Transcript Text Formatter Workshop

EGOWS 2008
Systematic forecasting of weather
“type” in the GFE
John Bally
CAWCR
GFE
How do we describe the weather?
Widespread
Precipitating Weather Forecast Process Map
Deep and Shallow
Instability
Potential Type of
Precipitating Weather
Upper Level
Moisture
Snow
Level
Probability of
Precipitation
Weather Grid
Cloud
Cover
Expected Intensity
of Precipitation
GFE
Will the precipitation be convective ?
GFE
The model upper RH fields have skill.
Is there enough moisture for rain?
GFE
Diagnose weather type from instability,
cloud cover and upper RH
GFE
Start with the Probability of Precipitation
from a multi model ensemble.....
GFE
Match the statistics of ensemble PoP to
climatology.......Limit PoP to 70%
GFE
Use PoP to delineate areas of weather
and assign coverage
GFE
Make sure that the PoP and the most
expected precipitation amount match.....
GFE
What effect will the wind field have on
Precipitation ?
GFE
Increase forecast precipitation where
convergence will trigger showers.....
GFE
And set the weather intensity from the
expected precipitation rate.....
GFE
08 Jun 18Z
The previous day, with a cold front
crossing the west Australian coast .....
>>
09 Jun 00Z
>>
09 Jun 06Z
>>
09 Jun 12Z
GFE
08 Jun 18Z
Forecasters expect more instability than
the model shows......
>>
09 Jun 00Z
>>
09 Jun 06Z
>>
09 Jun 12Z
Feature Based Precipitating Weather Process
Define Feature
Track Feature
Associate Weather
Type with Feature
Potential Type of
Precipitating Weather
Probability of
Precipitation
Weather Grid
Expected Intensity
of Precipitation
GFE
Draw in the axis of the cold front...
GFE
And again 12 and 24 hours later....
GFE
Make contours from these lines... Value
is the time of wind change .......
GFE
Use select tool to dynamically pick out
the area from 3 hrs ahead of the
change to 6 hours behind it .....
GFE
08 Jun 18Z
The cold frontal zone moving over
western Australia .....
>>
09 Jun 00Z
>>
09 Jun 06Z
>>
09 Jun 12Z
GFE
08 Jun 18Z
Use this “edit area” to assign “showers
and storms” potential weather type
near the front .....
>>
09 Jun 00Z
>>
09 Jun 06Z
>>
09 Jun 12Z
GFE
08 Jun 18Z
Again, calculate the forecast weather
from potential weather type, PoP and
expected precipitation intensity ...
>>
09 Jun 00Z
>>
09 Jun 06Z
>>
09 Jun 12Z
GFE
08 Jun 18Z
Also increase wind speed near the
change line... often under-forecast by
model......
>>
09 Jun 00Z
>>
09 Jun 06Z
>>
09 Jun 12Z
GFE
Now we produce some words....
Weather Types are described by........
Coverages : How much of specified area is impacted ?
Intensities : how strong ?
Attributes : optional features of the weather type
Coverage, intensity and attributes combined .... e.g.
“Widespread heavy showers with hail”
GFE



Sample and summarise the grids ....
Take a time-series of grids and perform mathematical
operations, including:

Averaging, Min/Max, deciles etc

Collection of weather keys
These operations produce samples, which represent the best
numerical description over the space & time in question.
e.g. Wind : ((5, 10), (300, 340))
MaxT : (22, 28)
Weather : ((SctSH+ 32%), (PaRa- 20%), (AreasTS 10%))
GFE Describe the situation over time.....




Examines the grids at the minimum time resolution, e.g. 3
hourly resolution
Identify times in the sequence where the samples change
significantly
Choose significant transition points and produce one subphrase for each time chunk.
e.g. “Light wind becoming northerly 15-20 knots during
the morning then tending southwesterly in the evening.
GFE

Combine weather within a type.....
Look at similarity of weather between subphrases
Is it worth describing a transition from the chance of light
rain to the chance of moderate rain?


Generally combine across one coverage OR intensity

Combined more aggressively in complex situations
GFE Describe the stats of each period...

Take the statistics for each chunk (time period)

Produce a scalar, vector, or weather description


Wind ((5, 20), (300, 340)) 
“north to northwest wind up to 20 knots”
Wx (isolSh-) 
“isolated light showers” or “chance of a light shower”
GFE



Talking about changes.....
Transition style for trends while over-time,
e.g. “becoming light around midday”.
Over-time style for isolated events,
e.g. “Rain during the evening”.
Use transition words......eg
 Wx (isolSh-, WideRa)  “isolated light showers
increasing to widespread rain”
 Wx (isolSh-, WideSh)  “isolated light showers
becoming more widespread”
GFE
and produce the worded forecast ...
GFE
with some help from our testing
infrastructure ...