Title: An%20Automated%20Synoptic%20Typing%20System%20Using%20Archived%20And%20Real-time%20NWP%20Model%20Output
1An Automated Synoptic Typing System Using
Archived And Real-time NWP Model Output
- Robert Dahni
- Meteorological Systems
- Central Operations and Systems Branch
- Bureau of Meteorology
- 19th International Conference on Interactive
Information and Processing Systems for
Meteorology, Oceanography, and Hydrology, Amer.
Meteor. Soc., Long Beach, California, February,
2003. - 11 February 2003
2Overview
- Background MENTOR
- Synoptic Classification manual correlation-based
map-pattern eigenvector-based - Tools Synoptic Typer Map Browser
- Examples weather variables associated with
synoptic types - Future Developments
3Background
- MENTOR (Ryan et al, 2003)Mentor is a web-based
system which allows forecasters to record in
real-time their assessments of likely
meteorological problems of the day, forecast
difficulty and their estimates of the value of
objective guidance entries accumulate in the
Mentor database, and can be quickly analysed and
searched by forecasters to assist in subsequent
forecasting decisions. Automatic synoptic type
classification is an important element of the
system.
4Manual classification
- Treloar and Stern (1993)
- Direction, strength and curvature of the surface
flow - 50 synoptic types
- 0900 hours EST MSLP station data (1957-2002)
- SE Australia
- Spreadsheet (Excel) computation
- Updated using NCEP grids (1948-2001)
- Interpolated to station locations
- Updated synoptic types (Stern, 2003)
5Correlation-basedmap-pattern classification
- Jasper and Stern (1983) seasonal sampling 22
years38 synoptic types - Updated using NCEP grids (1948-2001)
- 2.5o resolution (SE Australia)
- 00UTC MSLP analyses
- Correlation thresholds(0.7, 0.75, 0.8, 0.85,
0.9, 0.95) - Number of keydays (lt100)
- Number of synoptic types(10, 15, 20, , 90, 95)
- Minimum group size (1)
- Resources IDL 5.5 and UNIX server
NCEP grids
correlate
correlation matrix
derive
keydays
catalog
synoptic types (csv)
analyse
Daily data (years) Disk or RAM (Mb) CPU (hours)
54 780 15
synoptic types (binary)
statistics
6Eigenvector-based classification
- Dahni and Ebert (1998) automated objective
synoptic typing - Simple pattern recognition scheme with fields of
MSLP as input - METANAL 00UTC MSLP analyses 1.5o resolution
1970-1993 - Principal components and cluster analysis
techniques - First 5 principal components 20 clusters
Melbourne
7Eigenvector-based classification
NCEP grids (MSLP, 850 hPa temperature, 1000 and
500 hPa geopotential height and wind,
precipitable water, OLR)00, 06, 12 and 18UTC
analyses 2.5o resolution 1948-2001
8Synoptic Typer
Graphical User Interface
- Interactive (GUI-based) mode for development
- Developed on PC (Windows) using IDL 5.5
- Cross-platform (Windows, Linux, UNIX) application
9Synoptic Typer
- Non-interactive (batch) mode for operational
implementation (UNIX) - Existing C module used to extract NWP grids
from real-time the NEONS/ORACLE database - Automatic synoptic classification of real-time
NWP model output (e.g. GASP, EC and LAPS) - Real-time synoptic type guidance stored in the
Forecast Database - Automatic synoptic type for the MENTOR system
STNNUM, FCST_TIME, SYNT 086071, 2002092600,
7 086071, 2002092700, 7 086071, 2002092800,
18 066062, 2002092600, 2 066062,
2002092700, 13 066062, 2002092800,
12 040842, 2002092600, 1 040842, 2002092700,
3 040842, 2002092800, 13 014015,
2002092600, 3 014015, 2002092700,
11 014015, 2002092800, 14 009225, 2002092600,
8 009225, 2002092700, 8 009225,
2002092800, 12 023090, 2002092600,
7 023090, 2002092700, 7 023090, 2002092800,
5 094010, 2002092600, 11 094010, 2002092700,
5 094010, 2002092800, 16
10Map Browser
Graphical User Interface
- Interactive
- NCEP grids
- Vector, Barb or Streamline
- Derived fields
- Tropical Cyclones
- Synoptic Types
- Mean fields
- Interpolate data
- Weather variables
- Batch mode
11Example (manual classification and Melbourne
rainfall)
Treloar and Stern (1993) Synoptic Types50 NCEP
grids, Years1948-2001 Days19724 Rain Days gt 30
mm 127
Synoptic Type Freq () Rain Daysgt 30 mm ()
41 4.0 18.1
43 1.5 17.3
27 3.4 15.0
12Example (correlation-based map-pattern
classification and Melbourne rainfall)
NCEP grids Years1948-2001 Days19724 Threshold0.
90 Synoptic Types50 Rain Days gt 30 mm 127
Synoptic Type Freq() Rain Daysgt 30 mm ()
49 1.1 18.1
21 2.1 11.8
28 1.5 8.7
9 2.0 7.9
46 1.3 7.9
13Example (correlation-based map-pattern
classification and Melbourne heat waves)
NCEP grids Years1948-2001 Days19724 Threshold0.
90 Synoptic Types50 Heat Wave Days 136
Synoptic Type Freq() Heat Wave Days ()
17 2.1 22.8
34 1.6 12.5
8 2.0 10.3
4 3.4 9.6
14Future Developments
- Synoptic Types operational implementation,
multiple input fields, correlate sequence of
days, extension to other regions - Associate Weather Variables with Synoptic Types
significant rainfall, heat waves, fog events,
forecast errors (verification) - For further information go to the following web
site - http//www.bom.gov.au/inside/cosb/mss/projects/syn
optictyper/