Visit the complete Western Waters Digital Library
 back to results : previous : next

 Handbook of conceptual cloud models for South Dakota
 Title Author Date of Original 1973 Object Type Physical Description illustrations Notes Restrictions This document is in the public domain. No copyright permissions are needed. Acknowledgement of the Arthur L. Rangno Cloud Seeding Collection as a source is requested. Geographical Coverage Collection Contributor Repository Ordering Information Reproductions are not available Transmission Data Scanned with a document scanner to tif. Converted to jpg, size of 1000 pixels wide by proportional height, sharpened and color channels adjusted for readability and contrast. OCR performed during upload into CONTENTdm. Document Number Full Text HANDa:x::c: OF COH::EPl'UAL ClOOD H)l)ELS FOR OOU'l'H DAmTA Prepared for OOJTH DAtt1l'A WEA:r~R eatrR>L COItiIOOICfi ty Robert D. Elliott NORl'H AMERICAN WEATHER OONSULTANl'S Ma.y 1, 1973 TABLE OF 00NTEN1'S I CONVECTION IN GENERAL • • • • • • • • • • • • • • 1.1 Convection in a Conditionally Unstable Air Mass. 1. 2 The Develop1leJlt of Precipitation in Convection • 1.3 Fa.ctors Governing the Type and Degree of Culm.Ilus Developnent . . • . • • • • • . • • • • • • • • ~ 1.l~l 1.1~1 1,2-1 1.3~1 1.4 Ldrge Cumulus Cloud Dimensions, Movements, and Structure 1,4.1 1. S Cloud J::ineJUdtics • • • . . • • 1. S~l 1.6 Precipitation Efficiency • • . • 1.6_1 1, 7 Stages of Cumulus Developnent . • 1, 7~l 1. 8 Application of Numerical Models . 1.9 Some Cloud Seeding Results 1. 9_1 Chapter I _ &mmary References II 2.1 ,., ,'..3. '.5 OOU!'H DAn:trA CUMULUS Cumulus Climatology of South Dakota • Cloud Catcher Observations • . ••• Hail Suppression • • . • • . • • • • Conceptual Cloud Models for Seeding • Seeding Hodes • • • • Chapter II - SUnrnary References 2.1.1 2.1~1 2.2_1 2.3_1 2.4_1 2.S~1 1.1.1 Ldpse rate and cloud diagram • • • • • • • • • • • • • • 1.1~2 1.1.2 Concepts of cumulus growth • • • • • • • • • • • • • • • 1.1-4 1.1-3 Sounding parMleters discussed in text in connection with Figure 1.1.4 •• • • • . • • • • . . • • • • • . • • . 1.1~6 1.1-4 Types of soundings and associated profiles of upiraft and cloud water content • • • • . • . • • • . . • . • • • • 1.1 ~7 1.2-1 Flow diagram of primary cumulus cloud physics states and processes 1.2_2 1.2_7 1. 2-2 Oooerved concentration of ice forming nuclei . . • • • • • • 1. 2_5 1.2_3 Typical curves showing the number of effective ice fonning nuclei per gT~ at various activation temperatures for four different nucleating agents • • . • • • • • 1. 2_4 Ratio of ice crystal concentrations to ice nuclei concentrations versus temperature ••• • • • • 1.2_9 1.2_5 Variation of Fletcher's three_dimensional diagram showing the relation I:e~en condensation, freezing, and sublimation as a function of saturation ratio, temJ:)eratures, and particle size • • • • • • • • • • • • • • • • • • • . 1.2-10 1.2-6 Ta-s diagram showing the oonditions of fOIllloation of varioUB shapes of snow crystals • . . 1.2-12 1.2_7 Vertical velocity for hail case .•.. . . . . . . . • • • 1.2-14 1. 2_8 Rainwater for hail cue . . • • • • • • • • . . . . . . • • 1. 2_14 1.2-9 Particle distribution in naturally and JIg! nucleated clouds 1.2-15 1. 2_10 Particle distribution in naturally and JlgI nucleated clou:ls (with dyndtnic effects) • • • . • . • • • • • • • • • • • 1.2-18 1.3-1 Schematic soundings illustrating the influence of various atmospheric processes on the alteration of the air. • • 1.3_2 1.3_2 Typical soundings and related cumulus d.evelopnent •••• 1.3-3 1.3_3 Schematic illustration of processes in formation of a deep unstable layer by differential advection. • • • • • • • • 1.3-5 1.4-1 Cloud ge~try • • • • • • • • • • • • • • • • • • • • • • • 1.4_2 1.4_2 Hydrodynamic pressure on surface of cylindrical convection cell and resultant superimposed updraft on right side of cell •• • • • • • • • • • . . . • • • • • • • • • • • 1.4_4 1.4_3 Schematic diagram showing sequential dewlOiZOOnt of diverging echo couplets •• • • • • • • • • • • • • • • • 1.4_6 1.4_4 The twisting and stretching of a vortex filament. • 1.4_7 1.4_5 Features of a squall cloud. • • • • • • • • • • • • 1.4-8 1.4_6 Tracks of individual rainstorms and squall lines of May 20_21, 1949 • • • • • • • • • . . • . • • • • 1.4-10 1.4_7 Features of II nocturnal thunderst:orm . • . • . • • • 1.4-11 1. 5_1 Updraft traject:ory and cloud outline in horizontal and vertical planes produced from a single fbed source 1. 5_2 ii 1.5_2 Plan view of cloud outline based upon Figure 1. 5_1 assumptions but for a discrete IllOving source producing new towers 1. 2. 3. etc. at positions I, 2, 3, etc•• at successive times I, 2, 3. etc. ••••••••.••• 1. 5_3 1. 5_3 Cloud profiles for various types of cloud systems under deep w.um and deep cold advection synoPtic situations. 1.5_4 1.6_1 Scatter diagr4llJ,!) of wind shear versus precipitation efficiency for 14 thunderstorms which occurred on the High Plains of North America 1, 6_4 1.7_1 Illustrating pulsation (discrete) variations •• • • • • 1, 7_2 1. 7_2 Sequences of daily events in convection. • • . • • . • . 1. 7_5 1.9-1 !'klan rainfall of seeded clouds, control clouds, and mean difference in each lO_minute interval relative to seeding time • . • . . . . . . . . • • • . 1. 9-2 2.1-1 Typical cumulus type soundings • • • . • • • • . . . 2.1_2 2.1_2 Wind roses of Rapid City for various levels. • • • • 2.1_3 2.1_3 Monthly normals of precipitation and annual trend of precipitable water • • . • • • • . • • • • • . • • 2.1w4 2.1_4 Annual precipitation (inches) • • • • • • • • • • • • 2.1_5 2.1.5 Average growing season (AprU-8eptember) precipitation (inches) 2.1_5 2.1_6 lbur (local mean time) of maximum frequency of thundersto:nne, June_August .•••••••••••••••.••••• 2.1_7 2.1_7 Isochronel5 of peak hourly precipitation during am:mer in South Dakota • • . . . • • • • . • • • • • • • . • • • . 2.1-8 2.1_8 Isochrones of secondary peak in hourly precipitation during summer in South Dakota • • • • • • • • • • • • • • • • • • 2.1_9 2.2_1 Scatter dh.gram comparing logarithm of radar_estimated rainfall and logarithm of cloud depth for no_seed, silver iodide seed, and salt seed cases. . • • • • • • • • • • • • • • 2.2_2 2.2_2 Dynamic effects in a precipitation-eloud diagram • . 2.2_3 2.2_3 Radar volume precipitation in acre_feet according to multiple regression equation • . • • . • • • • • . 2.2_6 2.2_4 Cloud activity when the low center 1llOV'88 eastward to the north of South Dakota . • • . . iii .• 2.2-13 2.2~5 Cloud activity when the low center IOOveS eastward to the south of South Dakota. . • • • • . • • • . • • • • • • 2.2..13 2.3~1 Height distribution for hailers, non.hailers, and entire sample •• • • • • • • . • • • . • • . • . . . • • • • 2.3_2 2.4.1 Stages in transient cloud, steady state back feeders, and steady state front feeders ••• • • • • • • • • • • • • 2.4_2 2.5_1 Illustration of seeding methods used for various types of clouds •.. • . • . • • • • • • • • • • • . • • • • • • 2.5.2 1.6_1 Values of storm JOOisture .flux, storm duration, total water, and storm efficiencies for four major stonns . • • • • 1. 6.2 2.2_1 Summary of wind, cloud code, and seeding characteristics associated with advection typ9 and cloud height 2.2_8 2.2.2 Summary of wind, cloud code, and seeding characteristics associated with advection type . • • • • . • . • . • • 2.2_10 2.2_3 Summary of Wind and seeding characteristics associated with cloud codes •••••.•••.•....•.•••• 2.2 _12 2.2_4 Average vertical wind shear for various radar cloud top ranges iv 2.2_14 CHAPI'ER I • COOVECTION IN GENERAL 1.1 Convection in a Conditionally Unstable Air Mass Culwlus clouds are ClOst likely to develop in an air mass having ~ lapse rate lying between the wet and dry &diabatic temperature values extending through a thick layer. Figure 1.1.1 illustrates the typical situation'as laid out on a height temperature diagram. The air mass 'is stable al::ove cloud base with respect to the dry process,. but unstable with respect to the wet process. Any small upward motion lifting air which is saturated up al::ove the cloud b4se can initiate convection, even though the air al::ove and. l:elow this wet zone is dry. The environmental air mass is said to be conditionally, or convactively, unstable. In the classical parcel method, a saturated air parcel at cloud base rises unmixed upward along the wet adiabatio (heavy dashed line in the figure). The temperature of this parcel is higher than that of the environment (the heavy solid line), therefore the parcel is rooyant and acoelerates upward through the positive .'lIea (area where the parcel temperature ex­oeeds the environmental temperature, marked + in the figure). The upward momentum propels the parcel al::ove the positive area and. into the negative area (where the parcel temperature is less than the environmental temperature, marked • in the figure). Here deceleration occurs and. the parcel finally OOlDeS to rest at the level when the two areas (as represented on il T.; diagram) billance one another. If the top of il developing cloud behaved ilS does this parcel, the whole cloud would look like the cloud-l outline. This height is seldom ilttilined. The parcel llSthod ilSSumes that all of the potential energy of the conditionally unstable air IDil8S is converted directly into the kinetic energy of upw,ud motion of the parcel. whereilS considerable energy is ilctU41ly dissipated in interactions with the environment. As the real cloud rises it continuously mixes 80me of its environment into itself, il process cililed entrilinment. The excess of temperature, upwilrd momentum, and con· densilte with the rising parcel is thus reduced towilrd that of the environment. In the case of an isolated cumulus cloud, the ascent temperature curve in its heart resembles that shown by the dotted line. 01000_2 represents its outline. Notice that the anvil starts to spread in the negative area. In this area the cloud r:e_ comes negiltively buoyant, decelerates, and spreads laterally, The concept. of erosion of cumulus towers through entrainment. of drier environ_ mental air had been accepted for a decade or more before quantitative calculations of entrainment based upon field observations were llldde by StOi.1rJel (1947, 1951), 1.1-1 45 , 40 II 35 anvil of cloud 2 30 25 HEIGHT !.,."..' (K ft.) '" 20 ,., 15 ( ICLOUD 21 10 5 AM -60 -50 -40 -30 -20 -10 o +10 +20 +30 2-+------.-------.-----.----.-----.-----.-----.-----.-----.-----.-~ -70 TEMPERATURE (e) Fig. 1.1-1 Lapse rate and cloud diagram. The heavy solid, ddShed, and dotted lines represent temperature lapse rates on the temperature height diagram. The lightly sketched cloud outlines use the SillIE vertical scale but un arbitrary horizontal scale. Austin (1948), Austin .mel. Fleisher (1948), and }bughton and Cramer (1951). The most frequently observed pattern of fully developed oonvection is more likely to resemble that shown in Figure 1.1-2(a). The arrows pointing into the sides repre_ sent entrainment. The upward pointing arrows indicate updrafts. There is upward acceleration of air through much of the cloud depth. Near the top there is de_ celeration and rapid lateral expansion, althoUl;Jh entrainment may also continue. &aller oonveetion cells have a limited life and resemble a puff more than a oontinuous jet. Scorer and Ludlam (1953) called this bubble oonvection and develop­ed the b.1bbl.e theory of oonvection. Laboratory experiments were conduced by Scorer and Ronne (1956), Scorer (1957), Woodward (1959), and Saunders (1962) in which oon­vection bubbles ware simulated in tanks. Turner (1965) simulated dIl evaporating cloud in the laboratory. Turner (1957) and Levine (1959) emphasized the vortex ring type circulation within the rising" blbble as illustrated by Figure 1.1.2(b). The attached stem dangling l':eneath gives it the appearance of a jet combined with a vortex ring. Environmental air is brought in from the sides. The vortex ring expands aloft at a oonstant angle as sho-.m. by the dashed lines. Simpson and Wiggert (1969) consider the cap of an updraft inside a maritime cumulus tower to be roughly spherical in shape and to have an internal vortical circulation as shown in Figure 1.1_2(c). In "one-dimensional'" numerical convection IIlOdels, the type presently most often used for guidance in weather modification, the dr paroel is usually repre­sented as an elemental circular disc of finite radius which rises in finite steps, with a given amount of entrainment in each step, usually expressed in terms of the fractiondl increase in the total mass of the IJ.pdraft parcel per unit of rise. The entrainment is oo:oputed at each step, and is assumed to be inversely propor_ tional to the radius of the updraft, a relationship which has been established by laboratory experiment. Numerical 1llOde1s by Ha1tiner (1959), MAson and £mig (1961), and Squires and. Turner (1962) produoe realistic results. The starting plume concept by Turner (1962) was employed by Davis (1966) as the basic for a one-dimensiondl (vertical) 1llOde1 which can be used effectively in the field both for the control and the evdluation of seeding experiments. The Experimental Meteorology Branch (EMS) one­dimensional model was developed by Simpson and Wiggert (1969) also for use in field work. The practical model ltlOst relevant to northern Great PlailU:i cumuli has been developed by Hirsch (1973), and will be discussed in more detail below. 