ADVANCES IN ATMOSPHERIC SCIENCES, VOL. 25, NO. 1, 2008, 31{42 Simulation of Terrestrial Dust Devil Patterns GU Zhaolin¤1 (顾兆林), QIU Jian2 (邱 剑), ZHAO Yongzhi3 (赵永志), and LI Yun2 (李 云) 1Department of Environmental Science and Technology, School of Human Settlements and Civil Engineering, Xi'an Jiaotong University, Xi'an 710049 2School of Energy and Power Engineering, Xi'an Jiaotong University, Xi'an 710049 3Department of Chemical Engineering, Tsinghua University, Beijing 100084 (Received 12 June 2006; revised 29 September 2007) ABSTRACT Introducing the surface properties [initial vortex, ground temperature and surface momentum impact height (SMIH)] for the boundary conditions, dust-devil-scale large eddy simulations (LES) were carried out. Given three parameters of initial vortex, ground temperature and the SMIH based on Sinclair's observation, the dust devil physical characteristics, such as maximum tangential velocity, updraft velocity, pressure drop in the inner core region, and even reverse flow at the top of the core region, are predicted, and are found to be close to the observations, thus demonstrating the ability of the simulation. The physical characteristics of di®erent modeled dust devils are reproduced and compared to the values predicted by Renno et al.' theory. Even for smaller temperature di®erences or weaker buoyancy, severe dust devils may be formed by strong incipient vortices. It is also indicated that SMIH substantially a®ects the near-surface shape of terrestrial dust devils. Key words: dust devil, dust storm, large eddy simulation (LES) DOI: 10.1007/s00376-008-0031-7 1. Introduction rising plumes in the CBL (e.g., Battan, 1958; Ryan and Carroll, 1970; Cortese and Balachandar, 1993; Shapiro Dust is a main component of atmospheric aerosols. and Kogan, 1994; Renno et al., 1998; Michaels and With several special wind systems, such as dust Rafkin, 2004) called \swirling rising plumes" (SRPs) storms, dust devils etc., the mineral dust particles orig- (Zhao et al., 2004). There is a super-adiabatic lapse inate mainly from the great desert areas of northern rate near the ground surface, which is the driving force Africa, Southwest America, and Asia. Mineral aerosols for dust devils. Near the surface, warm air parcels spi- play a key role in atmospheric chemistry, ecology, and ral toward the centre while absorbing heat from the the Earth's radiative balance (Wang et al., 2004; Shen surface. The radial inflow of warm air into the rising et al., 2006). plume results in the concentration of ambient vorticity. With a low pressure and a warm core, dust dev- Typical observed diameters of dust devils may de- ils are vertical vortices that occur in the atmosphere's pend strongly on geography. For Arizona dust dev- convective boundary layer (CBL), usually short in du- ils, an average diameter is likely to be in the order ration. Dust devils can swirl dust, debris and sand of tens of meters (Sinclair, 1973). The heights of dust to high levels. They are concentrated vortices, oc- columns in dust devils are usually less than 600 m but, curring on various surfaces of the Earth (Vatistas et in desert conditions with deep mixed layers, and the al., 1991). Their frequent appearance in some areas thermal updrafts associated with them have been ob- suggests that they are signi¯cant in the evolution of served to extend as high as 4500 m (Sinclair, 1966). dust storms (Greeley et al., 2003). However, this is so Renno et al. (1998) proposed a thermodynamic theory far only speculation, without direct observational evi- of dust devils and presented a range of thermodynamic dence (Cantor et al., 2002). Dust devils may arise from parameters against Sinclair's (1966) observations. ¤Corresponding author: GU Zhaolin, [email protected] 32 SIMULATION OF TERRESTRIAL DUST DEVIL PATTERNS VOL. 25 Leovy (2003) pointed out that computer simula- to many factors, such as soil cover, physical char- tions have the potential to be an important tool for acteristics, composition, topographical features, and interpreting laboratory and ¯eld observations of dust weather (Renno et al., 2004). Based on the dust- devils. Computer flow simulations have been per- devil-scale quasi-steady LES mathematical model, the formed to gain insight into the dynamics of bound- present study aims to reproduce a variety of observed ary layer vortices. In a laboratory simulation of tur- dust devil shapes by adjusting three key parameters, bulent convection, Willis and Deardor® (1979) found namely the incipient vortex, SMIH (surface momen- that dust-devil-like vortices were most likely to occur tum impact height), and ground surface temperature. at the vertices of the cellular convective pattern, where The quasi-steady LES model and the choice of three vertical velocity reached its peak. parameters for boundary conditions are described in Kanak et al. (2000) found the existence of