A Roughness-Detection Technique for Objectively Classifying Drops and Graupel in 2D-Image Records

Abstract

The development and evaluation of a new computerized technique for classifying drops and graupel in two-dimensional (2D)-image records is described. The method is unique because images are classified as drops or graupel on the basis of their exterior roughness rather than shape. The technique involves using the method of least squares to fit a fourth-order polynomial to the outside curvature of each half of large, symmetric, circular images. Formulations for determining variance and polynomial coefficients are reviewed. Roughness criteria determined using 2D-C and 2D-P cloud data in a quadtree analysis of maximum variance of the polynomial approximations and image diameters are illustrated. A method for determining the radius of ?center-out? images is also reviewed. Size distributions formed by combining 2D-C and 2D-P data for either drops or graupel are illustrated with error ban based on Poisson statistics. Two different methods of calculating water content based on size-distribution information for particles with diameter greater than 150µm are demonstrated. An independent evaluation of the objective classification technique using 2D-C and 2D-P cloud data shows that the polynomial classification of images as drops or graupel preformed sufficiently well to give a population of size-distribution parameters and water contents that were generally not statistically different from those obtained by human classification.