1.1_3 40 .,........~-------..,...--------::::::oo-- 30 10 (a) 20 ,, I I - I ~ .... 15 :r C) ":r' 10 (b) (c) .... :r !:! ":r' 20 Fig. 1.1~2 Concepts of cumulus growth. The light lines represent idealized cloud boundaries drawn to fit the structure of the cumulus lllOdels as represented by heavy lines and arrows. Numerical experiments (Malkus and Witt~ 1949; Ogura, 1962~ 1963; Murray and Anderson~ 1965) using equations of motion containing eddy viscosity terms and applied to a two-dimensional grid have yielded time dependent pictures of convec_ tion which emphasize its mushroom character as in Figure 1.1-2(bl. Orville (1965) has included surface heating ani parameterized cloud physics in an elaborate two­dimensional model for convection over a heated mountain slope. His model shows clear~y the mushroom character of the cloud rising owr the mountain peak. When he introduces vertical wind 5hear~ the convection cell is tilted downwind, and eventUAlly breaks away from. the mountain~ with a new cell forming at a later time. Physical observations of updraft MId liquid water content~ plus ttw results of convection model rurus~ show that sounding types can be related to cloud types. Figure 1.1_3 shows some key sounding parameters which relate to such cloud features as the vertical profiles of updraft and cloud water content. Figure 1.1-4 presents various types of sOW"ldings and associated updraft and liquid water content profiles. In the ~r left hand corner is d case havirg a large llT. The maximum updraft is 30 mfs~ a high value. The peak L'lC is 4. gfbiJ. The top reached 50 Ut. In the upper right hand corner is the SIlDle temperature soW'lding with a large saturation deficit in the environment. The LW:: is much reduced due to entrainment of drier air. Although there is less water~ the updraft is greater and the top rises higher. The weight of any liquid or solid water in an updraft produced. a negative bloyancy. In this case there is less liquid 'lfater~ therefore a greater net bloyancy. ACcordirg to the entrainn:ent modal~ the ma.gn1tude of llT h48 no bearing upon the liquid water va1ues along the profile~ only the environmental tenq::erature, humidity~ and the entrainment rate ma.tter.· However~ the greater t:.T the higher the top rises because of the added bJoyancy. The middle figure on the right is a case haviTl;l a small tlT~ extending through a good depth~ and having a small saturation deficit. The updraft is less and the top lower than for those cases already discussed~ wt the LW::: is higher. In genera1~ the twight to which the cloud top will penetrate is quite sensitive to the shape of the tlT and SO· profiles. A slight inversion may stop ascent~ even though there is a considerable positive area just above this level. In fact~ the added b.1oyancy produced by the release of heat of fusion in cloud seeding can provide a '"break_ through- as will be discussed later. The middle and lower figures on the left hand side show how the shape of the up::lraft profile is altered when the distribution of tlT with height is varied from 1.1-5 50 40 ~ " I- 30 :I: ~ ':"I: 20 10 -50 -40 -30 -20 -10 0 .,0 T - TEMPERATURE ( CI I i I I 0 10 20 30 U - UPDRAFT SPEED ( M/sec: I I I I I 0 2 4 6 LWC - LIQUID WATER CONTENT (9/ k91 OF CLOUO Fig. 1.1-3 Sounding parameters discussed in text in connection with Fig. 1.1-4, Tw is the wet adiabatic curve starting from cloud base at 10 lCft. Ie is the enviroI'llDElntaJ. temperature curve. DP is the dewpoint curve. toT is the temperature difference between Tw and Te. 3D is the maximum tempera_ ture dewpoint difference4' expressed as (I, satura.tion deficit (g/J:g). 1.1..Q .. (ICi ..1 1M II.) STRONG INSTA8ILITY. WET ENVIRONMENT lIT /' t EffiCIENT t TYPE ../ '-- Fill. 1.1-4 'I'ypo<: of $0"'1:11 1.t-.l profiles of updr.ft _ <'loud ~t.. -.'...,1. iioofftr to Fi'l. 1.1_2 tor ~ _ ."..le... t)J / ---v---. STRONG ~ INSTA81L1TY. / /-'"\ 0'" ,.,,"'""' WEAK INSTABII.lTY ~I , 'NEAK INSTABILITY, ,I: SMALL UPORAfT RAOWS 4':. that shown in the top left h4nd oorner figures. The updrAft profile shown in the middle figure results in an efficient rain ·producing cloud while that of the lower figure results in an inefficient producer. In the latter case a larger portion of the cloud w",ter IOOveS into the anvil and evaporates than in the fOImElr case. In general, upiraft speed, height of the top, and water content, are all en~ hanced by a large tiT, and reduced by a large entrainment rate. The rate of entrainment is inversely proportional to the updn.ft ndius. The consequences of this is illustrated in the lower right hand diagram in Figure 1.1~4. The l!lou:nd.ing is identical to that in the sketch ab)ve, yet because the up:lraft radiw: is half of that above, the cloud top is much lower, as is also the peak updraft and the peak LW::::::. The cloud physics processes which lead to precipitation have effects upon the updrafts dyn4llLics. As the clooo water is converted to ice, the heat of fusion in­cre~ es the cloud temperature, and. ite buoyancy. In the rather delicate balance between the drag or weight of the suspended liquid cloud. water and the thermal buoy~ ancy, a few tenths of a degree increase in the virtual temperature of the updraft can mean meters per second increase in the peak updraft speed, and can sometimes result in several thousnad feet incre~e in cloud top. This leads to more conden­sation, and therefore to IOOre precipitation. Another factor influencing updrafts is the disposition of precipitation. If the precipitation products fall direotly down through the updraft, they tend to accumulate for some time just ab)ve the level where the updraft speed is '" maximum. The weight of these accumul.ated particles in turn can suppress the updraft. On the other hand, if some or all of the particles fall outside of the updraft, then its buoyancy is enhanced and the updraft is in. creased. The effect of silver iodide seeding on euntulus cloud dynamics is the "result of two factors: 1) the release of heat of sublimation and 2) the rapid conversion of cloud. droplets to precipitation particles. The first factor always adds buoyancy to the updraft. The second may also do so where the precipitation particles fall outside of the updraft. These factors can be modeled numerically, as will be dis_ cussed later. 1."1-8 1. 2 The Dewlopnant of Precipitation in Convection The oloud physics processes leading to various types of precipitation are quite involved, and will be covered in an introductory ~er only. Figure 1.2_1 diagrams: the more important cloud physics states and processes involved in the transfo:rtM.tion of water vapor to' rain reaching ground level. At the start, in the lower left hand side, an air parcel enters the cloud in an updraft. The par_ cel contains d populdtion of condensation nuclei. The condensation nuclei are largely hygroscopic sea salt or sulfate particles. Their source is 1:JJbbles over the ocean surface or wind-blown loose so11 particles, or industrial emissions. The nuclei tend to fom a water hull, and condensation occurs upon them when the air beCOlllelS supersaturated as in up:I.rafts. The distribution of condensation nuclei by size, or the size spectrum, varies with geography. The spectra tend to fall into two main types: the oceanic and the continental. The overall concentration of continental condensation nuclei is about three tines that of the maritime nuclei up to 5000 feet above the surface, with this contrast decreasing upwards, according to Squires and Tocuaey (l966). Since only a slight supersaturation Iuds inmediately to condensation, the cloud base is very close to where the lifted air o.chieves 1001. relative humidity. This is due to the fact that the higher vapor pressure over the small droplets (higher than over a flat wdter surface) is counteracted by the lower­ing of the vapor pressure due to the presence of salts in the droplet. Condensa_ tion is indicated in the second box up. Cloud droplets of dbout ten micron size are grown quite rapidly, but essentially no larger ones are grown unless there are present a few oversize nuclei. The fourth box up on the left side is labeled_coalescence. This refers to the process whereby the smaller cloud droplets collide and coo.lesC8 with the larger droplet6, thus ledding to fewer and larger dro}:lS. The. process depends upOn there being differences in droplet size and therefore in tenninal velocity. This process is more active in maritime than continental clouds becaw;e their nuclei spectra is of the type favoring it. The drops fonned in this manner achieve a 50 micron radius. At this size they have a sufficiently large terminal velooity to fall as dri'lizle from a cloud, but not down through an updraft. Given sufficient time in a high water content cloud, rain drops will form by this process. The term hydrometeor 48 used here refers to these precipitation particles, but not to cloud droplets. In most models for the seeding of cumulus, use is made of the C041escence clooo hydrometeor generation process, these hydrometeors being later subjected to freezing. I:essler (1967) studied the possible interactions between convective type 1..2.-1 Fi~. 1. 2-1 Flow diagram of prilnaxy cumulus cloud physics states (shown in double lined bo:ms) and prooesses (single lined boxes). 1.2_2 motions and cloud physics mechanisns. He developed 4 parameterized version of the conversion of cloud water to hydrometeor water by collision dJld coalescence, based upon the principle that when the water content of the cloud reached a certain level, conversion would oonmence dIld continue at a rate proportional to the excess of cloud water content above this critical value. Srivastava (1967) applied Kessler's parameterization to the study of cumuli. Berry (1968) performed a theo­retical study of the coalescence process and ·produeed a practical parameterized equation for growth by coalescence Which includes the type of condensation nuclei distribution. In developing a practical IIIDdeI for use in field tests, it is important to express complex processes in tenns of a few observable parameters. Davis (1966) and Weinstein (1970) made use of [essler's parameterized hydrometeor development in their convection models. Simpson and Wiggert (1969) also applied Kessler and Berry's work to the EMS model. The inputs to the hydrometeor growth equations in both cases included the cloud liquid water content. Figure 1.2_1 diagram shows an arrow (marked 1.) pointing to the right from. the small hydrometeor box to the .sccretional growth box. Accretion technically refers to the sweeping out and collection of cloud droplets by 41\ ice crystal; however, it is sometimes used to 1nd1cate growth by collection of additional mass by falling rain drops. In the all warm precipitation process depicted in this branch, the rain drops may grow to sufficient size (about 0.7 em. diameter) to break up .sero­dynamically and form smaller drops. This process is indicated by the box marked r41ndrop shattering. If this occurs in an updraft, the sma11er drops may 4Soend for awhile, and grow by accretion. The process may then be repeated. This is the esaen08 of Langmuir's (1948) warm cloud chain reaction. R4in drop observations indic.ste that the conoentration of drops in different diameter classes decreases exponentially with drop diorneter. The concentration is also related to rainfall intensity; Marshall and P.slmer (1948) give the relationship: where N is the drop concentration in mUllber per lQ3 per lI'Il\ di&lll'lter range, D is the ch.ss diameter in nm, J.. is 41RO. 21 rmn-l where R is the rainfall rate in lDlIl per hour, and No is the intersect of the curve with the zero side axis. Ey employing a characteristic distrib.1tion fitted to the MarshaJ,l~Pa1mBr equa­tion, and a simplified relationship between rain drop terminal velocity ~ the 1.2_3 square root of the drop diameter, and assumptions concerning the representative­ness of the mean diameter, I:essler worked out a par&lleterized version of the accrea tion process which has been employed in practical c:umul.us cloud seeding models. The accretion rate is proportional to the product of the liquid cloud water content and the hydroaeteor water content to the .875 power. Both of these quan­tities can be measured by aerial sampling techniques and appear dB terms in computer models. In addition, the hydrometeor content can be measured by radar. It can be shown by silnilar arguments to those above that the radar return is related to the 1. 