WMOD00318 (2) - Laserfiche WebLink

Home Browse Search

840 JOURNAL OF APPLIED METEOROLOGY VOLUME 34 ...............---.-.--.-.... --......-........---...... I I ! ..\" '. , , , . 'riL . '\, " " FIG. 4. Grid configuration of 3D nested-grid model. Grid I is the largest hatched grid (~ = ~y = 16 km), with grid 2 being the next largest hatched region (~ = ~y = 4 km), and grid 3 being the dark shaded region (~= ~y = I km), tion more as parcel temperature lowered. This effect would have been greatly reduced if horizontal disper- sion was considered in the simulation as it was in the RAMS simulations presented in the next section. Im- mersion freezing is an insignificant contributor to ice formation for the conditions simulated. Considering this factor and the high overhead of implementing such a process in larger-scale simulations, it was omitted as a potential ice formation process in the mesoscale sim- ulations. While the parcel model simulations cannot ade- quately simulate the seeding situation, they do indicate that the concentrations of ice crystals formed based on laboratory results are in line with those inferred to have formed from seeding (Deshler et al. 1990). Namely, the ice crystal concentrations in the seeded plume are predicted to be between 10 to more than 100 L -I . 4. Mesoscale model simulations Mesoscale model simulations were done using a three-dimensional version of RAMS. Preliminary 2D sensitivity tests (not shown) demonstrated the impor- tance of releasing the seeding material in the correct location to optimize artificial nucleation processes. One problem found with the 2D simulations was that the SL W was very transient and the resultant precipitation was increased only when the seeded material was re- leased in a region of SLW. Use of the 3D version of the model allowed for a more realistic depiction of the spatial variability of the liquid water. These three-di- mensional simulations also allowed a variable initial- ization that alleviates some of the problems inherent to horizontally homogeneous initializations. a. Model description The numerical model used in this study was a version of the RAMS cloud model developed at CSU (Tripoli and Cotton 1982, 1989; Cotton et al. 1982, 1986). RAMS was configured using the nonhydrostatic, fully compressible momentum equations; a thermodynamic energy equation; and equations for liquid- and ice- phase precipitation processes. The predicted variables included the three velocity components; the Exner function 71"; the ice-liquid water potential temperature Oil (Tripoli and Cotton 1981); pristine ice crystal con- centrations; and mixing ratio of total water, rain, pris- tine ice crystals, graupel particles, and aggregates (Cot- ton et al. 1986). Note that pristine ice crystals in the Cotton et al. ( 1986) parameterization represents freshly nucleated, vapor-grown ice crystals. Ice crystals formed by seeding, while not being pristine in the purest sense, are placed in the pristine ice crystal category of the model. Potential temperature, temperature, cloud droplet mixing ratio, water vapor mixing ratio, and pressure are calculated diagnostically (Tripoli and Cotton 1982), Horizontal and vertical turbulence are parameterized using an eddy viscosity closure scheme, as described by Tripoli and Cotton ( 1982). The equa- tions are integrated numerically by a time-splitting procedure for a nonhydrostatic, compressible system (Tripoli and Cotton 1982). A Klemp and Wilhelmson ( 1978) radiative-type lateral boundary condition was used. A terrain-following sigma-z vertical coordinate system was used following Gal-Chen and Sommerville (l975a,b). A comprehensive overview of the micro- physics model was given in Flatau et al. (1989). Natural primary ice formation in RAMS followed from two models that quantify four ice formation mechanisms. The deposition / condensation-freezing model of Mey- ers et al. ( 1992) was used to describe nucleation from the vapor state below and above water saturation. This model is based on data from continuous flow diffusion chambers relevant to both deposition nucleation and the sorption form of condensation freezing. It is pos- sible that this parameterization also accounts for the natural immersion freezing IN population. Ice crystal concentrations are specified as a function of ice super- saturation, which varies with temperature and vapor mixing ratio forecast by the model. Contact freezing was also quantified following Meyers et al. ( 1992). The total potential numbers of contact freezing nuclei ef- fective per liter of air is based on a compilation ofvar- ious laboratory data. The fraction of this potential ac- tivity realized in any time step is determined by the collection rates of ice nuclei by cloud droplets due to the combined effects of Brownian collection, thermo-