Defining cognitive decision making processes in forecasting: a knowledge based system to generate weather graphics

Over recent years, Stern (2002, 2003, 2004a&b, 2005) has been involved in the development of a knowledge based weather forecasting system. The purpose of the present paper is to utilise the system to define and evaluate the cognitive decision making processes involved in the generation of extended-range day-to-day weather forecasts. Specifically, the weather forecasting product examined in the present work is that of weather graphics.

Fourteen different weather graphics are utilised. These are graphics representing:

Sunny

Dry

Haze

Windy

Partly Cloudy

Cloudy

Dust

Fog

Drizzle

Rain

Snow

Shower

Thunder

Cyclone

The weather graphics are generated from an algorithm that has a logical process to combine features of the temperature, precipitation and phenomena (for example, fog, thunder, and dust) components of the forecasts with features of the forecast synoptic type (strength, direction, and cyclonicity of the surface flow).

As an illustrative example, how the system incorporates NWP uncertainty into forecasts is now described. This is achieved:

(1) For weather forecasts 7 days in advance, by averaging the weather suggested by climatology, with the weather suggested 7 days hence by the NWP guidance;

[In the case of obtaining temperature components of the forecast suggested by the NWP guidance, this is done by utilising the "perfect prog" approach to statistically interpret the NWP guidance; in the case of obtaining precipitation components of the forecast suggested by the NWP guidance, this is done by averaging the 'raw' NWP precipitation forecasts, with forecasts obtained by utilising the "perfect prog" approach to statistically interpret the NWP guidance.]

(2) For weather forecasts 6.5 days in advance, by averaging that forecast generated previously (7 days in advance), with the weather now suggested 6.5 days hence by the NWP guidance.

(3) For weather forecasts 6 days in advance, by averaging that forecast generated previously (6.5 days in advance), with the weather now suggested 6 days hence by the NWP guidance.

...

...

(14) For weather forecasts 0.5 days in advance, by averaging that forecast generated previously (1 day in advance), with the weather now suggested 0.5 days hence by the NWP guidance.

... thereby systematically adjusting for new information while reducing the weight given to the earlier NWP guidance & climatology.

The National Oceanic and Atmospheric Administration's (NOAA) Global Forecasting System (GFSx) NWP model (http://www.arl.noaa.gov/ready/metdata.html) provides output that includes forecast data forecast data every 6 hours from forecast hours 0 to 180 on a 1 degree latitude/longitude grid covering the globe. In addition, this data is mapped to northern and southern hemispheric 257 x 257 grids. The data is updated 4 times per day.

The knowledge based forecasting system was utilised to objectively interpret the output of the GFSx model statistically in terms of local weather at Melbourne (maximum temperature, minimum temperature, probability of precipitation, amount of precipitation, a weather graphic depicting the expected weather during the morning, and a weather graphic depicting the expected weather during the afternoon) in order to rigorously establish how successfully the system incorporates NWP uncertainty into forecasts.

A 100-day trial of its performance is presently underway, with the knowledge based system generating twice-daily forecasts out to seven days in advance.

To illustrate, the forecasts leading up to Saturday April 2 are now given (the observed weather was fine, mainly sunny and windy with a minimum temperature of 20.2 deg C, and a maximum temperature of 31.6 deg C).

The format in the table below is [time forecast was produced, morning graphic, afternoon graphic, minimum temperature, maximum temperature, amount of precipitation, probability of precipitation]:

PM MAR 26, Cloudy, Cloudy, 16, 26, 0mm, 40%;

AM MAR 27, Cloudy, Cloudy, 18, 30, 0mm, 46%;

PM MAR 27, Shower, Shower, 20, 30, 1.3mm, 60%;

AM MAR 28, Shower, Rain, 20, 30, 3.4mm, 71%;

PM MAR 28, Cloudy, Cloudy, 18, 32, 0mm, 39%;

AM MAR 29, Windy, Windy, 20, 30, 0mm, 24%;

PM MAR 29, Sunny, Partly Cloudy, 20, 30, 0mm, 14%;

AM MAR 30, Sunny, Partly Cloudy, 20, 32, 0mm, 9%;

PM MAR 30, Sunny, Partly Cloudy, 18, 32, 0mm, 7%;

AM MAR 31, Sunny, Partly Cloudy, 18, 30, 0mm, 9%;

PM MAR 31, Sunny, Partly Cloudy, 18, 32, 0mm, 6%;

AM APR 1, Sunny, Sunny, 18, 32, 0mm, 4%;

PM APR 1, Sunny, Sunny, 16, 32, 0mm, 2%;

AM APR 2, Windy, Windy, 18, 32, 0mm, 4%;

After the first 55 days of the trial (Feb 14, 2005 to April 10, 2005), the overall percentage variance explained by the forecasts so generated is 44.3% (compared with 43.1% for the official forecasts).

Specifically for temperature, the percentage variance explained by the 523 minimum temperature and 578 maximum temperature forecasts so generated is 66.3% (compared with 66.7% explained by the official forecasts). The Root Mean Square Error of the forecasts so generated is 2.68 deg C (compared with 2.67 deg C for the official forecasts).

Specifically for precipitation, the percentage variance explained by the 578 quantitative precipitation and 578 probability of precipitation forecasts so generated is 22.3% (compared with 19.5% explained by the official forecasts).

Website

To view the graphic generator, go to:

http://www.weather-climate.com/graphicgenerator_sevendays_version5.html

or

http://www.weather-climate.com/graphicgenerator_15days_version7.html

References

Stern H (2002) A knowledge-based system to generate internet weather forecasts. 18th Conference on Interactive Information and Processing Systems, Orlando, Florida, USA 13-17 Jan., 2002.

Stern H (2003) Progress on a knowledge-based internet forecasting system. 19th Conference on Interactive Information and Processing Systems, Long Beach, California, USA 9-13 Feb., 2003.

Stern H (2004a) Incorporating an ensemble forecasting proxy into a knowledge based system. 20th Conference on Interactive Information and Processing Systems, Seattle, Washington, USA 11-15 Jan., 2004.

Stern H (2004b) Using a knowledge based system to predict thunderstorms. Presented at International Conference on Storms, Storms Science to Disaster Mitigation, Brisbane, Queensland, Australia 5-9 Jul., 2004.

Stern H (2005) Using a knowledge based forecasting system to establish the limits of predictability. Submitted to 21st Conference on Interactive Information and Processing Systems, San Diego, California, USA 9-13 Jan., 2005.