A Simple Physically Based Snowfall Algorithm - PowerPoint PPT Presentation

About This Presentation
Title:

A Simple Physically Based Snowfall Algorithm

Description:

Develop complimentary snow amount/ratio code for use in Bufkit ... The layer snow ratios from step one are weighted by the percent of column ... – PowerPoint PPT presentation

Number of Views:100
Avg rating:3.0/5.0
Slides: 36
Provided by: NWS6
Category:

less

Transcript and Presenter's Notes

Title: A Simple Physically Based Snowfall Algorithm


1
A Simple Physically Based Snowfall Algorithm
  • Daniel K. Cobb Jr.
  • Science Operations Officer
  • WFO Caribou, ME

2
Introduction
  1. Motivation and Goals
  2. Description of Algorithm
  3. Example Case
  4. Summary
  5. Future Work
  6. References
  7. Questions

3
Motivation Goals
  • Improve on 101 snow ratio assumption
  • Incorporate aerial and temporal variation of snow
    ratio over a storm.

4
Motivation Goals
  • Develop a Snow Amount SmartTool for GFE
  • Physically based population of snowfall from QPF
  • Good base tool in terms of collaboration
  • Develop complimentary snow amount/ratio code for
    use in Bufkit
  • Excellent Interpretation/interrogation tool for
    forecaster

5
Motivation Goals
  • HISTORY
  • Initial interest began in 2000.
  • Idea further inspired by
  • Top-Down microphysics of Baumgardt
  • Crosshair approach of Waldstreicher
  • Canadian snow ratio decision tree algorithm by
    Dubè
  • Snow density diagnostic of Roebber

6
Algorithm
  • SNOW CRYSTAL BASICS
  • Crystal habit depends
  • Primarily on temperature
  • Secondarily on relative humidity
  • Largest crystals (dendrites) form at temperatures
    between (-12C and -18C)
  • Crystal growth rates are also the largest in this
    temperature range.

7
Algorithm
8
Algorithm
  • To a first approximation, the amount of cloud
    mixing ratio formed in any layer will be related
    to its relative humidity and vertical motion.
  • This provides a basis for inferring the amount of
    crystal habit any one layer will contribute.

9
Algorithm
  • FOUR STEP PROCESS
  • Layer snow ratios are calculated for all
    available NWP levels based on temperature.
  • The vertical motion of each layer is scaled based
    on the relative humidity of the layer.
  • A column total vertical motion is calculated as
    the sum of the scaled layer vertical motion.
  • The layer snow ratios from step one are weighted
    by the percent of column vertical motion and
    summed to obtain a base snow ratio.
  • The base snow ratio is then multiplied by the QPF
    to obtain snowfall.

10
Algorithm Example
T -25C ? -5 µbs-1
T -15C ? -10 µbs-1
T -5C ? -5 µbs-1
Consider a 3 layer cloud with the following layer
average temperatures and vertical motion First
map temperatures to a snow ratio
11
Algorithm Example
12
Algorithm Example
SR 81 ? -5/-20 µbs-1
SR 241 ? -10/-20 µbs-1
SR 91 ? -5/-20 µbs-1
Layer temperature has now been mapped to snow
ratio (SR) The percent layer contribution to
vertical motion is now being calculated.
13
Algorithm Example
SR 81 ? -5/-20 µbs-1 8.0 0.25 2.0
SR 241 ? -10/-20 µbs-1 24.0 0.50 12.0
SR 91 ? -5/-20 µbs-1 9.0 0.25 2.25
The weighted layer snow ratios are summed up over
the cloud yielding the base snow ratio. The snow
ratio would then be 2.0 12.0 2.3
16.3 Or 161
14
Algorithm Example
The snowfall is obtained by multiplying the snow
ratio by the QPF. A QPF of 1.50 and the
calculated snow ratio of 161 would yield 1.50
16 24 inches
15
Example (2004Jan19)
  • Localized heavy snowfall from pivoting inverted
    surface trough and eastward extending upper low.
  • SOOs neighborhood was ground zero with 21 inches
    of rather fluffy snow!
  • Maximum snowfall rates approaching 3 inches per
    hour occurred at about 15Z on Jan19th.

16
(No Transcript)
17
(No Transcript)
18
(No Transcript)
19
(No Transcript)
20
(No Transcript)
21
(No Transcript)
22
CarSnowAmt SmartTool
  • Collaborators
  • Dave Novak (ERH, SSD)
  • Jeff Waldstreicher (ERH, SSD)
  • Tom Lebvre (FSL)
  • Test version now available from STR
  • Currently useable with Eta80, Eta40, and WSEta.
    (GFS80 coming in OB4)

23
(No Transcript)
24
(No Transcript)
25
(No Transcript)
26
Snow Amount Bufkit
  • Planned incorporation into Bufkit
  • Currently exists as Perl program which uses
    Bufkit files to perform calculations
  • Compliments GFE SmarTool by allowing forcaster to
    critique the answer.
  • Additional precipitation type logic currently
    being developed.

