Laserfiche WebLink
<br />I <br />I <br />I <br />I <br />I <br />I <br />I <br />I <br />I <br />I <br />I <br />I <br />I <br />I <br />I <br />I <br />I <br />I <br />I <br /> <br />One of the intriguing findings from recent bin <br />microphysical studies has been identifying the <br />importance of in-cloud turbulence in the onset of <br />coalescence growth (e.g., Khain and Pinsky, <br />1995). Hall (1980) developed an explicit bin <br />model for ice. Farley and Orville (1986) <br />developed a bin representation for precipitation <br />ice used by Farley (1987) to document the <br />importance of variable ice densities (and low- <br />density growth) in hailstorm growth. The past <br />decade or so has seen a dramatic increase in the <br />formulation of bin models with many size <br />categories (typically 20 - 35 size bins) for multiple <br />ice habits (e.g., Reisin et a/., 1996; Khain and <br />Sednev, 1996; Khvorostyanbv and Sassen, 1998; <br />Ovtchinnikov and Kogan, 2000), made possible by <br />the exponential growth in computer speed, <br />memory, and storage capabilities. These state-of- <br />the-art microphysical models are providing <br />exciting new insights into the complexities of cloud <br />microphysics and their interactions with the <br />dynamics and other physical processes within a <br />relatively idealized framework (e.g., Ovtchinnikov <br />et a!., 2000). It is not possible, however, to use <br />such expensive approaches within fully interactive <br />NWP models, and it may not be so for another <br />decade or more. <br /> <br />7. NWP MICROPHYSICAL SCHEMES <br /> <br />Before the 1990s, those operational NWP <br />model that included one or more prognostic <br />variables for cloud condensate were likely using a <br />variation of the Sundqvist (1978) and Sundqvist et <br />a/. (1989) formulation, which considers fractional <br />cloud coverage within a grid box 25-50 km in size. <br />Cloud condensation or ice deposition (depending <br />on temperature) is initiated once the relative <br />humidity exceeds a threshold value. An algorithm <br />is employed that partitions further moistening <br />between increasing cloud mixing ratios within the <br />cloudy portion of the grid box, and increasing the <br />relative humidity and fractional cloud coverage in <br />the cloud-free part of the grid box. Simplified <br />relationships are used to determine the <br />precipitation rate from cloud condensate (like a <br />sum of autoconversion and collection), <br />evaporation of cloud water and rain, and the <br />Bergeron freezing of cloud water to ice. An <br />expanded version of this. scheme that included <br />more microphysical interactions between cloud <br />water, cloud ice, and precipitation, similar in <br />design and motivated in spirit by LFO, was <br />devised by Zhao and Carr (1997, ZC) and <br />implemented into the 80-km Eta in 1995 (Zhao et <br /> <br />a/., 1997). Cloud water is assumed at <br />temperatures above OC or if the top of the cloud <br />layer is warmer than -15C, otherwise cloud ice is <br />assumed. <br /> <br />In these schemes, precipitation <br />instantaneously reaches the ground and is not <br />stored in the atmosphere. This is a reasonable <br />assumption for rain, but not for snow and for most <br />ice crystals. Furthermore, the threshold relative <br />humidity needs to be tuned at different model <br />resolutions, and it is not clear whether the <br />underlying assumptions of the moisture partitioning <br />make sense at resolutions finer than 20-25 km. <br />The ZC scheme was replaced with the <br />implementation of the 12-km Eta model in <br />November 2001 by a new prognostic cloud <br />scheme that attempts to emulate aspects of higher <br />resolution models with a modest increase in <br />computational expense (Rogers et a/., 2001; <br />Ferrier et al., 2002). The prognostic variable is <br />total condensate, and only it is advected. Within <br />the microphysics, arrays store the fractional <br />composition of cloud water, rain, and ice (snow <br />and cloud ice), where it is assumed that their <br />fractional contribution is unchanged during <br />advection. A simple, first-order "box" algorithm is <br />used to partition between the storage of <br />precipitation within the box and fall out through the <br />bottom. Assumptions are made concerning the <br />characteristic size of the precipitation ice particles, <br />which is assumed to vary as a function of <br />temperature (Ryan, 2000), but which can increase <br />in size when the number concentration exceeds an <br />upper limit. Mixed-phase processes are <br />calculated. The bulk density of ice is (crudely) <br />estimated by keeping track of the growth of the ice <br />through accretion of liquid water versus <br />depositional growth, such that their fall velocities <br />increase when the bulk densities increase (Bohm, <br />1989). Extensive lookup tables are used to store <br />solutions to complex calculations of different <br />moments of rain and ice distributions, from which <br />the microphysical rates can be calculated quickly <br />and accurately during the integration. Liquid water <br />freezes at temperatures colder than an assumed <br />threshold temperature. Assumptions are made <br />about the cloud ice field that circumvents the need <br />to calculate ice nucleation processes. <br /> <br />8. UNRESOLVED ISSUES IN MICROPHYSICAL <br />MODELS <br /> <br />Below is my list of the major unresolved issues <br />in microphysical modeling (not intended to be a <br /> <br />9 <br />