Explicit Forecasts of Winter Precipitation Using an Improved Bulk Microphysics Scheme. Part Ⅰ: Description and Sensitivity Analysis GREGORY THOMPSON, ROY M. RASMUSSEN, AND KEVIN MANNING February 2004 Monthly Weather Review 519~542

1.Introduction Explicit bin-resolving cloud models(Rasmussen et al 2002, hereafter RG) Bulk microphysics parameterization 利用Reisner et al.(hereafter RRB)作為修正基準 目的：改善顯示(explicit)的即時過冷水滴預報，藉以防止飛機積冰(aircraft icing)的現象發生 Note : 利用bulk model來替代explicit model，節省計算時間

2.Description of the bulk microphysics parameterization Initiation of cloud ice Autoconversion Treatment of graupel Intercept parameter of the snow size distribution Intercept parameter of the rain size distribution

Initiation of cloud ice Fletcher 1962 Cooper 1986 Meyers 1992

Note : 在最近的研究指出ice的形成Si >5%，T<-5℃

Autoconversion Kessler 1969 Qr is rain content,Db=0.3, Nb is cloud droplet concentration, Qc is the cloud water content Berry 1968 (Berry and Reinhardt 1974) υ is a gamma distribution parameter

Treatment of graupel 改進冰雹形成中兩個最大量： Rutledge and Hobbs 1983

Intercept parameter of snow size 因為固定的截斷參數(N0,S)使得過多的雲水被消耗 發展一個跟混和比和single moment相關的關係式(Swann 1998) 平均粒子直徑跟溫度一起增加，且N0,S和λ減小時，溫度也會增加(Houze et al. 1979) Houze et al 1979

灰粗線/虛線是利用Houze et al.1979 細實線/虛線是Reisner et al.1998

Intercept parameter of rain size distribution Droplet size drizzle drops rain (雨滴的形成過程) 在大部分的single moment scheme中，都是假設雨水突然由雲水形成而且呈現Marshall-Pamler size distibution and intercept parameter(8106(m-4)) 考慮drizzle to rain

粗實線是利用上頁公式畫出來的終端速度和雨混和比圖，虛線是利用N0,r= 8106(m- 4)做出來的圖 大雨滴 小雨滴 粗實線是利用上頁公式畫出來的終端速度和雨混和比圖，虛線是利用N0,r= 8106(m- 4)做出來的圖

3.Approach Mesoscale model (MM5) 2-D idealized configuration 120points, spacing 10km 39σp level, flow is 15ms-1 Bell-shaped mountain 1km high with 100km half-width

敏感度實驗的設定

4.Sensitivity tests

CONTROL experiment

b.Bin model Initial cloud ice是利用Cooper(1986)

c.Ice initial:Meyers and Fletcher experiment Fletcher experiment減少了cloud ice 在鋒值的數量 Meyers experiment增加了cloud ice 的數量

d.Autoconversion:Kessler1, Kessler5, Berry, and BandR Kessler1(qco=0.110-3) Kessler5(qco=0.5 10-3) Berry(1986) BandR(Berry and Reinhardt autoconversion 1974)

e.CCN spectra:Maritime, Continent1, and Continent2 experiment Maritime(Nc=50cm-3) Continent1(Nc=200cm-3) Continent2(Nc=500cm-3)

f.Treatment of graupel: Graupel1 and Graupel2 experiment Graupel1:original exponential distribution, No,g=4 106m-4, and Reisner rimed snow-to-graupel conversion Graupel2:Riming growth of snow must be 3 times larger then depositional growth before creating graupel

g.Version2 把Fletcher, Kessler5, Graupel1一起使用 可以得到一個非常接近Graupel1的結果，所以在這三組實驗中以Graupel1比較敏感

h.Snow intercept parameter:SON and SONV experiment SON(No,s=2 107m-4) SONV(No,s=f(T))

i.Rain intercept parameter:RON and RONV RON(No,r=11010m-4) RONV(No,r=f(qr)) Note : combine SNOV

j.Final experiment BandR, Graupel2, SONV, RONV一起使用

k.Deeper/colder cloud system CCT=-25℃ CCT=-60℃ Seeder and feeder system

CCT=-25℃

CCT=-60℃

Seeder and feeder system 上層都是由冰晶跟雪組成的溫度較低的雲 下層是混和雲 在經過兩小時之後上層的雲開始影響下層 在下層會有一些過冷毛雨被冰覆蓋落在地面

Conclusion Primary ice nucleation using Copper(1986), replaces the Fletcher(1962) curve Autoconversion using Walko et al.(1995), replaces the Kessler scheme A generalized gamma distribution for graupel (eq. 8) replaces the exponential distribution; the intercept parameter depends on mixing ratio (eq. 10) instead of constant; and riming growth of snow must exceed depositional growth of snow by a factor 3 before rimed snow transfers graupel The intercept parameter of snow size distribution depends on temperature replacing the mixing ratio dependency The intercept parameter for rain size distribution depend on mixing ratio, thereby simulating the fall velocity of drizzle drops as well as raindrops