Model Diagnostics at HPC

1
Model Diagnostics at HPC

• Jessica Clark
• David Roth
• Forecasters
• NCEP/HPC/FOB
• http//www.hpc.ncep.noaa.gov/
• Camp Springs, MD

2
Products HPC Issues

• QPF
• 6 hourly through D3
• 48 hr QPF D4-5
• QPFPFD
• Flood Products
• Excessive Rainfall (94e)
• QPFERD
• River Flood Outlook
• Winter Weather
• Snow/Freezing Rain Probability graphics
• QPFHSD
• Winter Storm Summaries
• Low Track
• 3 hourly Surface Analyses
• D1-2 Basic Weather Products
• Fronts/Pressures
• Instantaneous Pcpn Potential and Pcpn Type

• Medium Range (D3-7)
• 24 hr Front/Pressures
• Max/Min/12 hr PoPs
• PREEPD and PMDEPD
• Hawaii Narrative
• Model Diagnostics
• 500mb prog for West/E Pac
• PMDHMD
• NDFD Chat Coordination
• Tropical Weather
• Backup/guidance for TPC
• Public Advisories for inland tropical systems
• South Amer. Caribbean text products
• Selected Cities/Travelers
• Daily Weather Map
• Air Quality Narrative

3
Steps in the Process

• Initialization
• Model continuity/wavering over time
• Model Biases
• Climatological considerations, including
  teleconnections
• Any consensus amongst current guidance?
• Continuity with previous shift/forecast

4
Model Initialization
5
Targeted Obs
D5 feature of interest
D5 spread in the ensembles
Day 5 forecast
Day 1 feature
Area of interest on D1 associated with the D5
feature
In the winter months, HPC can request Targeted
Observations be taken in order to improve the
data coverage in an area associated with a system
which has considerable uncertainty associated
with its forecast on D3-7
6
Surface initialization
7
850 hPa Initialization
8
500 hPa Initialization
9
Precipitable Water Initialization
10
250 hPa wind initialization
11
Model continuity wavering over time
12
Run-to-Run Comparisons
13
Day-to-day comparisons
14
Comparison across multiple days
15
Model Biases
16
Example from the ECMWF
Models can be too retrogressive with closed
cyclones shows up regularly on the ECMWF
particularly from 120 hours onward. In the case
of the ECMWF, a southward bias also appears due
to its relatively stronger mid-level ridges
17
Climatology Considerations in determining model
preference
18
Presence of warm/cold ENSO phase
19
Teleconnections with an anomaly center
20
Teleconnection-favored pattern
21
Is there a model consensus?
22
Use of ensemble members from the NCEP/GEFS,
ECMWF, and Canadian models in determining
preference
23
NCEP ensemble member spaghetti plots
24
Overall spaghetti plots
25
Ensemble low clustering
26
Model Diagnostic-related Products
27
Products relating to Model Diagnostics

• PMDHMD Model Diagnostic Discussion which covers
  model choices over the next 3 days, issued twice
  daily
• PREEPD/PMDEPD Extended Forecast Discussions
  which cover the days 3-7 period both in model
  choice and sensible weather impact
• 500 hPa graphics for 36-168 hours
• Fronts for days 3-7 for the United States and
  vicinity overnight

28
500 mb D3-7 Progshttp//www.hpc.ncep.noaa.gov/med
r/medr500.shtml

• Generated twice daily
• Blend generally used Manually editable
• Graphics Description model percentages and
  level of confidence
• Issued to web only
• Comparison fields between HPC H5 progs and 00Z
  GFS and 00Z NCEP Ensemble Mean

29
Medium Range Pressure Forecast
30
HPC vs models in the days 3-7 pressures