Susan Joslyn and Buz Hunt: Decision Making
in Weather Forecasting
Abstract
This talk will provide an overview
of a new project studying decision making in weather forecasting. Although modern
day weather forecasting is based on complex numerical models, their predictions
do not always agree, due to differences in inputs and assumptions. Each forecast
is associated with a degree uncertainty. Forecasters combine model predictions
with other information, such as current conditions and climactic trends, to
produce a forecast. Ideally model uncertainty would be included in the decision
process, although the current procedures tend to neglect this factor. The authors
are part of a multidisciplinary team of faculty from the Statistics, Sociology,
Atmospheric Sciences, Psychology and Applied Physics that has been assembled
to develop better ways to assess and present model uncertainty to weather forecasters.
Our focus is on military weather forecasters who are already overwhelmed by
a large volume of information and must make decisions rapidly. As the psychologists
on the team, we will determine the best method by which to communicate uncertainty
to military forecasters. The first stage of the process is to determine how
decisions are made in this domain and to determine the information sources upon
which decisions are currently based. We will present preliminary results from
observations and verbal protocol analyses that lay the groundwork for this 5-year
project.