ounding nalog etrieval ystem Ryan Jewell Storm Prediction Center Norman, OK S A R S Sounding Analog Retrieval System
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
ounding nalog etrieval ystem
Ryan Jewell
Storm Prediction CenterNorman, OK
S A R SSounding Analog Retrieval System
What is SARS?
SARS is a forecast system based on sounding analogs.
The algorithm matches forecast soundings to a large database of proximity soundings associated with severe weather.
SARS finds matches using a small number of parameters and parameter ranges determined by a calibration process.
Name inspired by MARS – Map Analog Retrieval SystemGreg Carbin – http://www.spc.noaa.gov/exper/mref_mars/
Safe to use!
Used experimentally at the SPC. Integrated into NSHARP (Sounding displays)
RUC and NAM plan view display (Model Grids)
What is SARS?
Two types: Hail and Supercell/Tornado
Hail: 1148 Severe hail proximity soundings (Observed)
Supercell Tornado: 938 Supercell proximity soundings (RUC) (Under Development)
Types of SARS
Hail SARS can forecast:
1) Probability of SIG (≥ 2.00”) hail.
2) Maximum expected hail size (≥ 0.75”) .
Matching Sounding Database
• Includes 1148 observed hail soundings 1989-2006.• Within 100 nm and +/- 2.5 hrs either side of 2330Z (21-02).• Had to be in same air mass as storm.• Modified for surface conditions (if needed).• Thrown out if contaminated by outflow, etc.
Expansion of dataset used in Jewell and Brimelow
(WAF 2009).
Severe Hail Proximity Soundings
Matching Sounding Database
Assume dataset is “representative.”
Spans all seasons
18 years of data
All regions of the CONUS
1989 - 2006
A function of climatology and quality of soundings.
SARS Calibration Method
1 Matching Parameters – Relevant parameters associated with severe storms (various measures of instability and shear associated with hail).
2 Define initial ranges for each parameter to be used in search. (Example +/- 500 CAPE)
3 Test each sounding independently against the database, analyze matches.
4 Adjust parameters and ranges until the desired result is received.
Desired Result = The majority of matches agree on a particular type and magnitude of severe weather, and it verifies.
If a sounding is associated with 3.00” hail, most of the SARS matches should be very large hail.
Determine matching parameters and ranges
Example – Calibration for hail SARS.
Remove 1 sounding…test against remaining soundings (1147).
Calculate skill scores for parameter set #1 and range combination # 1...
Test various combinations of parameters and parameter ranges.
8 different parameters with 5 ranges each = 58 or 390,625 combinations.
SARS Matching Parameters
Most Unstable (MU) CAPEMixing Ratio of MU Parcel700-500 mb Lapse Rate
500 mb Temperature0-6 km Bulk Shear
Final list of matching parameters(out of about 20)
Notably showed little or no skill:Freezing Level
Wet Bulb Zero Heights0-3 Storm Relative Helicity (SRH)
SARS Parameters Ranges
Significant Hail Parameter Ranges(Resulted in best skill scores)
MUCAPE +/- 40%
Mixing Ratio of MU Parcel +/- 2.0 g/kg
700-500 mb Lapse Rate +/- 1.5 C/km
500 mb Temperature +/- 7 C
0-6 km Bulk Shear +/- 9 m/s
Large ranges, but all 5 must overlap.
Performance
SARS SIG Hail Algorithm
SARS Skill Scores
Significant Hail (≥ 2.0”) ForecastTotal Soundings = 1148
Hit MissFalse
Alarm
Correct
NullNo Matches
Found
486 84 97 475 * 5
CSI TSS POD FAR
0.729 0.683 0.853 0.166
NOTE: Highest skill score AND highest % with matches
* 1 Tie
SARS Skill ScoresSignificant Hail Forecast - Filtered
Remove Golf Ball (1.75”) and 2.00” reports (near 2” threshold)
Total Soundings = 889
CSI TSS POD FAR
0.843 0.784 0.945 0.113
Hit MissFalse
Alarm
Correct
NullNo Matches
Found
477 28 61 318 * 4
NOTE: Highest skill score AND highest % with matches
* 1 Tie
Performance
SARS SIZE Algorithm
Mean value of SARS binned by report size – Observed vs. Forecast
MEAN STDEV: 0.43”MEAN STDEV: 0.43”
Correlation (All): 0.68 r2= 0.47
Correlation (Filtered): 0.75 r2= 0.56
SARS MATCHING EXAMPLES
(one HAIL of a year!)
National Record - July 23, 2010 – Vivian, SD
Contours = # Matches
Color Fill = % that are SIG (≥ 2.00”)
RUC Model
GRIDDED SARS EXAMPLE
Mean SARS Hail Size (inches) RUC Model
KS Record (dia) - Sep 15, 2010 – Wichita, KS
7.75” Hail
2.75” Hail
4.25” Hail
Contours = # Matches
Color Fill = % that are SIG (≥ 2.00”)
RUC Model
CIN
CIN
CIN
RUC ModelMean SARS Hail Size (inches)
CIN
CIN
CIN
RARE EVENTS
Put AZ Hail Case HERE
Contours = # Matches
Color Fill = % that are SIG (≥ 2.00”)
RUC Model
Put AZ Hail Case HERE
Mean SARS Hail Size (inches) RUC Model
Contours = # Matches
Color Fill = % that are SIG (≥ 2.00”)
RUC Model
SUPERCELL SARSForecast Soundings
F5 Tornado
F4 Tornado
F4 Tornado
F5 Tornado
SARS Summary
The SARS method can be applied to various types of severe weather (hail, tornado, wind).
SARS forecasts storm REPORTS! Local biases in reporting WILL be reflected in SARS!
SARS may miss rare events if they have not been accounted for in the database, but may also find rare events and heighten awareness.
SARS is conditional…cannot predict whether storms will form (capping, forcing issues).
And oh, by the way…accuracy of SARS heavily depends upon the forecasts models.
This slide intentionally left blank.
ND Record - Jul 14, 2010 – Sioux County, ND
1.75” Hail
1.75” Hail
Related Documents