6 power of the hydrometeor water content. In the models employed by Davis, Weinstein, Simpson, and Hirsch, the ice pro_ cess is introduced by freezing all of the accumulated hydrometeor water at some specific glaciation level, or through a narrow temperature ...one. For example, it might be made to occur :tetween -SC and -lSC for silver iodide seeding, and _2OC and -3OC for natural seeding. Physically, such a process first involves some type of nucleation, but for purposes of computations, a glaciation temperature is em­ployed, one for natural nucleation, and mother warmer one for artificial nuclea_ tion. Under these assuraptions, the particles follow the path to the right of the nucleation box lll4I"ked 2. The frozen hydrometeors continue growing by accretion, becoming graupel or even hail particles. It is possible for some or all of the nucleation to produce small ioe crystals when then grow rapidly by the diffusional (sublimational) process. This occurs along path 3 in Figure 1. 2_1. The equation for crystal growth by vilpor diffusion includes the local saturation vapor pressure over ice and factors related to crys_ tal form and size. Within a supercooled cloud, the difference between the vapor pressure in the cloud and that over ice is greatest at -lSC. Therefore, diffusion­al growth is roost rapid ther~. The measurements of the conoentrations of natural ioe nuclei are made in a variety of ways, all of which give data in "general agreement. Some typical observ_ ed ranges of concentrations are shown in Figure 1.2-2 from Mossop and Ono (1969). An exponential average curve is also indicated. Although the majority of ice nuclei are grains of platey silicate minerals of earth origin according to Kumai (1961), it is recognized that other sources include man_made as well as extraterrestrial origin, Rosinski (1967). Most observations are based upon counts of small ice crystals falling to the floor of cold boxas in which samples of outside air were introduced into 4 cloud 10 10 0: .'.".. :::; 0: 10 .'". dz LIMITS OF MEASURED ICE NUCLEUS CONCENTRATION -5 -10 -IS -20 -25 TEMPERATURE 1°C) -30 Pig. 1.2_2 Observed cxmcentation of ice forming nuclei. 1.2_5 of supercooled droplets. Sometimes the crystals are grown in solutions of various types once they have f.ulen to the bottom in order to make them more visible. The box temperature is varied with different samples to determine the number of effec_ tive nuclei at various temperatures. The same procedure is being used extensively in testing the nucleating power of various types of artificial nuclei. Figure 1. 2-3 is d so-called nuclei production curve .mel contdins typical curves for dif­ferent nuclei souroes. The ordindte is the number of effective nuclei produced per gum. of Ag! smoke produced. The abscissa is the temperature in degrees celsius. The equivalent curve for natural nuclei ill' also shown. This is based upon the assumption that we are lIldking ... comparison with utificial nuclei in a ooncentrd­tion of 10-13 grams per liter, a vdlue in line for well diluted silver iodide smoke plumes. From the general shdpes of the curves one would judge that nature's defi. cit in ice forming nuclei oocurs at the warmer temperature. At temperatures colder than about .25C there is a high natural production of ice crystals. In fact, seed. ing of stable clouds colder than about _25C can create an excess of ice crystals which fail to grow to precipitation size, and thus leAd to a decrease in precipitation. Contact nucleation, the first sub-heading listed in the nucleation box occurs when nuclei strU:e oloud drople"ts. This is most likely to occur when there is a high concentration of nuclei. The second. process, sublim.lltion, occurs under more nol1lla1 nuclei concentrations when the air is saturated with respect to ioe. This process forms the basis for the area of effect model (Elliott, 1969) cloud physics which was developed for orographic clouds. This n:del introduces new sets of ioe particles at successive 2C nucledtion steps as the air rises beyond the threshold level of nucleation and duplicates what is observed in cold boxes as to effective nuclei concentrations at diff.erent temperatures. Each set of ice particles is dllowed to grow by diffusion of water vapor onto the ice particle (diffusiondl growth box Figure 1.2-1) and later by riming (accretional box). The growth rates are sinlplified by making them dependent upon particle size. The model is most re. presentative of actual processes in clouds which have colder bases and therefore lower water content colder than typical South Dakota summer cumulus. But in these cumuli tMrc is a considerable concentration of ice crystals present in their upper parts. This is IIlOst apparent in the high frozen anvils. The third sub-heading, condensation/freezing, i8 the process where ice form­ing nuclei are first dissolved in cloud droplets, then produce freezing. This 1.2_6 / / / / / / / Equ Ivolent curve for natural nucl,l AgI - NH4I(5.54% l ijAQI -NH.I 114.7%) -- / .... ' __ LW-83 '" -- ~­.'!! / / , ,, , " I I I I I I I I I I I I I I I I 10'0 10 15 .g.o ~ 10 14 ..... ~ u :> z 10 13 '" ~... ~ 1012 ~ lO" 10 10 -5 -10 -15 -20 TEMPERATuRE (Oe) -25 Fig. 1.2_3 Typioal curves showing the number of effeetive ice forming nuclei per gram at various dctivation temperatures for four different nucleating agents. 1.2_7 m4y also occur in the cold box and in clouds, particularly if artificial nuclea. tion is done using a silver iodide.sodium. iodide complex rather than a silver iodide-ammonium iodide mixture or pyrotechnics. Particles which have gro' 0-JI- :!: C)~ § g~ "Zl.-.. "a.", .w..:xw: o~i~w-aJ. ~ 1:\-aJ. 0=-J>::-->JJ 8-=J(>oiwi''"~" Z'" u :x: 0 :x: Ulll 10' 10 -5 -10 -15 -20 -25 TEMPERATURE Ie) -30 -35 Fig. 1.2w4 Ratio of ioe crystal concentrations to ice nuclei concentrations versus temperature. -10 -20 1.0 10 o 1-------t---t--/l-4 MOSTLY L10UID 0.001 0.01 0.1 a VAPOR y..,1---7'--+-~-'--ft-*----7 - 30 1.3 ".> .-:..::.:::::.:.:.::..:~:::.:-t-r~:71.2 0 1.1 ¢-' 1.0 ~ 0.9 ~~ ~:'-J"--__';:::==;Z:==;Z:==~0.8 v~¢- - 3~.00l 0.01 0.1 1.0 10 100 ".>~ PARTICLE RADIUS R IN MiCRONS f + 10 Fig. 1.2..5 Variation of Fletcher's three_dimensional diagram showing the relation between condensation, freezing, and subli..ma.tion as a function of saturation ratio, teIQI:erature, and particle size. This differs from Fletcher's primarily in thdt the condensation surface begins at a sattlration ratio of uni ty rather than 1. 05. 1.2_10 as condensation or freezing nuclei# even at large s~rsaturations. Some super. saturation is required even with large particles for this to occur. A particle not lying on the curve is out of equilibri~ and there will be a change in state taking p1aoe either through condensation and/or freezing, or through evaporation, as indicated in the diagram. At temperatures below ab:lut .12.5C direct sublimation (deposition) of vapor on ioe will occur. This occurs at colder temperatures even when there ls sub6aturation with respect to water (but supersaturation with respect to ioe). As has been discussed by St. JWand# Finnegan and Burk4.rdt (1970)# the nuc1ea. tion process is oomplex and involves nunerous interactions between a population of nucleating smoke# dust and smoke arising from many sources, and the liquid water droplets in the cloud. Throughout the process there is kinetic coagulation, coagu_ lation through Brownian movement and through electrophoresis (with charged smoke particles) between the three sets of particles. The ice embryos grow by diffusion and by contact lIB do the larger frozen drope fonned by contact nucleation. Cold b:>x observations show that nucleation is more abundant with a higher water content (Steele and Davis# 1969; Donnan et al. # 1970) as would be expected from the ab:lve considerations. There is a considerAble vdriety in the fom of ice crystals which are formed either in the cold OOx, or in the atmosphere. There are resultant differences in growth rates, accretion rates, and terminal velocities. Ndkaya (1954) fOWld in laboratory experiments a relationship between crystal type, or habit, and the temperature of formation. There was some dependence also upon the supersaturation with respect to ice. Figure 1.2-6 shows the regions of preponderance of different crystal forms. The dendritic form with its lllAny hexa_ gonal branches, is the form o.ften depicted as an ideal snow crystal. It occurs most frequently at _15C, where the vapor pressure differenoe over water and ice is greatest. This suggests that at this temperature water is being transferred most rapidly from supercooled cloud droplets to ioe, and that the dendritic shape multiplies the surface area upon which water vapor deposits. This is indeed the cuse; dendrites grow several times faster than do other forms. At both high and lower temperatures, the vapor excess is reduced, and .growth is less. With moder. ate vapor exoess, hen.gonal plates predomilldte (at about _12C and _17C). Scrolls or cups, and spatial plates appear at still warmer and colder t8Jlli:Oratures. The needles appear at temperatures between zero and _SC. 1.2-11 :t: OC:NORITIC <> SECTOR AND PLATE • THICK PLATE • SPATIAL PLATES I NEEDLE XIRREGULAR NEEDLE alCOLUMN oSCROL.l., OR CUP T. w Fig. 1.2_6 Ta-s diagram, showing the conditions of formation of various shapes of snow Czystalsi W is a line giving the saturated vapor pressure with respect to supercooled water (taken from Nakaya, 1954). 1.2-12 The dynamic effects of seeding cumulus, such as the enhanced buoYMLcy, in­creased updraft, and rise in cloud top, has a oomplex effect upon precipitation. In most clouds it increases condensation and conversion of clom droplets to ice. In lMny, it increases precipitation. In sone cases the seeding-produced increase in updraft so limits the totdI in_cloud. time for the ioe particles that they do not grow to sufficient size to be counted as precipitation particles. Hydrometeor particles are large eno~h to have 4Jl awrecia.ble terminal velo_ city and. therefore to fdIl relative to the updraft. In a convective up:iraft strongly tilted ~ verticdI wind shear, they fa.ll coq:U.etely out of the updraft. In a vertical updraft, they fall wck down through the updraft, or rise with it, or if the temindI velocity equals the updraft, remain suspended at one level for a while. Weinstein (1970) has constructed a vertical updraft numerical model in which the changes with til'll8 in hydrometeor content and in other- quantities under natural conditions, and under the condition of artificial nucleation, is eompl1ted for the whole vertical column. This time dependent model represents a consider­able refinement over the steady state convection model because of the ability to handle the return of hydrometeor water down the column. The computations no longer apply strictly to the updraft cap, but to the entire colUlllll. Weinstein dIso in­cluded the effects of evaporation at cloud top and in the subcloud layer. Wisner, Orville and Myer (l972) have also constructed a time dependent roodel which is quite relevant to the northern Great Plains. In this model the freedng of the liquid hydrometeor water is predicted by means of Bigg's (1953) relation_ ship between drop freezing, drop size, ani temperature. Also incl\ded is a separate treatment for hail and rain, and for subcloud evaporation of precipitation. Figure 1.2-7 shows the vertical velocity field computed for a given sounding. Note the rapid developoant of the updraft during the first half hour, followed by its decay, with a downdraft developing in lower layers after ~he first hour due to rain eVd_ pordtion. Figure 1.2-8 shows the rainwdter content of the cloud. Hedviest rain occurs at hour 1. The distribution of particles within d moderdte size cloud. is shown schemati­cally for the ndtural nucleation and seeded cases in Figure 1.2-9(a) and (b). The up::traft and rainout rBg'ions are separated due to the effects of the wind shear which is usua11y present on South Dakota cumulus days. It is assumed that the same up:;iraft structure (shown by solid arrows) is present in both cases, Le., thare is no dynamic effect produced by the AgI seeding. The broad. open arrows indicate the 1.2-13 Fig. 1.2~7 Verticdl velocity for hail case. The contour interval is I m sec~l Fig. 1. 2_8 Rainwater for ha.il case. The contour intervdl is 0.25 gm. kg-I. 1.2_14 HEIGHT I KHI TEMP lei -4.0. (a) Naturally nucleated cloud ( b) AgI Nucleafed Cloud Fig. 1.2-9 Particle distribltion in ~turally ani AcrI nucleated clouds. 