transnormalization

This procedure is termed transnormalization. Using variables that are transnormalized is a necessary, but not sufficient, condition for the joint distribution to be multi-normal. However, tests to date have shown that joint normality is a good assumption whenever the variables have been selected by nature, as opposed to having been selected artificially by man. Predictors selected by man, such as time of day, are included in the TRP equation as "marginal" variables.

1977, Proceedings of the 7th Technical Exchange Conference, El Paso, Texas, 30 November-3 December 1976

The three main procedures that comprise this multivariate transnormalized method are transnormalization, correlation, and regression conditional probability. The transnormalization process (a nonlinear process) transforms the observed or raw data into the standard normal variable with the same cumulative climatological probability as the raw predictor. Transnormalization ensures normal distribution of the predictor.

1978, Meteorological and Geoastrophysical Abstracts