A Goodness of Fit Analysis for Functional Fits to DSD samples collected at Badlands National Park

Direct measurements of rain DSDs were taken using a Parsivel disdrometer at the Badlands National Park in South Dakota in the summer of 2011. The rain events were categorized as light, moderate, and heavy based on the rainfall rates. Heavy events that included hail were considered an additional category. Rainfall events were also classified as either convective or stratiform based on the radar data. Histograms of raindrop size samples show a significant separation between the hail-producing events and the rainfall events which did not include hail.

This presentation includes the results obtained from determining the gamma DSD parameters for the disdrometer datasets collected, using the modified maximum likelihood (ML) method developed here. The parameters obtained for events with hail are compared with those for non-hail events. Other potential functional forms for the DSDs (beta, lognormal and Weibull) are also considered. An objective procedure for evaluating the goodness-of-fit of the different estimators is also applied.