WMOD00438 - Laserfiche WebLink

Home Browse Search

the graupel formation stage and even suggest it could be followed to the fallout of precipitation at cloud base. In moving up scale to larger clouds. one must proceed with caution. Very large clouds are often associated,with severe weather and present a greater operational hazard. which must be considered. The complexity of larger clouds would also make an accurate evaluation of seeding effects more difficult. Convective complexes larger than cumulus congestus but smaller than the typical mesoscale convective system would appear to be the best candidates for continued experimentation. In working with convective complexes. it will not be possible to use as simple a physical hypothesis as was used in HIPLEX-1 to relate dry ice seeding to additional rainfall below cloud base. However. H IPLEX-1 results provide confirmation of our ability to produce precipitation embryos in clouds through the addition of artificial nuclei and. therefore. serve as a foundation for future experiments. The key questions to be considered in such a renewed experi- ment would include the roles of both artificial and natural precipitation embryos. The larger clouds will permit some additional choices regarding seeding treatments. e.g.. seeding at the -1 5 0 C level versus seeding at the -8 0 C level. At first glance. it would appear that seeding as soon as the cloud tops rise above the 0 0 C level would provide the maximum time advantage to the artificial ice crystals. but growth rates and crystal habits have to be considered. The time advantage enjoyed by embryos formed from columns around -5 0 C. as compared to embryos growing from thick plates initiated near -12 0 C or dendritic crystals produced near -15 0 C. may be offset by a more rapid growth to the riming threshold. a greater riming efficiency for precipi- tation embryos grown from ice crystals originating at lower temperatures. or both. These considerations would apply to both natural and artificial embryos. Furthermore. embryo concen- trations and their impact upon LWC would have to be considered. Riming would be slowed or stopped if LWC were reduced by seeding. Unfortunately. the available physical models of solid precipitation embryo growth do not agree on the LWC required to support riming. Therefore. even if one assumed that riming growth would be preferable to aggrega- tion of ice crystals. the available data are not suffi- cient to specify the embryo concentration that could be introduced before their competition for the available cloud water would choke off their growth. Considering that the microphysical processes fake place in clouds with both organized and turbulent internal motions. it is seen that physical hypothesis to be tested could be very complicated. All of the above considerations cover only micro- physical effects in a single convective cell. It would also be necessary to consider physical transport of artificial and/or natural embryos from one convec- tive cell to another and the possibility of dynamic seeding effects. Possible dynamic effects in High Plains convective complexes include added buoyancy through latent heat release in seeded cells. and the effects of precipitation-induced downdrafts upon cells developing subsequently within the same convective complex. We foresee. then. a need to test a complex hypothe- sis of seeding effects involving the concentrations of the artificial and natural precipitation embryos as functions of space and time. their riming efficiencies. the supercooled water available in both seeded and unseeded clouds. and some dynamic effects. Testing such a hypothesis would tax existing numerical cloud models to their limits. Indeed. the limitations of the best current cloud models are such that they could provide only indications of what the seeding effects might be. Therefore. field experimentation remains the ultimate test for any hypothesis proposed. It would not be possible to evaluate such a complex experiment with complete success using physical studies alone. so it appears that some statistical design would have to be used. While the physical design of the experiment and data collection are real problems. there is little doubt that the main obstacle to success in such an experiment would lie in the great natural variability of convective clouds. This implies that the statistical design would have to include some covariates (predictor variables) to be used to predict what the seeded clouds would have done if left unseeded. For clouds of the size being considered. the most likely covariates would be drawn from mesoscale observations. especially the mesoscale moisture convergence and vertical moisture flux at low levels. HIPLEX results from Texas showed low-level moisture convergence can be a powerful predictor of convective shower devel- opment. The experience gained on such programs as PROFS (Prototype Regional Observing and Fore- casting Service). on the real-time combination of satellite cloud pictures. radar data. and observations from mesoscale surface stations suggest that an experiment could be developed to make real-time predictions of convective storm formation. Using the initial conditions to derive covariates for the statisti- cal analysis. it might be possible to obtain significant results with a reasonable number of experimental units (40 or 50). In a typical High Plains setting. obtaining 40 to 50 experimental units would take about three summers in the field. Without predictor variables. natural variability in convective rainfall would require such an experiment to be very long 2