1.2_15 general movement of the particles. Some time after their fonnation in the updraft, the smaller ones move into the anvil while the larger ones descend through cloud base. The left Mnd diagrams show the generalized vertical distrib.1tion of liquid water content, disr69Arding the descending rainfall branch. In the naturally seeded case we see first the fOImation of cloud droplets just above cloud base and a steady increase in cloud water content with ascent. At the same time the coalescence mechanism gets underway and h,ydrometeor water coumenoes to accumulate. The h,ydrometeor water starts to freeze at sul::tz;ero temperatures, and is mostly all ice at temperatures under ~25C. The particles grow to graupel size, and some perhaps to hail size, e.specia1ly in larger clouds where there is more t1J:re for growth in the updraft. Ice crystals are formed by activation of natural ice forming nuclei and become abundant at temperatures below about _250. Those particles whioh are fonned earlier in the updraft are most likely to grow largest. They will achieve sufficient terminal velocity to fall from the updraft as graupel, Mil or snow. On Edling below the freezing level they melt rapidly, turning to rain. Hail of sufficient size will not melt. Beneath cloud base evaporation of falling" parti_ cles occur, especially of the smaller ones, which descend more slowly and Are there­fore exposed to relatiwly dry air over a longer period of time. Ths smaller parti­cles which are flushed out into the .mvU haw an excellent chance of evaporating through mixing with the drier air 410ft. There Are three basic ways in which the particles formed in updr.afts are eva_ porated before reaching the ground as precipitation: 1. Evaporation of cloud droplets due to mixing with subsaturated air entrained into the updraft. 2. Evaporation of rain beneath cloud base. 3. Flow into the anvil.with ultimate evaporation. The precipitation efficiency (to be defined in detail later) of -the cloud is reduced by the losses due to each of these factors. In the AgI seeded case (Figure 1. 2~9(b) more ice crystals form, and' do so at a lower level in the cloud. Moisture which under natural conditions would increase ioe in the form of graupel and hail, or simply remain as supercooled liquid cloud droplets, instead is converted to ice crystals. In the descending rain column there is therefore more snow, less graupel, and less hail. The snow has a .sma.ller teminal velocity than the graupel, or hail, and this leads to JIl)ving the precipi_ tation column forward; however, the earlier start of growth has a contrary effect. The evapora,tion of rain is inorellsed greatly if the precipitation Ilrea is moved forward so that it falls from the anvil rather than down through cloud. With more ice crystals and less graupel, there is also Il greater mass of particles flowing into the anvil, oilgain :because of the difference in tenainal velocity. It oan :be seen that for seeding to produce more preoipitation at ground level its dominant effect has to be the earlier production and growth of an abmdance of precipitation size particles whioh will fallout of the cloud before getting into the anviL Such early production also reduces losses due to entrainment evaporation of cloud droplets. In smaller clouds where there is little or 1'10 wind shear, the updraft behaves IlIOre as an ascending- b.ibbl.e which rises with the growing particles within it. but on reaching- its equilibrium. level (with the environment) COllIDEInoes to desoend due to the downward drag of the particles. Thus. Figure L 2~9 might serve to illustrate the process under these ciroumstances if the horizontal scale represented time. &naIl clouds which ice up IlI4Y separate into two parts, II. lower rainshaft, or virga. and an upper oloud consisting of small ioe crystals whioh simply evaporate. This process is more graphic with vertical wind shear where the top drifts away from. the rain shaft. seeding of small clouds whose tops are warmer than -lex:: may produce extensive overseeding, oonverting them to small ioe crystals with no rain shaft at all. Still, lower clouds may not rise t<:l sufficient height for the ice crystal mechanism to :be operative. Yet, if their base 'temperature exceeds about +lOC, there is a good chanoe that they will rain out by the coalescenoe mechaniSlll before 00lQ­pletely evaporatillQ'. This is es~c1ally true: in a maritime air mass. Clouds larger than those of Figure L 2-9 tend to rise vertically even with considerable wind shear, dUG to the strong upward flux of OpPOsing low level hori­zontal lllOmentum. The rain shaft, however, is UBually diverted from falling back down directly through a steady stdte up:lraft. Airflow in larger stOI1llS is illus. trated in the next &ection. If the dynamic effects of AgI seeding are to be considered also, then reference should be made to Figure L2_l0. The seeded case particle distribution (Figure L2.10{b) is similar to that shown for the case with no dynamio effeot Figure 1.2-9(b), except that the upward extent of particles is much greater. The updrafts are enhanced due to dynapdc effects as indicated by the long up:iraft arrows shown in (bl, which leads to a higher top. Therefore, a greater total nUlllber of particles are produced .md also the total mass of water in the form of liquid and solid particles is greater. In a sense the dynamic effect of seeding is the enhancement of the efficiency with 1.2-17 HEIGHT (K ft.) TEMP IC) 40 -55 (0) Naturally nucleated claud 40 -55 30 -36 20 -10 10~_~ .........J. +10 .::::= .-. -=-.---: fL lbl AgI Nucleoted cloud (with dynomic effects) Fig. 1.2_10 Particle distribution in naturally and AgI nucleated clouds (with dynamic effects). 1.2_18 which the potential energy of the air mass is converted into updrafts and into more condensate. The early tranaformation of liquid particles to ice provides this extra. energy. This positive hctor has to be a.dded to the list of negative evaporation factors presented above when the effects of seeding on precipitation are to be considered. Another difference of Figure 1.2~lO(b) frOJn Figure 1.2-9Ib) is that the rain shaft is thrust forward because the raised top portion of the cloud is entraining higher forward lllOIDElntum. In this case it is shown falling out ahead of the main cloud mass. This leads to roore rain evaporation. In addition, many of the additional ioe crystals are thrust out into the anvil, and their mass lost through evaporation. Ha.il s~ression, or reduction, depends upon the seeding produced increase in ice crystals at the expense of hail particles. The figures make it clear that this will normally be the case with seeding, regardless of whether rainfall is increased or not. The integrated radar return is often employed to III;lasure precipitation (in acre feet usually) falling from a cloud. It should be pointed out that if seeding merely incteased the m~lIber of precipitation particles, but With the total mass remaining constant, then the radar return Would indicate lees total precipitation. This is due to the hct that the radar return is very sensitive to the drop radius (6th power law) so tlult the increased nur:lber of drops (proportional to the reduction in r 3) does not offset the radius effect. Another feature of the radar return is the increillSed reflectiVity in the presence of hail. The misinterpretation of radar return due to ignoring hail when present can be serious. The foregoing disoussion is rather simplified in nature, and is presented for the purpose of providing general information concerning the distribution of partioles within clouds and their fallout to ground level. It suffers roost from the CO!lIPres. sion of several stages in a cloud's life cycle into one figure. Seeding effects not only alter the updraft dynamics but the area and the duration of the cloud. For example, it was found in northern Arizona (Slnith and Weinstein, 1971) that ice was not an important constituent in non.seeded cumulus clouds until the dissipation phase; whereas, it was an il\lPOrtant constituent in updrafts when seeding has occurred. The duration and area of seeded showers exceeded those of non-seeded ones. 1.3 Factors Governing the Type and Degree of Cwnulus Developnent In Section 1.1 it was stated that cumulus clouds are generated in an unstable air lIl.4SS havirY,;l' a lapse rate lying. between the wet and dry adiabatic values. The actual generation of cumulus pre.su~se8 the existence of an initial lifting of the air mass when it is saturated at cloud base. This can be caused by d thermal bubble ascending from heated ground, or as a lift as air is driven over a hill, or up a gravity wave incline. It can also be caused by the lift of warm moist air over a spreading wedge of rain-chilled air near the grourd. The actual initiation occurs at the lifting condensation level (LCL), or at the convective condensation level (ceL). These levels may be different, depending upon the initial distribu. tion of low level moisture. In addition to these factors, the lapse rate itself can be altered by surface heating, surface moistening, radh.tional cooling aloft, and differential horizontal temperature advection. The manner in which these processes can destabilize an air mass is shown schematically by the soundings on Figure 1. 3.1. The inherent insta. bility of an unstable air mass Ny be increased by the lifting of the lower layers by a front, or by an orographic barrier as depicted in Figure 1.3-1(e). The }'eight that cumulus clouds reach is often govented by a subsidence inversion created by warming aloft as depicted in Figure 1.3.l(f). Other processes are identified in the figure legend. The range in the degree of developnent of cumulus varies from that of the cumulus humilis (the least intense), to that of the cumulonimbus (the most intense). In the case of the cumulus humilis there is a very shallow, moist unstable layer topped by a stable layer or an inversion as depicted in the schematic sounding shown on Figure 1.3.2. The depth of the development is usually limited to one to three thow;and feet. This type of developlWmt is often noted in portions of high pressure areas where there is much subsidence. The tops of cumulus hUllli.lis are usually at a unifoIlD height due to the strong stable layer or inversion. In the case of the cumulonimbus, the depth of the unstable layer often QXtends to the tropopause and in some cases the growth is vigorous enough to penetrate into the stratosphere. In between these two extremes, there occurs various degrees of cumulus developlWmt, in relation to the level and depth of the unstable layer, relati ve to other cumulus cloud types. The growth of cumul.\l5 med.iocris is often lilllited by an inversion. In some cases the developaent may be vigorous enough to penetrate the inversion. If there is an unstable layer above the inversion, the , "m" 1m ", .00 .00 mb mb "~ . , p >d'.<, } I "\ ,1000 3~Thermals \ ; ~ThermOI$ 1000 T- (b) , Radiationol Coollno 111'I'\ Af Cloud To, .00 " \" -- 0 Nocturl'lol 500 mb ... 1.: '. \ . Clouds IS ,.~ \ \ 1000 1000 T_ T-I" 1'1 5:;gt---""21;----;f'=, 1000L --..:L_ 1000 L ~__ Fig. 1.3-1 Schem4tic so~ illustrating the influence of various a.tmospherlo processes on the alteration of the air mass, (a) surface heating with addition of sensible heat, (b) combination of surfaoe heating, eddy transfer of sensible heat and latent heat release, (0) radiational cooling aloft, (d) wazm &ir advection at low levels 1, and cold air advection aloft 2, (e) enhancement of instability from lifting, (f) warming aloft by subsidence. 1.3_2 500 m' 500 m' 1m'" , YcI, \, '\\ ~\, , T c T_ T_ T. c:ill ~~CU Med 500 m' Ym,\. Yd",- \, ,,\ ~\ T,o,,, \ 500 m' T-T_ Cb Cal Fig. 1.3_2 Typical. BOW"LdingS and related CUIm11us developrent. cloud will develop into d cumulus congestus, dnd, if conditions dIe fdvordble, into a cumulonimbus. The cumulus mediocris is considered potentid11y responsive to seeding under oonditions where the added heat of fusion from seeding is sufficient to increase the buoyancy of the system so that it Will penetrate the inversion. Simil.uly, growing ~UB ?onges~us dIld cumulonimbls, under fdvorable temperdture conditions, would be expected to develop to greater height than they would have if not seeded. One of the essential processes creating an air I'll4SS with d deep unstable la,yer is differentidl horizonta.l temperdture ddvection in the lower and upper troposphere. A schem4tic illustration of d COIIIllOn situation for cumulonimbus and squall line developoent east of the Rocky Mountains is shown in Figure 1. 3-3. l~ 3-4 Fig. 1.3_3 Schematic illustration of processes in formation of II deep \ln8table layer by differential. advection. The dashed lines are the mean isothe:rms for the layer from the surface to sao mb. The dotted lines are the mean isotherms soa rob to 2S0 mb. The open arrow represents mean flow surfaoe to sao rob; the solid arrow, sao mb to 2sa mb. 1.4 lArge Cumulus Cloud Dimensions Hovel:leIlts. and Structure There have been few actual observations relating cloud diameter, cloud height, and updraft or downdraft. Some of the cloud depth.diameter measurements that have been reported are sUlllllldrized in Figure 1. 4-1. The general relation between cloud depth 4l'Id diameter in a hurricane, as reported by Ha1ku.s (l960), is shown at the top of the figure. The dashed line in the center of the figure outlines the general depth-diamster relationships reported by Hosler, et al. (1967) for several Pennsyl­vania cumulus. The dashed line at the far right encolllp4.