1968 FLORIDA CUMULUS SEEDING EXPERIMENT: NUMERICAL MODEL RESULTS

Abstract

A one-dimensional numerical cumulus model was tested against data from a randomized seeding experiment made in South Florida in 1968. Fourteen GO clouds were studied. Nine were seeded by pyrotechnics with 1 kg of silver iodide each, while five were studied identically as controls. Various seeding subroutines and assumptions regarding the ice phase are compared. The experimental aircraft data are used to guide the modeling assumptions and to select the most realistic ones. Seedability and seeding effect correlate to 0.96 for seeded clouds in the three best models. A high correlation is found between seedability and radar-measured rainfall increase from seeding. Also, a high correlation is found between model predictions of the difference in precipitation fallout between seeded and control clouds and the measured rainfall differences, although the model predictions are much smaller in magnitude. A calculation is undertaken showing that coalescence within the cloud body on descent of the raindrops easily accounts for the discrepancy. The model predictions for each GO cloud are discussed in comparison with actual measurements on the cloud. The 1968 experiment was found to subdivide into two periods, one fair and one disturbed, with quite different effects of seeding. The two periods and corresponding cloud behavior are compared. It is concluded that the disturbed period was less favorable for seeding because of higher unseeded cloud growth and strong wind shear. Implications of this result for future modeling efforts are discussed.