For current pollen predictions methods, readings are taken at the end of the day to approximate the conditions that will be for the next day. This delay produces inaccurate results because conditions for pollen release change from day to day. Using a ragweed model and historical pollen measurements, a program was developed in 2003 by University of Tulsa undergraduates in computer science. This program predicted the day's ragweed pollen levels for today based on information gathered from the National Weather Service as well as the historical pollen measurements. Using the system as a foundation, we constructed a generalized pollen forecaster for any pollen season as long as sufficient historical pollen measurements are available.
This generalized pollen forecaster, called PollenPro, allows a user to input regional conditions for a particular pollen and predicts a pollen level for the current day. PollenPro works with multiple pollen level variables, such as current wind speed, direction of wind speed, relative humidity, and temperature. The most important data are historical pollen levels, dating back a minimum of 5 years. Taking this empirical model into account a prototype program was designed for the ragweed season.
To make the program more generalized so that all pollen seasons can be evaluated by this program, it requires the ability of the end user to be able to modify the influences each of the formula pieces have on that particular forecast, the forecast database, and the websites of the particular region of forecast. These modifications along with a better storage mechanism constitute a major product revision from the original ragweed forecaster developed previously. The first revision addresses the need for better sources of weather data. The National Weather Service recently launched an XML feed service. With this new data stream, we place the program input in a standardized format for any weather location. Data access and storage are more user friendly, with the allowance for customization by the user. Finally, the program can be executed on a variety of different platforms.