WMOD00318 (2) - Laserfiche WebLink

Home Browse Search

APRIL 1995 MEYERS ET AL. 835 program. A common scheme (e.g., Plooster and Fukuta 1975) has been to use the laboratory-measured ice nu- cleus effectiveness spectra (ice 'crystals per gram re- leased), if available, to specify the number of ice crystals formed, with no regard to mechanism or a consider- ation of the potential time required for nucleation. This represents the simplest scheme for use in mesoscale models. Other researchers have sought to increase the complexity of implementing ice initiation. Young ( 1974b) included two mechanisms, deposition and contact nucleation (via Brownian and phoretic trans- port processes), in a detailed microphysical cloud model. Only contact-freezing nucleation was used in published nonspecific seeding simulations (Young 1974a), and the nucleus activity spectrum employed was not representative of most that have been measured for field generators. Hsie et al. ( 1980) simulated both contact-freezing and deposition nucleation in a study of seeding effects in continental cumulus clouds using a 20 time-dependent model. The nucleation mecha- nisms were applied to the laboratory measured effec- tiveness spectra of a nucleus for which the ice formation mechanisms were not known. The relative contribution to ice crystal formation by each mechanism was based on the theoretical work of Cooper ( 1974), which has not been directly verified. None of the above-men- tioned models explicitly included condensation-freez- ing nucleation. The nature of ice-nucleation mechanisms, rates, and activities can directly and indirectly influence the na- ture of numerical cloud model predictions of seeding effects. Orville et al. ( 1984) found unexpected dynamic effects in AgI seeding simulations of stratus clouds. This clearly related to the presumed nucleation activity and rates of ice crystal formation (instantaneous). Orville et al. noted the potential inadequacy of the nucleation scheme employed and its importance in the realization of quantitative results. Blumenstein et al. ( 1987) de- veloped time- and temperature-dependent empirical expressions from experimental data obtained in the CSU isothermal cloud chamber and applied these in Rauber's ( 1981 ) 20 orographic cloud model. Her lab- oratory study demonstrated a slow ( up to 100 min for completion) condensation-freezing nucleation process by AgI-Nal aerosols at near water saturation and a fast (within seconds) process when unspecified water su- persaturations were induced. Also, nucleation activity increased in the supersaturated case. Assumptions of either the slow or fast nucleation process in Blumen- stein's simulations yielded large differences in the amount and targeting of seeded snowfall. Lacking from this study was a more detailed knowledge of the exact response of the aerosols to differing magnitudes of su- persaturation. The numerical study presented in this paper takes advantage of the recent laboratory study ofa particular ice nucleus aerosol that was used in a well-documented cloud seeding case study. It was sought to implement the high level of detail of ice nucleus response to varied temperature, humidity, and cloud conditions obtained in the laboratory study into mesoscale simulations. The goal was to have the capability to predict both the pre- cipitation effects and observable cloud microphysical effects of seeding for comparison to observations. b. Case study The case selected for the modeling study was the 18 December 1986 SCPP case study that has been ex- amined by Deshler et al. (1990). The day was char- acterized by a weak split-front, where the upper-level cold surge crossed Sheridan, California, (located at the base of the Sierra foothills) at 1500 UTC preceding a surface cold front that crossed Sheridan between 1800 and 2100 UTe. This environment has been noted to be favorable for the local production of supercooled liquid water (SLW) (Reynolds and Kuciauskas 1988), and consequently, three experiments were conducted between 1800 and 2000 UTe. Seeding was conducted in the SLW region at approximately -60C in three 37- km seedlines. Most of the natural precipitation was orographically produced with peak precipitation rates observed at Kingvale, California, to be 1-1.5 mm h -1. Deshler et al. (1990) inferred a precipitation seeding signature of 0.1-0.5 mm. A range of hydrometeor species were observed dur- ing this case study including pristine ice crystals, rimed particles, aggregate snowfall, and rain. Figure 1, based on results reported by Deshler et al. (1990), depicts an idealized x-z section of the measured distribution of ice and liquid water species at 1730 UTC (before seeding). Large aggregates were observed over a wide region west of the Sierra crest at temperatures warmer than -lOoe. Rimed crystals and graupel were noted in more localized regions further up the barrier at this time. Pristine ice crystals predominated at higher al- titudes and directly over the mountain crest. Pristine ice crystal concentrations were observed to be less than 10 L -] (mean values), with maximum values as high as 20-.30 L -I during natural conditions. Ice crystal concentrations over 100 L -1 (peak values) were ob- served in the seeded cloud. The ice nucleus aerosols used for seeding were AgI- AgCl aerosols produced by combustion of acetone- based solutions of AgI (silver iodide), NH41 (ammo- nium iodide), and NH4Cl04 (ammonium perchlo- rate). The generation of AgI-AgCl aerosols in this manner has been described by DeMott et al. ( 1983), who noted optimization of ice nucleation efficiency when NH4Cl04 was used at the 30 mol % level with respect to AgI. Except for the use of 3% AgI by weight instead of 2%, the solutions burned in SCPP aircraft- mounted generators during this case study were the same ones used by DeMott et al. ( 1983) and DeMott ( 1994) in laboratory studies. Potential differences in particle size distributions (PSDs) generated at th lab-