Abstract
An objective method of predicting the category of 15βday and monthly rainfall and mean temperatures in terms of terciles and quintiles over England and Wales is developed. The method selects those areas in the Northern Hemisphere where anomalies on 500 mb monthly mean charts preceding warm months have been significantly different from the anomalies in the same areas preceding cold months at the same time of year. For this purpose data over the period 1945β72 have been used and Welch's test has been applied to assess the significance of the difference between the two means.
Having found the areas which have been significantly different before warm and cold months, the anomalies at all grid points within these areas are added for each year of the data, taking due account of sign, and the years then ranked in terms of the total anomalies. The result is a list with most of the cold years at the top, the average ones in the middle and the warm ones at the bottom. If the same procedure is carried out on current data and the current month included in the ranking, a forecast of the probable temperature category can be made.
A similar procedure is used for rainfall, and for both elements on the 15βday time scale. A test involving some 250 forecasts over a period of a year shows excellent results.