A New Algorithm for Inversion of Aerosol Size Distribution Data

Abstract

The determination of an aerosol size distribution from conventional data is an ill-posed problem. The general characteristics of the aerosol size distribution inversion problem are discussed, and several existing methods are assessed. A new algorithm, in the spirit of Twomey's constrained linear inversion algorithm, employing the concept of generalized cross validation is developed. Extensive numerical tests on simulated data from a Marple impactor show superior performance of the method. Different versions of the algorithm are available based on different choices of the function spaces in which the assumed distributions lie. The new algorithm offers promise as the most accurate now available for inversion of aerosol data.