Transcript An Overview of Precipitation Type Forecasting Using NAM and SREF Data Geoff Manikin

An Overview of Precipitation
Type Forecasting Using NAM and
SREF Data
Geoff Manikin
Mesoscale Modeling Branch
NCEP/EMC
Thanks to:
Jun Du – SAIC/GSO @ EMC
Keith Brill – NOAA/NCEP/HPC
Brad Ferrier – SAIC/GSO @ EMC
Pete Manousos – NOAA/NCEP/HPC
12 km
80 km
40 km
Operational NCEP Precip Type Algorithm
• Based on Baldwin and Contorno (1994)
• Examines a vertical thermodynamic structure that a falling
hydrometeor encounters as it falls to the ground
• first determines if precip is generated as supercooled
water or ice and then uses decision tree approach
• identifies “warm” (> 0ºC) and “cold” (< 0ºC) layers
•Computes area between 0 and Tw to identify layers
•Diagnoses snow if:
•Coldest T at any level with p > 500 mb is  -4°C and
•Area between Tw and –4°C < 3000 deg m
•Diagnoses freezing rain if:
• coldest T in saturated layer is > -4°C and Tsfc < 0°C OR
area between –4°C and Tw > 3000 deg m
One Significant Issue with the operational scheme:
• it intentionally overpredicts ZR/IP to
have high POD
• area in this sounding
between Tw and –4 C
is > 3000
• The algorithm would
predict ZR for this
sounding
The goal: offer alternative
precipitation type forecasts from the
single operational NAM model run.
The alternate solutions:
1) revise the NCEP scheme – perform check
on “warm area” between Tw and O°C
2) Ramer algorithm – based on the ice
fraction of precip reaching the ground
3) explicit frozen precipitation variable from
the model microphysics
Percentage of Frozen Precipitation
• Explicit Field from the NAM Microphysics (upgrade 11/01)
•Gives % of precip reaching the ground that is frozen – not a
probability!
•Ranges 0 (liquid) to 100 (frozen)
•High values indicate snow or sleet
•Low values indicate rain or freezing rain
- Web Page contains
nine geographical
regions
- Page is updated
for 00/12z NAM
cycles
Data available
for all 1300+
Eta BUFR
locations
Click on a star
(bypass option
available on
main page)
Sample Precip Type Meteogram
2-m
T, Td
10-m wind
precip
% of frozen pcp
Ramer
Alt. NCEP
NCEP
950-700mb
temps
Philadelphia Verification
fcst
obs
Syracuse Meteograms and Verification
fcst
obs
Need to Offer Alternate Algorithms
in Gridded Form
Run Multiple Algorithms on Single Run
Account for Initial Condition Uncertainty
in Addition to Algorithm Uncertainty w/ the SREF
How to Tally the Dominant Type for
Entire Ensemble?
Consider 5-member ensemble with all
5 algorithms predicting rain or snow
Member Rain Snow
1
3
2
RAIN
2
2
3
3
2
3
SNOW
SNOW
4
2
3
5
5
0
14
11
SNOW
RAIN
RAIN!3-2 SNOW!
SREF “Mean” Dominant P-Type
New
Product
Current
Product
Issues to be Resolved
Require 0.01” of mean precip to give a type?
Explicit Method
Often Gives
More ZR
Use 2-m T
instead of
skin T ?
Summary
• NCEP p-type scheme performs well but has
biases that invite an approach using multiple
algorithms
• Precip type meteograms are available at
http://www.emc.ncep.noaa.gov/mmb/precip_type
• Dominant p-type field will be available from
operational NAM this winter – provide NCEP and
Ramer algorithm output as well?
• Dominant p-type will also be available in the
SREF mean file