Results of On-Top Glaciogenic Cloud Seeding in Thailand. Part II: Exploratory Analyses

Abstract

Randomized, cold-cloud, rain-enhancement experiments were carried out during 1991?98 in the Bhumibol catchment area in northwestern Thailand. Exploratory experimentation in 1991 and 1993 was followed by a demonstration experiment, limited to A-type experimental units, to determine the potential of on-top silver iodide seeding for the enhancement of area (1964 km2) rainfall. Analyses in a companion paper (Part I) established that the Thai cold-cloud demonstration experiment, evaluated according to its original design, failed to reach statistical significance in the time allotted to the experiment, although the probabilities that the seeding effects were positive on the treated cells and units are 72% and 79%, respectively. The results of exploratory examination of the entire demonstration experiment, including both A- and B-type experimental units, are presented herein. The exploratory studies involved both cell [392 seeded (S) and 335 nonseeded (NS)] and unit (35 S and 35 NS) analyses, a bivariate analysis of the joint effects on cells and units, and the analysis of pooled results from the exploratory experiment and the entire demonstration experiment. The results of these exploratory studies strengthen the case for seeding-induced changes in rainfall that were indicated in the evaluation of the a priori demonstration experiment. A multiple regression analysis to account for some of the natural rainfall variability suggests, however, that the apparent seeding effect has been overestimated by about a factor of 2 (i.e., +92% versus +48%). Temporal plots and analyses of unit rain-volume rates and cumulative rain volumes for seeding effects revealed stronger statistical support for convective masses within the unit not having seeded ancestry, as determined by radar, than for convective clusters with seeded ancestry. This result suggests that the effect of seeding, which begins with the directly treated cells, is propagated to nonseeded clouds within the unit. Enhanced downdrafts and/or ?secondary seeding,? as discussed herein, are posited as possible propagation mechanisms. Partitioning of the data by a crude aircraft measure of coalescence intensity revealed that the rain volume from NS units increased as coalescence intensity increased, whereas the greatest mean S rainfall was observed in the moderate coalescence category. The apparent seeding effects were >100% for units having clouds with weak to moderate coalescence and were nonexistent for units having clouds with strong coalescence. This was true also upon analysis of the cell sample. The implications of this and all results are discussed in the context of the conceptual model guiding the experimentation.