In this practical exercise, you will gain experience in evaluating the skill of a GCM prediction system, and compare it to a simple statistical method. You will be led into thinking about different measures to evaluate forecasts. During this course, the correlation coefficient between forecast and observed has been widely used as a measure to indicate the overall accuracy of a set of forecasts. An advantage of this is the wide use and understanding of the correlation coefficient as a statistical measure. However, this practical exercise will help you appreciate the importance of a range of measures of the forecast attributes (the issue of forecast evaluation is a broad subject and details are beyond the scope of this course).

Exercise 3 can be completed using Microsoft Excel or another similar application. You will need the datafile that was made available to you at the end of Exercise 2.

Download Exercise 3 (PDF Document)