vertical section 2. The air flow ¯eld of a modeled dust devil is vortices in the CBL by means of large eddy simula- simulated and analyzed in section 3. Six cases of mod- tion (LES). Their model used a 35-m horizontal grid eled dust devils are reproduced in section 4 and com- spacing, which is approximately the average diameter pared with Renno et al. (1998) thermodynamic theory of extra-large dust devils, and may be classi¯ed as a of dust devils. A summary is given in section 5. CBL-scale model. Although this model could simulate the initiation of weak vertical vortices of dust dev- 2. LES model and boundary conditions ils, it cannot be used to study the ¯ne flow structure, or the influence of surface and atmospheric parame- Figure 1 shows several kinds of terrestrial dust dev- ters on dust devils. Therefore, smaller scale models ils observed on di®erent surfaces, e.g., bushes, shrub are needed to solve these problems. In Kanak's re- layers, deserts, and even air¯elds. The typical upside- cent work (2005), a high-resolution (2-m grid spacing) down cone-like dust devil is common in nature. The LES of the CBL was performed in order to examine dust devil with spray|regular dust cylinder in the in- the physical characteristics of dust-devil-scale vortices. ner core with sprays of flat, upside-down cone shapes Although the general features of the simulated dust- near the ground|might have high tangential veloc- devil-like vortices agreed well with observations, the ity and near-bottom momentum roughness. Bowl-like model was limited in a number of aspects, such as the dust devils occur on smooth ground. The inflow air pressure drop inside the simulated vortex core was not layer near the ground is very thin, and the bottom of as great as observed, and the simulated vortex wind the dust devil is relatively flat. If the ground is very speeds were less than 10 m s¡1. It was also suggested smooth (such as in airports near deserts), no obvious that initial conditions of the simulation might have an draught spirals out, and the dust devil has a regular important influence on the formation of vertical vor- column shape. With a low pressure inner core and tices. swirling plume, dust particles might become airborne, Zhao et al. (2004) compared a CBL-scale simula- making the dust devil vortex visible if the surface is tion for convection cells with a dust-devil-scale simu- composed of loose material. lation for the air flow of dust devils and found that, It follows from the dust devil images in Fig. 1 that given appropriate initial conditions, such as heat flux the evolution, shape and intensity of dust devils may and initial temperature ¯eld, the convection cell could depend on surface properties, which might be present be initiated, and thus easily allowing the development within the boundary conditions and initial conditions of air flow for dust devils. Di®erent states of convec- in the current simulation. In the following paragraphs, tion take di®erent amounts of time to develop dust the three parameters (background vorticity, momen- devil air flow ¯elds. A three-dimensional, unsteady, tum roughness and ground temperature) are justi¯ed high resolution model is presented to investigate the as the surface properties. air flow evolution of a modeled dust devil, using ¯ne A number of investigators have proposed or pro- grids|the radial grid spacing smoothly stretched from vided evidence that larger scale convective circula- 0.1 m at the center to 4 m, and the vertical grid spac- tions, that are not initially rotational, can provide ver- ing stretched from 0.1 m on the surface to around 10 m tical and/or tilted horizontal vorticity (e.g., Ryan and at the top. Smagorinsky's (1963) scheme was used for Carroll, 1970; Cortese and Balachandar, 1993; Shapiro parameterization of the sub-grid viscosity. This model and Kogan, 1994; Shapiro and Kanak, 2002; Michaels might be a dust-devil-scale model. The simulation re- and Rafkin, 2004; Kanak, 2005). The vortex size of sults show the air flow shape of dust devils, which are a dust devil is proportional to the value of the back- strongly a®ected by ground surface friction. ground vorticity (Renno and Bluestein, 2001). Hence, Many studies have shown that on a small scale, in dust-devil-scale simulation, the background vortic- the shape and intensity of dust devils are sensitive ity, or incipient vortex, could be selected as a param- NO. 1 GU ET AL. 33 Fig. 1. Patterns of terrestrial dust devils. Image sources: (a) http://www.aeriane.com/images/; (b) http://www.inflowimages.com/code/showimage.asp?img=../chasereports/dustdevils/dd3.jpg; (c) http://www. saweather.net/photographs/vortex/; (d) http://www.weatherpictureoftheday.com/. 313K ifz ≤ 800m Tin = 313+ 0.015（z − 800) K else z p= gβρ ( T ( z ) − 313) dz ∫0 in ∂(rvr )/ ∂ r = 0 ∂(rvz )/ ∂ r = 0 ∂(rvt )/ ∂= r 0if v r > 0 − z/ C 2 -1 vCet=− 1(1 )msif ⋅ v r ≤ 0 Fig.