The Binormal Distribution: A Distribution for Representing Asymmetrical but Normal-like Weather Elements

Abstract

A new probability distribution, which has some of the advantages of the normal distribution but avoids the constraint of symmetry undesired in many applications, is presented. The distribution, called binormal, has three parameters, as the standard deviations on the two sides of the most probable value are different, but it also includes the Gaussian distribution as a special case. A test for symmetry and a method for the estimating the parameters are also shown. The distribution will be employed to express probabilistic temperature forecasts, but its climatological application for modeling certain temperature distributions is also useful and recommended.