL WEATHER SERVICE RALEIGH NC 905 PM EST MON FEB 16 2004 MODEL REPRESENTATION OF THE COASTAL LOW DEVELOPMENT WILL LIKELY BE HIGHLY SENSITIVE TO CONVECTIVE FEEDBACK. THIS IS EVIDENCED IN THE ETA...WHERE THREE SEPARATE LOW CENTERS ARE SEEN OFFSHORE BY 21Z TUESDAY...EACH CO-LOCATED WITH A MAXIMUM IN CONVECTIVE PRECIPITATION. Eta QPF
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.