$ses dome depth.diameter relationships reported by Simpson, et a1. (1967) for individual tropical cumuli. The two areas in the lower portion of Figure 1.4_1 were obtained from Plank (1969) of Florida cumuli. The lower left area outlines the dept;h diameter relationship on a typical cloud. cover day between 0900 and 1000. The other .nea denotes the relationship for 1400-1500. The figure su;;KIests that there is often a simple 1 to 1 relationship between cumulus diameter and depth. However, the figure also suggests that \mder certain circumstances cumulus depths considerably exceed diameters. This figure does not include the smaller cumulus such as humilis where diametere may be several times the depth. A$ will be seen later, there is a natural process whereby small adjacent clouds merge into a rapidly grow1n;;J cloud. Although there is little information on updraft velocity in the nore se1l9re convective storms, some limited information suggests maximum updraft velocities much greater than those reported in the Thunderstorm Project. Malkus (1960) reported pilot estimates of occasional enco\mter of updraft and downdraft velocities on the order of 50m.s· 1 in hurricanes and typhoons. Occasionally, irdirect means have been utilized to estimate vertical velocities. In a recent article by Booker, at al. (1969), vertical velocities in a severe storm were estimated at 45 m s-l from the data obtained when a balloon transponder package was entrained in the low level echo free area of the system. Typical updraft vertical velocities below the cloud base of severe convective stonos have been observed in the Hailstorm Models Project conducted by the Institute of Atllt)spheric SCiences (lAS), South Dakota School of Mines and Technology. For the hail cases of 1968 the maximum draft below 1::1000 base ranged from 2.5 m/s to 15 m/s. The Hailstorm Models Project has also confirmed that in the air mass th\m_ derstom the maximum updra.ft area exists along its trailing edge, whereas, in the squall line thunderstorm the maximum updraft is in the right front quadrant. With 10 ! w ~ ~ 0 :>g "w >g "..J: "W 1.0 J: ~ :> ~X "~ Hurricane Daisy Tropical Co uli , -..... penns:::;.n~ __:::: .l. .) ./ / /-.-., r .--,:., / /' ...... Florido .... / Cumuli (PM) ........ Florida ./'') Cumuli tAMI/ / / / / / / ./ L/ 1.0 10 0.1 '-__'--'-----'----'--1-.J...J...L.l.__-'-_..l---1-....L..L.J..J...J..l-_---' .0.1 DIAMETER (kml. Fig. 1.4-1 Cloud geometry. General relationship between cloud top height and. cloud diameter. -, Hurricane Daily 27 August 1958, Ha.lkus (l960); , Hosler, at 41., PeMsylvania (1967); .----, individU4l trop~cal cwmlli, Siml::son, at 41. (1967); - • -, Florida CUlIIl11'us 0900_1060, Plank (l969); _ ••• _ Florida cumulus 1400_1500, Plank. (1969). air mass thunderlltorms, the prllll4ry inflow does not necessarily im:Dediate1y S'Witch to the front on 1:::eooming a travelling stonn 4S was observed by Garrahan, et al. (1969). Varioua observers have noted that in the vicinity of the updraft area l:elow the cloud base there are roll or scud clouds making it possible to determine the location of the updraft and to some degree the magnitude of the updraft by the appearance of these clouds. Although in many instances the below-cloud updrafts are not delineated by a single smooth core, measurements reported by Auer and Sand (1966) beneath a heavily precipitating cumulonimbus cloud did show a smooth curve. This is believed to confirm the conclusion by numerous authors that the more in_ tense thunderstorms possess a steady_state updraft in their IIIdtUIe phase. The environment nonnally contains horizontal momentum. The entrained air has horizontal momentum of its Olm which is mbed in with the horizontal momentum with. in the updraft. It is well mOlm that large convection cells on the average move to the right of the mean envirol\lll&t\tal wind vector through their depth. The cell speed (mean vector IIIdgnitudel is usually less than the enviro:nmant41 mean speed. On the basis that the cell has enough in':ernal cohesion to act as a barrier, that" is, that it resists and deflects the environmental wind, it has been concluded that a hydro-dynamic pressure distril::ution around the cell will result. Newton and Newton (1949) and Newton (1963) have investigated this possibility and calculated that dynamic pressures on the order of a fraction of a millibar are possible, and that these will induce vertical ItlOtions which will be superimposed on the buoyancy produced updraft system. Figure 1.4.2 shows the pressure distribution and the up­draft on the side to the right of the cell, which is assumed initially to move with the vector mean wind. This effect, according to Newton, generates reactivity and makes the cell appear to move to the right of the mean motion. Observations of inflow patterns suggests that this happens. New towers may often be seen feeding into the main cell in this region. To the left, and behind the cell as depicted in the figure, there is an in­duced downdraft. This lies in the general position of observed downdrafts in large cells. Sometimes a large cell moves to the left of the mean flow. Chagnon and Staggs (1970) found that hailstorms tended to I:e right.turning when connected with ql14Si_ sta.tionary fronts. Ahead of cold fronts they were ohen left~turning. In the first case the inflow was from the southeast, but ahead of a cold front, inflow tended to appear ItlOre frequently on the left flank. 1:4.-3 Fig. 1.4.2 Hydrod;ynamic pressure on surface of cylindrical convection 0811 (denoted ++ cmd =), and resulhnt superimposed updraft on right side of oell. 1,4_4 The dillcovery that large oells rotate with an appreciable angular velocity has suggested that the Hagnus 11ft force IndY Act to deviAte the rotating cell to the right only of the drstream within which it is eml::edded (Byersl 1942). fltiita and Grancloso (1968) have made A careful investigation of A split thunderstorm using Doppler Radar and other measurements. One part of the split had cyclonic vorticity and became right moving, the other had anticyclonic vorticity and became left moving. The splitting process is illustrated in Figure 1.4_31 taken from fltiit", and Grandoso. The Magnus force is a function of the ratio of the peripheral velocity of the cell to the airflow relative to it. For lArge slowly rotAting systems, this speed llldy produce a force of lArger size than the leftward deviAting force resulting from the geostropic imbalance between the environmental horizontal pressure gradient and the coriolis force of the slowly moving cell. It seems wtlikely tha.t the vertical component of vorticity would be sufficient in normal size cells to effect a right turning, although the I!luggestion by Barnes (1970) that the vorticity in the Ekman shelU zone, while being entrAined into t.he cell, is twisted into A vertical vector is interesting. Vorticity (vertical comPOnent) of 10.3 sec·l could be generated ir, the time it takes this luyer of air t? become entrained. This concept is illu_ strated in Figure 1.4_4. A self propogating system develope when a squall line is fonood in the rain chilled outflowing air beneath Ii thunderstorm. This air moves in advance of the storm as a peuedo cold front 1 plowing up the warm moist air of the lowest levels to a level where it is entrained into the cloud updraft. This is illustrated in Figure 1.4.5. The downdraft is generated as ndd_tropospheric air is drawn into tt's system by the action of rain which drags it do1rning houra. It 15 during this time that nocturnal thunderstorms prevail. Figure 1.4_7 shews the inflow occurring near cloud base in the upper air layer. The up:iraft structure akove is similar to that of the squall cloud. There m.iy even be a wedk squall line beneath the cloud, bJ.t it is CQDP:lsed of air cooled and desoend.:i,ng within the lower air JllllSS itself. Air from the u~r layer is not involved as in the case of the squall cloud, although some genuine squall clouds can 4nd do occur at night when rain chilling is intense. The implications of nocturnal thunderstorms with respect to cloud seeding are intriguing. One should expect a dearth of ice forming nuclei to be present simply because "the inflow is from a level well akove the ground, the ultimate source of most such nuclei. This would mean that cloud seeding could be more effective here than in the case of clouds drdw1ng air from the lowest levels of the atmosphere. On the other hand, ice forming nuclei could have been accu:aul.ating near the bdse of the upper layer earlier in the d4y, by essential.ly the same mechanism that water had accumulated there. By dawn, considerable moisture has been removed by rain from this upper layer, and subsidence is reducing its humidity. Where sequential soundings have been taken during convective days this sequence of events is clearly shown. It is not until the middle of the next day that the moisture supply of this UJ:Per layer be_ gins to be replenished from below. Fig, 1.4~6 Tracks of individual rainstoI1llBf squall line of 20_21 May 1949. Dashed lines connect rainfa.ll centers observed at same time (date/hour I::elow); dots and circles on tracks sha'-' positions at successive hours. Maximum hourly rainfall observed with each stom (inches) indicated ~ slanted figures. (After Newton and Newton, 1959). HEIGHT (K ft.) 40 .. -4r-----------..J 30 2' 20 10 STABLE LAYER S\lrf.-'-- --'-'--'-'CLL'-!...'----.!..U-U.<-'-'-'-'-'-L--'- _ Fig. 1.4_7 Features of 4 nocturnal thunderstorm. 1.4_11 1. 5 Cloud Kinematics The kinematics of cloud develoIXnent are interesting. They explain a good deal allout the shape of the convectively produced cloud. Figure 1.5_1 illustro'l,tes the vertico'I,l and horizontal shape of 0'1, cloud subject to the various typico'l1 intero'l,ctions with environmento'l1 flow noted in the figure. The source is assumed fixed in the earth. Figure 1. 5.2 shows the views for a typical moving source on the right for_ ward flank of the cloud base. The precipitation efficiency is dependent to some extent upon the sho'lpe of a cloud. If the precipitation products falling from the anvil hll into feeder clouds, or are otherwise ingested back into the updraft system, then efficiency is higher than if they f;lll out into dry subcloud dr. The difference is illustrated by the vertical sections of Figures 1.5.1 and l.S_2. The precipitation efficiency is greater in the 1. 5_2 case because of the more vertical orientation of the cloud. Figure 1. S_3 illustrates the manner in which feeder clouds o'I,re entn.ined into the main cloud system. These diagrdtQ6 are a log-ico'l1 extellBion of the kinematic con­sidl!! lratiollB entering into Figure 1.5-i ar.d 1.5.2. Two basic air ma:>s situatiollB are considered, the first where deep warm advection is occurring in a southerly flow just behind a high center; with the winds veering aloft. The second is where deep cold advection prevails, as in a post upPer trough situation; with the winds backing aloft. The winds are shown for these two cues in the upper panel of the figure. Also shown is a typical cumulus cloud verticlll velocity profile, for deep clouds (daahed line), and sha.llow clouds (solid line). The next panel do'Wn shows the type being considered. These vary from the right turning type (system motion to right of mean flow in which system is embedded), characteristic of deep convection; the no turning type (motion same as mean flow), characteristic of shallow Convection; to the rare left turning type which sometimes occurs in deep convection. The next P4llel down shows the low level flow vect:or (solid o'II"roll), the cloud system motion (hollow o'I,rrow), and t:he resultant inflow int:o the system (dashed arrow). The next two panels down show typical cloud. profiles, west_east and north_south sections, with the area of feeder clouds indicated. The bottom panel diagrams show in plan view the manner in which the feeder clouds msrge into the main system. The majority of feeders are observed to move in from the southwest through southeast quadrants, because right turners are the rule for large clouds. 1.5-1 _.L- ""'- X lui FlllEO SOURCE Of UPORAfT Fig. 1. 5-1 Updraft trajectory and. cloud outl~e in horizontal and vertical planes produced from a single fixed souroe. These features are based upon tha typical envirOl\lXWllnta1 wind (hodograph) 4n:l in-eloud updr4ft (profile) and. the dominant horizontal momentID controls indicated. This might be a typical shower cloud. The precipitation tends to f411 4he~ of the cloud. Fig. 1.5_2 Plan view of Qloud Qutline based upon Fig. 1.5_1 assumptions rot for d discrete ItDving souroe producing new towers 1. -2. 3. etc. dt positions 1. 2. 3. etc•• at successive times 1. 