27
Bufkit Example 2004Jan19
28
(No Transcript)
29
StnID Date/hour FcstHR QPF SfcT
SnR Snow CumSnw CumQPF

727130 040118/1800 0 0.000 -7.8
0.0 0.0 0.0 0.00 727130 040118/1900
1 0.004 -6.9 20.4 0.1 0.1
0.00 727130 040118/2000 2 0.016
-7.0 19.5 0.3 0.4 0.02 727130
040118/2100 3 0.024 -7.0 19.9
0.5 0.9 0.04 727130 040118/2200 4
0.024 -7.2 17.1 0.4 1.3 0.07 727130
040118/2300 5 0.024 -6.8 20.0
0.5 1.7 0.09 727130 040119/0000 6
0.020 -6.5 16.8 0.3 2.1 0.11 727130
040119/0100 7 0.020 -6.1 15.7
0.3 2.4 0.13 727130 040119/0200 8
0.020 -5.8 15.4 0.3 2.7 0.15 727130
040119/0300 9 0.028 -5.8 15.1
0.4 3.1 0.18 727130 040119/0400 10
0.035 -5.9 14.8 0.5 3.6 0.21 727130
040119/0500 11 0.039 -5.8 15.0
0.6 4.2 0.25 727130 040119/0600 12
0.043 -5.9 14.8 0.6 4.8 0.30 727130
040119/0700 13 0.047 -5.9 14.7
0.7 5.5 0.34 727130 040119/0800 14
0.047 -6.0 14.9 0.7 6.2 0.39 727130
040119/0900 15 0.047 -6.0 15.2
0.7 7.0 0.44 727130 040119/1000 16
0.043 -6.2 15.2 0.7 7.6 0.48 727130
040119/1100 17 0.039 -6.2 14.3
0.6 8.2 0.52 727130 040119/1200 18
0.039 -6.2 14.0 0.6 8.7 0.56 727130
040119/1300 19 0.043 -6.0 14.4
0.6 9.3 0.60 727130 040119/1400 20
0.047 -5.4 14.9 0.7 10.1 0.65 727130
040119/1500 21 0.051 -4.9 15.0
0.8 10.8 0.70 727130 040119/1600 22
0.055 -4.6 15.2 0.8 11.7 0.76 727130
040119/1700 23 0.051 -4.1 15.8
0.8 12.5 0.81 727130 040119/1800 24
0.043 -3.7 15.7 0.7 13.1 0.85 727130
040119/1900 25 0.039 -3.5 16.1
0.6 13.8 0.89 727130 040119/2000 26
0.035 -3.6 16.1 0.6 14.3 0.93 727130
040119/2100 27 0.031 -4.1 16.6
0.5 14.9 0.96 727130 040119/2200 28
0.028 -4.6 16.0 0.4 15.3 0.98 727130
040119/2300 29 0.024 -4.8 15.4
0.4 15.7 1.01 727130 040120/0000 30
0.020 -5.3 14.9 0.3 16.0 1.03
30
Verification (PQI) 2004Jan19
Date Time Snow Equiv Ratio
01/18 18Z 24Z 2.3 0.15 15.3
01/19 00Z - 12Z 9.8 0.61 16.1
01/19 12Z 18Z 8.6 0.42 20.5
01/19 18Z 24Z 0.8 0.08 10.0
Storm Total 21.5 1.26 17.1
31
Eta Forecast 01/17 12Z
Location Snow Equiv Ratio
Caribou 9.7 0.58 16.7
Houlton 13.8 0.84 16.4
Millinocket 12.6 0.82 15.4
Bangor 8.2 0.67 12.2
Eastport 9.9 0.83 12.0
32
Summary
  • Initial results A weighted average approach to
    snow-ratios works well.
  • Such an approach is computer calculation
    friendly.
  • Predicted ratios are very similar to those found
    using Dubè decision tree.
  • Decision trees are people friendly.
  • Applying snow-ratio diagnostic techniques
    improves forecast location of snowfall amounts as
    well as snowfall axes.

33
Future Work
  • Snow ratios up to 1001 have been observed
  • This is often the result of aggregates of
    spatially large dendrites. The aggregate being
    less dense than its constituent crystals.
  • Comprehensive snow study at WFO-CAR
  • Two sonic depth sensors
  • Measurements planned at 1, 3, and 6 hours.
  • ASOS LEDWI snowfall algorithm tests

34
References
  • Baumgardt, Dan, 1999 WintertimeCloud
    Microphysics Review. NWS Central Region,
    Available online at http//www.crh.noaa.gov/arx/m
    icrope.html.
  • Dube, Ivan, 2003 From_mm_to_cm. COMETs
    Northern Latitude Meteorology Webpage,
    http//meted.ucar.edu/norlat/snowdensity/from_mm_t
    o_cm.pdf.
  • Roebber, P. J., S. L. Bruening, D. M. Schultz,
    and J. V. Cortinas Jr., 2002 Improving Snowfall
    Forecasting by Diagnosing Snow Density. Wea.
    Forecasting, 18, 264-287.
  • Waldstreicher, J.S., 2001 The Importance of Snow
    Microphysics for Large Snowfalls, Preprints, 3rd
    Northeast Operational Workshop NOAA/NWS Albany,
    NY, Available online at http//www.erh.noaa.gov/e
    r/hq/ssd/snowmicro/.

35
Thank YouQuestion?
Write a Comment
User Comments (0)
About PowerShow.com