The objective analysis of relative humidity

Abstract

A method of objective analysis of relative humidity is described which makes use of non-isotropic weighting functions. These depend on the gradient of the background field in such a way that observations in a direction along closely packed isopleths have higher weight than those in a direction perpendicular to them. Rainfall predictions with the United Kingdom fine mesh 10-level model are found to be better when based on this humidity analysis than on one using isotropic weighting functions and the results for a particular case are described. The use of surface observations in the lowest layer of the model is discussed and it is shown that it is better to use the surface humidity mixing ratio rather than the surface relative humidity. Resulting forecasts of rainfall are better and do not exhibit diurnal variations found when using surface relative humidity observations.