2, 3. etc. The successive new towers are assumed to merge into one cloud mass composed of the so-mingled predecessor towers. Cloud is lost in the downwind wake of the anvil. The cloud appears . to move with the source motion velocity C. A continuous ooving SQUI"C8 would produce the same moving cloud outline. A -one puffer- cloud would I'IlOW along a sing'le tower tr~ectory as time progressed and its avenge m::>tion would be near the enviroIVOOntal mean for the layer. The precipitation falls more directly through the cloud than in the case of Pig. 1.4_2. It may also fall into front feedsr clouds. DEEP WARM ADVECTtON 0[[1' COLD ADVECTION TYPICAL HOOO'"'''''DLk::-. ~~: ~ 'NO 3D • ~ 7•• UPDRAFT 3D : 8:10 300 I( '/~~/ PROFILES I(f ~", '~4D '". I 0 MIs 10 3D M/S TYPE RIGHT NO LEFT RIGHT HO LEFT TURNING TIIRNING TIIRNING TIIRNING TURNING TIIRNING VECTORS "t:::> f/) 1) ~ c;l a w-£ 16 C1 !1 ill a a PROFILE w , w , w , w , w , w , H-S (J~ dJ fl H~~ Jj- PROFILE lJ H S H S H S S H S FEEOER a (J C9 D (3 CJ OIAGRAM Fig, 1,5_3 Cloud profiles for various types of cloud systems under deep warm and deep cold advection synoptic situations. 1,5_4 1,6 Precipitation Efficiency This term has appeared several places already in the text. It has been de. fined as the fraction of the condensate produced by 0 cloud system which ultimately reaches the ground as precipitation (Elliott, 1964). A more collIlIOnly etDP10yed de­finition in the Great Pbins is the fuction of the tot41 water (in liquid, solid, and vapor form) processed through the base of the cloud which f411s as precipitation to ground level, The btter definition is equivalent to the fonner provided that the updraft moves high enough to condense 411 of the water brought up through the bue. Foiling this, the latter definition gives a lower value. In general, the precipitation efficiency of cumulus clouds is reduced below 1,0 by various processes. The most illQXlrt(lJlt ones are: 1. Evaporation of cloud droplets due to entrainment of drier air. We have already seen that the smaller the cloud's updraft area, the more severe this effect is. A very large fraction of <.tIl cwnulus clouds produced by updrafts are reduced to zero efficiency by this evaporative process. If nucleation occurs before this develops, then the efficiency may be raised above zero. In this sense, the most illQXlrt(lJlt control on efficiency is produ::ed by the cloud top temperature, because this is related to the concentration of natural or artificial nuclei made available for rain production. 2. The updraft profile is such as to produce a large (lJIvil, which in tu.rn evaporates. This process has been mentioned in Figure 1,1.4 discussion. It is probably the most Uilportant factor .in efficiency in clouds of sufficient size that entrainment is not a domin(lJlt factor. The upira.ft profile C(lJI be chdnSled by the dynamic effects of seeding, and this can either decrease or increase the precipi_ tation efficiency. 3. The shape of the cloud plays a role in efficiency. A cloud leaning due to wind shear may have precipitation particles fallout of the side and into relative­ly dry air. It is less efficient due to the excessive evaporation of the pdrticles before they reach the groW1.d. Small cumulus having a. bent.over profile due to wind shear are also often torn assW1.der before they can develop into shower clouds. Any precipita.tion falling into the drier subcloud la.yer experiences evapora_ tive loss regardless of whether or not the aboV$processes are effective. This is an im,port(lJlt fa.ctor in the generation of SCIUllI lines as was discussed in Section 1. 3. It shows that UL increase in efficiency C4l'll'LOt be regarded as the sole goal of cloud seeding. If there were no subcloud eva.poration, there could be no self propogating mechanisms such as squall lines. 1.6-1 A practical precipitation efficiency formula definition (lAS 69.2) is given and where C is the storm speed, V n the mean surface Idyer wind component into (o~site the vector wind D) the storm, L is the storm width, q the mean specific humidity of the surface layer, P 2 - PI the surfaoe to cloud base pressure difference, 9 the gr.wity constant, T the storm durdtion, and R the mean rainfall over the area A. An example is shown in Table 1. 6.1, taken from Grand River Project data and appearing in the 1M3 69_2 report. Table 1. 6_1. Values of storm moisture flux, storm duration, total water, and storm efficienoeis for four major stoms. July 17 July 19 August 6 August 7 Moisture Flux Storm Dura.tion Total Water Storm Efficiency (109 qm/sec) (hrs) 110 12 9!I'Il IV 32 4.0 240 " 10 2.3 5 6 7 2.2 13 24 24 3.0 160 62 It is seen that efficiency as defined increases with size, although the top water producer (July 17) is not the most efficient. In general, efficiency does increase with cloud top height up to a point bByond which it may decrease. Very large clouds tend to throw precipitation size particles (or incipient precipitation particles) out into the anVil, where they may be lost through evapordtion. Cloud top temperature is actually a more significant parameter than cloud top height, although cloud baBe temperature (or dewpoint), updraft strength and wind shear in the vertical are also important. Clouds whose tops do not rise above the freezing level will not grow precipitation by the ice mechanism. If the base is cold enough, they will not grow to it ~ the coalescence mechanism either. These low clouds have zero efficiency because the cloud water condensed evaporates before there is any precipitdtion. L6_2 This formula or variants tlvareof have been used many times to estimate cloud efficiency. For eX41llp1e, Auer and Maurwitz (1968) computed the precipitatiop efficiency of a number of severe stonos. Figure 1.6-1 illustrates the h.rge vari_ ation in the precipitation efficiency of severe stonos with wind shear. It should be noted that the dyna:n1c effects of seeding introduces an entirely new element into the preoipitation picture which seriously oompromises the use of the efficiency concept in cormection with weather ~ification. A cloud whose dynamics have been affected by seeding is a new cloud. This cloud may be less efficient than it waa in its unseeded state, rot yet may produce more precipitation beoause of its enlarged size. 1.6.3 .O....RWln.III12) \'<"""""....11 2J·U~~\, 10 JUL~" .... _IHART~~kJ!!~1HAUSER19$71 ...... _(H[WTON 1986) 24 JU..E"; ......... ..-JUHE" 27J~T 21 N"E 1~~U.;t.- (CHISHOL,,11I70!_ K)20S040!)060 PRECIPITATION EFFICIEHC'r.'llt Fig. 1.6_1 Scatter diAgram. of wind shear versus precipitation efficiency for 14 thunderstorms which occurred on t.ha High PlAins of North America. DAta point.s from Auer and Karwitz (1968) indicated by dot.s. Da.shed line was free.hand fitt.ed to data. 1.6_4 1, 7 Stages of Cumulus Develoenent At any gi ven til'lle over a fairly large area there will exist a spectrum of cloud thicknesses and horizontal dimensions. At any given time cu:nulus clouds will be found. in different stages of. their life cycle. If coulds within II. given area are viewed over a period of time, it will be found that they go through a cycle of events. They start as a Sl'M.11 cloud puff, then grow to a fully developed stage, and finally decay to a fragroant, or to nothing at all, A single cumulus humilis or cUlJU.l1us med.iocris tends to go through such II. cyole in al:lout ft ve minutes. With more active convection such as that characterized by cumulus congestus, the cycle of events for a single cloud may cover ten minutes. Under strong convection, where the llWtimum davelopaent is a single CUIlnl1onimb1S, the life cycle may be 30 minutes or more. If clusters of cumulinimbi form a mesoscale orgllnization, then the whole cloud system may endure for hours. Figure 1.7.1 illustrates the situation where a new cloud forms and adds its mass to the old rermant, a pulsation phenomenon results. In .m.y given field of convection clouds, at a given time, there may coexist all of the species smaller than that of the lMXimurn develOpllent in the field with the smaller size clouds being more numerous by far. There is II. diurnal trend in this spectrum of sizes. In particul,n, '[he lll4Ximwn develoJm'lnt increases in size as the day progresses. Diurnal effects have been studied by Plank (1969) in the vicinity of Florida. His cloud photo mosaics show that characteristically a field of small clouds in the morning is replaced by large clouds, or groups of clouds merged together, in the afternoon. However, some of the smaller clouds continue in evidence into the afternoon. Plank used steorographic methods to canpare cloud depth to diameter. On the While, the depth diameter ratio was near one, but tended to be around 1, 5 just when the large olouds were appeari~. This suggests that as cumulus towers broke through a stable layer, or d small inversion, they suffered erosion as they penetrated into the drier air aloft. The concept that convective penetration into drier air aloft is a cumulative process, with early penetrating towers depositing their moisture in the upper layer thus paving the way for later penetrations, is an old one. For example, the narrow towers of a cumulus castel_ latus are often replaced. by eu:nuloniml::us clouds. Mathews (1972) has described a modal which simulates this repetitive process of c\Dulus penetration. If the final maximum developnent cloud species is an adequate rain producer, then cloud seeding of new towers ll8 they emerge can accelerate the cumulative process. If the situa_ tion is such that no natural penetration into tI-e drier upper layer will occur, then seeding may provide j~t enough im,pulse to bring this al:lout. 1.7..1 c Fig. 1. 7Ml Illustrating pulsation (discrete) variations. In C488 A there is no pulsation, it is a "oneMpuffer'" type. In case B there are six new pulses (feeders) from position 1 to 6 (a.bout every 15 minutes _ characteristic of back feeder or 41lout every 2S minutes _ characteristic of big front feeder). In cass C there is only one addition of a dis­cernible pulse after 1, and that is 4t 6. It is continuous feed super_ oell which can last an hour before being replaoed by .,. new pulse. During develoPDent the larger clouds "suck in" adjoining small clouds which merge with it. In this way the overall height.diameter relationship is pres.ervad. At an advanced stage, two or more large cells may merge into a superwcell, or a mesoscale line form. Included within the complex of a larger cloud system is an encircling area where cloud developnent is suppressed due to compensating downward motion aloft. Satellite pictures often show such mesoscale patterns of convection. Their overall dimension is typically 10.60 nautical miles. Systematic cloud seed. ing may indeed stimulate such a large scale developnent. As can be inferred from the above, when the maximum developnent is large, not all clouds go through the maximum developnent stage in their life cycle. There is neither ....ater nor sky room. available. Only a few achieve max.1mum developrlent; IlXlst are stunted in their growth and die out due to the mesoscale circulation patterns around large cloud systems. Some of the small cells are fortunate enough to be feeder clouds which merjde into,. and regenerate the larjde systems. The feeder system. leads to an apparent pulsation in the larger cloud,. with regrowth as each feeder merges with it. Thus, a very large mesoe.cale cloud system, or even a giant cell, may last for hours, with a pulsation period of 20_30 minutes. It would be of no use to seed small cumuli doomed to extinction. Only the feeder clouds or the up. drafts of the main clouds should be seeded. Scmeti:mes one sub-area covering perhaps a 1,000 square mile area will have larger ml'JdmUlll developnent clouds than in a second area. This situation may last for one half to one hour,. but then maxilDUln developnent dies away in the first area while becoming greater in the second area. This often occurs in mountainous regions. It may take the form of strong developnent over one set of mountain peaks, then decay, with an upsurge across a valley in another ranoe. Seeding can have an in­fluence on the manner in which one area develops, accelerating it in one area to the disadvantage of anotrer (St. Am.and and Elliott, 1972). A diuntal effect of primary illlPOrtanoe in the Great Plains area during stvlll'ller involves the movement of rain producing systems eastward through tha area. The case of a single squall line has been illustrated in Figure 1.4.5. Squall lines, thunderstorm lines, and generalized regions of precipitation tend to move eastward through the region from the Rockies to the KisslssiWi River Valley during the course of each active rain day. These systems, or their precursors, develop during midday in the higher mountains. During the afternoon they advance ea.stward across the western high plains, reaching the eastern high plains by tha early morning hours. These will be discussed in more detail in Chapter II. 1.7....3 The sequence of daily event, and of the lives of typicdl cloud types during a convection day, is illustrated. in Figure l.7~2. In the upper panel the fi.rst clouds of the morning are fonning. The lower dr mass is st~e, and saturated near its top. At places where the lower air if lifted ~ terrain features, or ~ waves, the impulse starts a cloud. fonning somewhere near the base of the upper layer. A bubble of cloud. moves into the upper layer; it is eroded away .md dissi_ pates in the dIy air within five or so ,minutes. Initially, the clouds are castel_ latus, tall, slender, and very transient. Later, as the upper air mass a.cctul'lUlates lOOisture brought there by predecessor clouds, the clouds become roore substantial. Real cumulus congestus, however, does not develop as a rule until the ground is sufficiently heated that thermals are spreading heat and moisture thro~ut the lower layer. The second panel shows this stage, with a typical cloud lifetime of 20 minutes or so. The third panel shows what happens with a superadiabatic lapse rate and strong thermals in the lower layer. Shower clouds develop out of cumulus congestus, with lifetimes of half hour or more. New towers aAl84Z' as old one5 dissipate, and anvils appear. Anvils are: the result of the spreading out of cu:uulus tops in stable layers. They may be comp;>sed of water droplets or ice crystals. A succeeding cumulus may puncture upward through an old <'!nvil cloud. In the fourth panel a steady state system is shown which may have a duration of hours. This panel depicts a squall cloud similar to that shown in Figure 1.4_5. The fifth ~l shows the nocturnal thundershower situation (see Figure 1.4_7). Here stabilization has set in in the lower layer and this cloud being fed frOID the bottom of the upper layer, which has become quite lIlOist by now. The fifth panel shows the situation during night hours sometime after convection has ceased. Anvil remnants are raining out, and subsidence is setting in, drying- out the ~r air During the day, depicted in 1. 7_2, cumulus developllent has gone through full stage in both the afternoon and at night. As will be seen later, afternoon full develoIEeJlt is most often encountered near the mack Hills, and early morning full developnent in the eastern portion of the state. On many cumulus days full develop~ mant is not aohieved at any time. HEIGH' IKIO Fig_ 1.1_2 ~ ot dally e..nU in """,~ion_ The 11d 11,.. t.... tlw t ...t tiQUAe &no scon:Ilng te_ntura. tlw e.. tlw dew_ po1nu. The d.u1wre rapidly eroded by entrainment the smaller its up:iraft radius, and the drier the environmental alr. Precipitation: • Cloud droplets form rapidly in updrafts where the air is supersaturated, growing by condensation to about 1O\J. diameter. • ~drometeors are formed by coalescence of different size cloud droplets. This can result in some rain without any fonna.tion of ioe. • Rain drops grow by the accretion of cloud droplets as they fall through 'hom. • Ice pa.rticles form by direct contact between liquid droplets and ioe fonn­ing nuclei (contact nucleation). • Ice particles form by diffusion of water vapor onto natural or artificial loe fonning nuclei (sublimation nucleation). • loe particles form by the freezing of hydrometeor drops or cloud droplets containing natural or artificial ice forming nuclei (condensation/freezing nucleation) • • Ice particles grow by diffusion (sublimation) most rapidly while they are .mall. • Ioe particles grow most rapidly by collision and collection of cloud drop_ lets (riming) when they are large and in a cloud. • Ice crystals and raindrops are subject to fragmentation by various pro­cesses. Clouds with relatively warm tops have many more precipitation particles than the available ice fonning nuclei would irdicate. • Sub-cloud. evaJXlration of rain is important. It cem result in d downdraft of evaporation chilled air. • The greatest evaporative loss of cloud wdter in the greatest number of cumuli is through entrainment of dry environmental air. • In large cumuli there is a large evaporative loss of particles as they evaporate in the anvil. Principle Effects of seeding: • The principle micropQysical effect of AgI seeding is to increase the number of precipitation size particles formed lower down in the cloud. • The principle dynamic effect of seeding is to add buoyancy to the updraft, increasing its strength and the height to which the top rises. The area of the oloud is also inoreased. • One microphysical effect of seeding is to increase precipitation by pro· moting its early removal from the cloud l::efore evaporative process can remove the water. • Another microphyGical effect (in very cold regions of deep clouds, or in some very sJll411 clouds) is to reduce precipitation by producing an ex. cess of undersized ice crystals w;'\ich are subject to ultimate evaporation. • One dynamic effeot of seeding is to enhance buoyancy, thus increasing the upiraft and raising the cloud top. This increases total. condensed water avail4ble, and total precipitation. • Another dynamic effeot (mostly in very large clouds with large natural up.. draftG) is to speed the flow of ice crystals into the anvil, thus reducing precipitation. • A cloud's precipitation efficiency can be defined as the fraction of water processed up through its base which falls as precipitation to the ground. Luge clouds tend to be more efficient than small ones. Cumulus Behavior • Large cumuli tend to move to the right of the mean air flow within which they are elllbedded. • Cumulus cloud heights tend to bear a I to 1 relationship with cloud dia. meter. Outrider or feeder clouds merge with a large growing cloud and thus add to its cloud area, maintdining this relationship. • A squall line is a self propagating mesoscale system. • Noct=41 thundershowers are the result of convective overturning of an upper layer of air at a time when the lower layer is stabilizing. • Feeder clouds can merge into the main cloud system from any direction, OOt preferably from the south~st, through south, to southeast. They are a key factor in the movement of the system. • Cumulus clouds tend to go through a life cycle of less than an hour. Only when large mesoscale cloud systems develop do they endure for hours. • There is a diurnal variation in the maximum life cycle developnent of clouds. It often reo.ches a nlAXi.nwm in the late afternoon following lI1IlXiPun surhoe heating, but may occur at night. • When mesoscale 01000 systems fom they suppress developrent in surroW'u::l.ing Bmdller clouds. Auer, A. H., and W. Sand, 1966. Updraft measurements bllneath the base of cumulus and cumulonimb1s clouds. J. Appl. Meteor., 5, 461-466. , D. L. Veal, and J. D. Harwitz, 1969: Observations of ioe crystal and ---re;-nuc1ei concentrations in stable cap clouds. J. of AtlIlOs. Sciences, 26, W. 1342_1434. Austin, J. M., 1948: A note on cumulus growth in a non_saturated environment. Jour. Met., 5, pp. 103-107. ---"tir=on::-.~J~u~d~t. ~l~~S~:~,2;~:~~3.A thermodynamic analysis of cumulus convec_ Barnes, S. L., 1970: Some aspects of a severe, right_acving thunderstorm deduoed frOID meso-network rawinsonde observations. J. Atmos. Sci., 27, pp. 634-648. Battan, L. J., and R. I:assander, Jr., 1967: SUnmary of results of a randomized cloud seeding project in Arizona. Fifth Berkeley Synp:>sium on Math. Statistics and Probability _ Proceedings 5:29_33 U. of Calif. Press. Berry, E. X., 1968: A parameterization ()f the collection of cloud drops. Proc. of the International Conference on Clo".rl Physics, Toronto, August 26_30, 1968, pp. 111-114. Bethwaite, F. D., E. J. Stnith, J. A. Wal"bJ.rton, and 1::. J. Heffernan, 1966: Effects of seeding isolated cumulus clouds with silver iodide. J. Appl. Meteor,S, p. 513. B:l.gg, E. :t., 1953: The supercooling of water. Proc. Phys. Soc. B, 66, p. 668. Booker, D. R., D. C. &11, H. E. Hart, and L. W. Cooper, 1969: Evidence of severe storm rotational characteristics obtained from superpressure balloon tra_ jectories. Preprints of Sixth Conference on Severe Local Storms. Amer. Meteor. Soc., pp 32-37. Byers, H. R., 1942: Nonfrontal thunderstorms, Dept. of Meteor., Univ. of Chicago, Misc. Repe. No.3, 22 pp. Chagnon, S. A., Jr., and D. W. Staggs, 1970: RBI radar_h4il relations for weather m:xiification experiments. Illinois State Water survey, Final Report on NSF GA_4681. Davis, L. H., 1966: Alterations of buoyancy in cumuli. PhD. Thesis, The Penn. State Univ., University Park, Penn., 98 W. Donn~f igI-~H.tr~'~i_~;~:~~~n::oi~t~~~s~d~;:~~~hn~~c~~:;~~o~:f~i:ncies function of LW::::: and generator flame temperature, a preliminary report. The Journal of Weather Modification, Weather Modification Assoc., Vol. 2, No.2, W·1SS_164. Elliott, R. D. and E. L. Hovind, 1964: The water balance of orographic clouds. J. of Appl. Mateor., 3, pp. 235_239. ----.:R.::::.::•-:•::::rc::;:h~in~:6;~po;~O~;;:~~~S~~~: ~~1~~:c~0=~~~~~=;~ ~~~tric Elliott, R. D•• I:. J. Br01ol1"l, and L. O. Grant, 1971: Transactions of a seminar on extended area effects of cloud seeding, Feb. 15_17, 1971, Santd Barbara, Cuifornia. Fujita, T•• and H. Grandoso, 1966: Split of a thunderstorm into anticyclonic and cyclonic storms and their motion as determined from numerical roodel experiments. Satellite and ~sO_IIlilteor01ogy research project, Dept. of Geophysical SCi. Univ. of Chic.ago Res. Paper 162. October 1966. Garrahan, B. M•• B. J. Carley, and T. J. Henderson. 1969: Areal inflow mapping and related time-lapse photography. Volume 11: Report on hailstorm models project. Report 69.1 lG' Grant No. G.A. _935, South Dakota School of Mines & Technology. Grant. L. 0., and others, 1969: An operational adaptation program of weather modification for the Colorado River Basin. Interim Report to Bureau of Reclamation, Dept. Atmos. Sci., Colorado State University. Gunn, R., and G. D. I:inzer, 1949: The terminal velocity of fall for water droplets in stagnant air. J. Meteor., 6. pp. 243.248. Haltiner. G. T., 1959: On the theory of convective currents. Te11us, 11, pp. 4.15. Heymsfield. A., 1972: The crystal terminal velocities. J. of Atm. Sci•• 29. pp. 1346-1357. Hirsch. J. H., 1972: A numerical clQud mdel • its use during Project Cloud Catcher. Third Conference on weather Hodific'ltion. Am. Meteor. Soc •• Rapid City. South Dakota, June 26_29, 1972. Hosler, C. L., D. C. Jensen, and L. Golc!shlak, 1957: On the aggregation of ice crystals to form snow. J. Meteor., 14, pp. 415.425. -"""""'the""dyn=a:u:; ~D::~~~~~~s ~~~io=. E. ~ :~~tl;~~:ar:F~:~s~;=~on of NSF GP_4743. Pennsylvania State University. Houghton, H. G., ani H. E. Cr(llOOr, 1951: A theory of entrainment in convective currents. Jour. Met •• 6. pp. 95.102. Howell, W. E., 1972: Discussion of "weather modification in Arizona in 1971" by Herbert A. Osborn _ and addendum thereto. U. S. Bureau of Reclamation, Division of Atmospheric Water Resources Management, Denver, Colo .• June 1972. I:ess1er, E., 1967: On the continuity of water substance. Tech. Hem. Iertm-Nssl 33. 125 pp. :toenig. L. R., 1968: Some observations suggesting ice multiplication in the .atmosphere. J. Atmos. Sci.. 25, PP. 460.463. I:oscieleki. A., and A. DelUlis. 1972: Seeding effects in convective clouds in western South Dakota. Third Conference on Weather Modification, Am. Meteor. Soc., June 26_29, 1972, Rapid City, South Dakota. I::umai, M., 1961: Snow crystals and the identification of the nuclei in the northern United States of America. J. of Meteor., Vol. 16, No.2, April 1961, pp. 139.150. LangIIlUir, 1., 1948: The production of rain by chain reaction in cumulus clouds dt temperatures above freezing. J. MeteQr., 5. pp. 175.192. Levine. J., 1959: Spherical vortex theory of blbble-1ike flX)tion in cumulus clouds. J. Meteor•• 16, pp. 653_662. Mall:us, J. S., ard G. Witt, 1959: The evolution of a convective element: a numerical calculation. The AtlOOaphere and Sea in Motion. Rockefeller Insti­tution Press, New York, pp. 425.439. Malkus, J. S., 1960: Recent developnent in studies of penetrative convection and an application to hurricane cumulonimbus towers. Cumulus Dynamics, Pergamon Press, pp. 65_84. Marshall, J. S., and W. MclC. Palmer, 1948: The distrirution of raindrops with size. J. Meteor., 5, pp. 165_166. Mason, G. J., and R. Emig, 1961: Calculations of the AScent of a saturated bJoyant parcel with lllixing. Quart. J. Roy. Meteor. Soc., 87, pp. 212_222. H4thews, P. A., 1972: A m:xJel of successive conwctive cloud developoent. Third Conference on Weather Modification, Am. Meteor. Soc., Rapid City, June 26_29, 1972. McNaughton, P. L., 1972: An area cloud seeding experiment in the north of Rhodesia, 1970_71. Rhod. J. Agric. Res., 10, pp. 91_103. Mossop, S. C. -. and A. Ono, 1969: Measurements of ice crystal concentration in clouds. J. Atmos. Sci., 26, pp. 130,:,,137. Murroy, F. W., and C. E. Anderson, 1965: Numerical simulation of the evolution of cumulus to_IS. DouglAS Report 1SM_49230, Missile and Space Systems Division, Douglas Aircraft Company, Inc., Santo Monica, Calif., 97 pp. Nakaya, U., and T. Terada, 1935: Simultaneous observations of the mass, falling velocity and form of individual snow crystals. J. Fac. SCi., Hokkaido Univ., Japan, Ser. 2, I, p. 19I. --' 1954: Snow crystals, naturl._""'_lOOC: Fig. 2.1_7 Isochrones of peak hourly precipitation during B~r in South Dakota (csr). 04 02 ........ _-- '9- ... _ _ "'_ (0__• -::=-.-::-.=----- ..._ ... - IO$TM_TloII:lCOOl:M11MWUlIOIbr-.ol:ZOO€ t • , Fig. 2.1_8 Isochrones of seoondary peak in hourly precipitation during sun:mer in South Dakota (cgrl. 2.2 Cloud Catcher Observations The Cloud Catcher data provide a unique set of detailed observations of. a con~ trolled seeding test operation applied to individual clouds. These data have been examined independently from lAS analyses, in an attempt to establish a useful synoptic climatology of cloud characteristics and of expected seeding effects. Before going further into this, some of the b5ses for evaluation of Cloud Catcher seeding results will be examined. Figure 2.2.1, taken from lAS 7205, is a scatter diagram plot of seeded and not seeded radar rainfall against clou:::l. depth. Although the scatter of points i6 large, the difference J:etween the AgI seeded and not seeded regression lines (straight lines on the log.log plot) is significant at the 7'f" level, and that between the combined Agl and salt seeded versus the not seeded regression is signifi~ cant at the Sf. level. The Agi seeded and not seeded regressions cross at ~ depth of ab::lut 28 r.ft, which represents an average cloud top height of 38 J:.:ft. This crossover is in keeping with the qualitative, conceptual cloud models discussed in Chapter 1. At greater depth negative microphysical effects prevail over any positive dynamic effects. This type of analysis does not provide a basis for sorting dynamic from microphysical effects. The reason for tr.is is worth detailing. Su;::p::lse the seeded cases have more ::-ain than non.seeded cases, relative to their clow depth. If there were no seeding produced change in cloud toP. then the net effect would be that shown by the arrow AS in Figure 2.2.2. The path of the arrow is direct from. the not seeded regression line to the seeded regression line. If, on the other hand, the seeding also raises tops, thus increasing the depth, then 'We should expect the increase due to greater depth to be that represented by the arrow N;. If this were the only precipitation effect of seeding, then it would not J:e revealed by this type of regression analysis at all. Only the additional non.dynamic effect re. presented by the arrow CD would J:e detected. Should the regression analysis indicate less precipitation for the seeded cases then a positive dynamic effect would still go Wldetected, even though it might, in fact, coW'\terbalance the negative microphysi. cal effect. Independent data are needed to establish the dynamic effect. Unfortunately. Cloud Catcher cloud top data a.1one cannot be used to make such a determination. However, 3. recent lAS study has sho'Wrt conclusively that echo area is greater for seeded than for not seeded clouds no matter what their height or depth. TMre is a J:elief that clou:::l. toP6 do not rise more thdn a 1000 feet or so with Agol seeding in this area, and that the area expansion is more significant. ~h1s aerial expansion 2.2;'1 • NO SEED IOPOO .... SILVER IODIDE o SALT j 1,000 gI ::J ri! 100 z II: ~ u 10 f- ''"" f- 5 Fig. 2.2Ml Scatter diagram eom.paring lognithrn of nddr..est1ma.ted rainfall and loguithm of cloud depth for no_seed, silver iodide seed, and ult seed cases. PRECIPITATION CLOUD DEPTH -~- Fig. 2.2_2 Dynamic effects in a precipitation-eloud diagram. 2.2_3 is nevertheless a reflection of the dynamic effect of seeding and has been noted in various reports on cloud seeding going back to the late 1940's. A seedi~­produced enhancement in area will show up as an increase in volUll'l9 precipitation. Any such area increase, not reflected in a rise in the top, will then be revealed in regression type llJ'Iilysi$ using cloud top height or depth as llJ'I independent vari_ able. We are therefore left with some question aa to just what extent dynamic seed. ing effects are covered by the regression analysis, but dpparently the major total effect of seeding is represented in. them. A regression line has little meaning near the ext1'eIl:rBS of the data set. The true relationship may very well be ew:'ilinear, not linear. The indicated crossover, if it exists, can be established only aft.er the accumulation of more da.ta. This can be said also of any crossover, or approach to zero effect, at some minimum depth. The overall statistical l:est fit of voluce precipitation to cloud. depth is given by the power law: precip""' J:: (depth) 4. The total rainfall given as a volume lOOasure (acre feet) should l:e proportional. to the cloud area, its depth (if the peak uprll'aft scales with the depth), and the duration of the cloud.. Since the area is related to depth sqwu-ed, a.nd duration to depth, one should expect volume precipita­tion to vary as the 4th power of the depth. If precipit4tion efficiency 41so varies with depth, then a 5th power law should prev4il, at least in the lower range of tops. If efficiencies diminish when tops exceed 40 Ut, a lower power law wouJ.d prevail there. The point rainfall at a single gage on the ground, on the other hand., should vary roughly as only the depth because it does not integrate an area, and it does not follow along under a cloud during its duration. A gage network would vary rough. ly as the 4th JX)wer, but it would ordinarily catch rainfall of many cells, not a single cell. The process of drawi~ a valid inference from available data by statistical pro­cedures can l:e more scientificdlly satisying when the procedure includes the valida. tion of a numerical model of the process involved. Numerical models are indeed playing an increasingly significant role in evaluation of seeding results, and in the design of new experiments. Cloud top height and other variables have been incorporated into a multiple regression in seeking for 4J\ operationally useful definition of precipitation, The first three years of Cloud Catcher data (1969.70.71) were examined for this purpose. The cloud top W4S used as a parameter because it is conmonly observed. Regressions were formed of loglO of the radar volume precipitation (RVP) against maximum. radar cloud top and combinations of aerological parameteIl5 representing the energy avail_ able for convection. The correlation between the loglO RYP and radar cloud ~op was strong: .840. The correlations between RVP and aerological par~terll' were weak.. The bell't correlation wall' .270 for net geostrophic warm advection between the 850 and 300 nib levels as detennined from the rooming sounding. The regression equation is: LogIO RVP "'" -2.1319 + .1302 (2:rl - .0022 (WA) where precipitation is given in acre feet, and 2:r - m.axiD:twn radar cloud top WA = warm advection parameter. Geostrophic warm advection occurs where the wind veers aloft, while cold advec. tion occurs where it backs aloft. The relationship is found in standard texts. The formula used here is: where s~ is the ll2an wind speed in m/sec between levels 2 and 1, lJ. gi is the angular turning in degrees of the wind in going from 2 to 1, and f is: f~gg = 5.2, ffgg = 3.0, ~gg = 4.5, f~gg = 3.5. Advective warming in the troposphere represents a net de_ stabilizing factor, since there is no corresponding warming alxlve the tropopause. Negative values of the warm advection par~ter represents cold advection in the tro_ popause, generally a stabilizing factor. The warm advection is therefore a lmdSure of the energy becoming availttble for convection through synoptic scale rootions 4I\d thermal distribJ.tioIlS. Figure 2.2_3 shows the regression predicted value of precipitation on a 1lUUC1mum radar cloud top-warm advection graph. In going through the range of cloud tops from 15 to 45 thousand feet the precipitation increases through three orders of IM.gnitude. In going through the range of increllSing warm air advection it decreases by alxlut a third of &XI order of magnitude. The decrease in precipitation with increased energy available for convection at first seems contradictory. However, with roore energy for convection there are stronger updrafts, and less tim for growth of precipita_ tion particles, thus increasing the possibility of a negative dynamic effect. Also, an inefficient type updraft profile lM.y be more prevalent due to the sounding char_ acteristics. The overall implication is that when there is a great deal of energy available for convect ion the atmosphere responds by producil'lQ the Imst rapid 40 1000 35 "AX. CLOUO TOP (J< it.) 30 100 25 \ \ \ \ \ '8 --/ / / / ",~ \ ./ / ./ ' ....... '- 6 _--- /' / --......... -...---"-"-"---2-" --------- F1g. 2.2.' Rodor val_ preciPltation in acre.feet accon/ing to multiPle _re.aion 6eq1u:aftti~o50n W('JaI ohlli..d...,ll-i.nee). The doehed line. are the f_cy of Cdse. in n'-r Per 20 '5+-------.-----__.-- -r-- -:-, -'00 -50 +50 +'00 COLO AOVECTlON --t WARM ADVECTION stabilization possible~ that is warming aloft (by convective updrafts) and cooling below (by rain' evaporation). As pointed out in Chapter I~ this type of stabilizing IDeChMiam is M inefficient precipitation (to the groW\dl producer. Figure 2.2.3 also shows the frequency of cases expressed as the number falling within a 6 J:ft.SOWA square in three years of not-seeded and salt seeding cases com­bined. Salt seeded cases are inclOOed as salt seeding presumably has no effect on cloud tops. A distinct peak in the frequency of occurrence falls at 28 to 30 Kft and 0 to 25 WA indices. Apparently the higher clouds are associated with greater warm advection than the lower, more mlDerous ones. Cold advection has a limiting effect on the developaent of tall clouds. Correlations between RIfP and direct IDeasures of soW\ding instability were poor. The morning soW\ding was worse than the afternoon sounding. Table 2.2-1 relates cloud and air mass characteristics to the warm and cold ad_ vection types~ and to convenient cloud top height categories. The IDeM wind direc­tions and speeds show a distinct pattern with respect to bacl:::ing and veering of the winds aloft~ .s.s might be expected from the manner in which the advection categories were selected. There was a considerable variation in direction between cases at the 850 mb level~ as mig~t be expected, but at higher levels directions tended to be con_ sistent. In the mean~ most of the backing in cold advection occurs from 850 to 700 mh. In warm advection most occurs from 850 to 700 also~ with a layer of backing from 700 to 500. The number of the various cloud types coded by lAS fall into distinct cloud top zones. The zone for tops of 28 Kft or higher produces the highest precipitation and contains all of the lAS code 5 and 6 cases; .Code SEf' is mesoscale clouds with back feeders~ and 6 represents a super cell. This same zone contains five out of eight lAS code 3EF clouds. Code 3EF clouds are steady state back feeders. The zone with tops under 28 J:ft contains 12 out of 13 transient clouds~ consisting of code 1 and 2 which are of short duration and consist of one or two O6lls. All six front feeders fall within the sector aboye 28 Kft and zero or positive warm advection. Front feeders~ usually code 5FF and sometimes code 3FF~ are lUldoubt. edly quite efficient in stabilizing the atmosphere, but are apparently inefficient in producing precipitation. Table 2.2_1 also summarizes the moan difference between seeded and not seeded precipitation by cloud top advection categories. The differences -were computed on the basis of averages of seeded and not seeded residuals from the multiple regression 2.2.7 T4bl.e 2.2_1. Slmmary of wind. cloud code. and seeding characteristics associated with advection type and oloud height. Warm Advection Cold Advection {300 mb 272 _ 18 263 _ 26 Mean wind 400 lab 261 - 15 266 _ 18 direction and 500 mb 262 _ 12 277 - 15 speed (m/s) 700 mb 279 • 7 284 _ 10 850 mb 217 - 7 315_ 7 SFF and 5fT _ 6 3fT and 5FF _ 0 { 28 K:ft height SIF and 6EIF _ 3 SIFa.nd6!F_2 or greater 3EIF _ 6 3ElF _ 4 NllIllber of land2_1 land2-0 type code 3FF and 5FF _ 0 3fTandSFF_0 Under 28 i::ft SElF and 6BF _ 0 SElF and 6BF _ 0 height SElF _ 2 SElF _ 1 land2_5 1 and 2 _ 7 1 Over 40 Kft _ .39 AgI seed _ not height seed.ltl81Ut 28 to 40 Ut residuAl. height + .51 - .27 differences Under 28 Kft height + .53 + .89 2.2_8 values. In all of the under 28 J:::ft ooxes the effects of the seeding appears to be an increase in precipitatton by one half to .9 of an order of magni tude (about +30 to +8~). In the 28 to 40 lit ~ there is found an indicated decrease for silver iodide seeding under cold 4dvection, l:::u.t an increase for wam advection. This suggests that it is worthwhile seeding clouds taller than 28 [ft provided there is deep waIlll advection. The seeding appears to make up for some of the natural inefficiency. There are only two cases in the above 40 I::.ft boxes, all in warm advection and the indicated decrease is therefore in doubt. It should be pointed out that the maximum cloud top temperature is pro~y the key parameter, not the maximum cloud top height. The height values shown apply to midsUll'UtlElr. An a
 back to results : previous : next