2140 MONTHLY WEATHER REVIEW VOLUME II9The Effect of Heterogeneous Soil Moisture on a SummerBaroclinic Circulation in the Central United States*JEROME D. FASTt AND MICHAEL D. MCCORCLE~Iowa State University, 2mes. Iowa(Manuscript received 17 July 1990, in final form 20 March 1991)ABSTRACT Thermally induced drculations, similar to sea breezes, may be established in the presence of horizontalgradients in soil moisture, soil type, vegetation, or snow cover. The expense of extensive observational networksand the relatively small-scale drculatious involved has made examining these drculations very difficult. Recentnumerical studies have indicated that sharp gradients in soil or vegetation properties may induce mesoscalecirculations in the absence of synoptic forcing. The current study employed a three-dimensional, hydrostatic mesoscale model to evaluate the effects ofhorizontally heterogeneous soil moisture and soil type on the passage of a summer cold front in the centralUnited States. Grid-scale condensation, precipitation, latent heat release, and cumulus convection are not accounted for in this model; moisture was affected only by advection, diffusion, and evaporation. Numericalsimulations demonstrated that evaporation of soil moisture significantly affected the boundary layer structureembedded in the baroclinic circulation. Although the position of the front was not altered, the thermal andmomentum fields were effected enough to weaken the front near the surface. Evaporated soil moisture wasadvected ahead of the cold front, far from its source R~gion. Moisture convergence was significantly enhancedin several locations, indicating that soil moisture may play an important role in modifying the spatial distributionand intensity of precipitation. The impact of surface inhomogeneities in soil moisture and soil type on the atmosphere is expected to behighly dependent on the particular synoptic conditions.1. Introduction The partition of energy between the sensible andlatent heat fluxes at the surface is a fundamental factorthat determines the evolution of the planetary boundary layer. The intensity of circulations in the boundarylayer is directly related to the magnitude and horizontalvariation of the sensible and latent heat fluxes. Thecomponents of the surface energy budget at the earth'ssurface can be significantly altered by terrain inhomogeneities. Thermally induced mesoscale circulationsforced by terrain inhomogeneities such as sea and landbreezes, mountain and valley winds, and urban circulations have been studied extensively by observational and numerical techniques. There has been an increasing recognition in the literature that other inhomogeneities in land character * Journal Paper No. J-14116 of the Iowa Agriculture and HomeEconomics Experiment Station, Ames, Iowa. Project No. 2804. t Present affiliation: Westinghouse Savannah River Company, Savannah River Laboratory, Aiken, South Carolina. t Present affiliation: Science Applications International Corporation, McLean, Virginia. Corresponding author address: Dr. Jerome D. Fast, WeslinghouseSavannah River Company, Building 773A, AI013, Aiken, SC 29808.c 1991 American Meteorological Societyistics may cause important circulations to develop.Thermally induced circulations established near horizontal gradients in soil type, soil moisture, vegetation,snow cover, or cloud cover, may also produce circulations similar in structure and magnitude to seabreezes. A nonclassical mesoscale circulation (NCMC)was defined by Segal et al. (1989) as a circulation produced by soil=moisture gradients, to distinguish it fromthe sea-breeze phenomena. This term will be used inthis paper as well. Most studies have focused on the potential impactof simple discontinuities in soil type, soil moisture, orvegetation, while neglecting synoptic forcing and thre~dimensional effects (Avissar and Pielke 1989; Mahfoufet al. 1987; Ookouchi et al. 1984; Pinty et al. 1989;Segal et al. 1988; Yan and Anthes 1988). Since thesestudies neglect synoptic forcing, the impact of NCMCson the boundary layer could be overpreclicted in manysituations. Large horizontal gradients in land characteristics assumed by these studies are resolved by a gridspacing of 5-10 km. Sharp horizontal gradients in soilmoisture are not entirely unrealistic. Strong soil-moisture contrasts can be produced by persistent weatherpatterns, convective precipitation, topograp~c influences, and agricultural irrigation. Synoptic forcing may be large enough to mask orsuppress NCMCs in numerical simulations as suggestedSEPTEMBER 1991 JEROME D. FAST AND MICHAEL D. MCCORCLE 2141by Segal et al. (1989); however, it is possible thatNCMCs interact with larger-scale cimulations underthe proper circumstances. Mesoscale phenomena primarily forced by instabilities in larger-scale synopticsystems, such as low-level jets, fronts, and convectionbands, could be affected by NCMCs. McCorcle (1988)and Fast and McCorcle (1990) initialized a coupledearth-atmosphere numerical model with typicalspringtime synoptic conditions to examine the effectof inhomogeneous soil-moisture content on the GreatPlains low-level jet. The magnitude and structure ofthe simulated nocturnal jet was very sensitive to sharpsoil-moisture gradients. The Penn State-NCAR Mesoscale Model was used to examine the influence ofsoil-moisture variation in the southern Great Plainsand the effect of the Mexican plateau on the evolutionand structure of the dryline, elevated mixed layer, andthe boundary layer in Lanicci et al. (1987). Vadablesoil moisture in the southern Great Plains was foundto be important in determining differential heating andgeneration of low-level instability in the prestorm environment. While most studies of NCMCs assume homogeneousland characteristics within a grid element 10-100 kmwide, large inhomogeneities of soil type, soil moisture,vegetation, and soil type are frequently observed onthis scale. Wetzel and Chang (1988) reported that formesoscale and global numerical models with a gridspacing on the order of 100 km, the subgrid-scale variability of soil moisture may be as large as the totalmean available moisture content in a particular region.The effects of soil moisture on the boundary layer maybe relatively transient because of the subgrid vadabilityin evaporation rate. Vegetation effects also may betransient because evapotranspiration depends on soilmoisture, density of vegetation cover, and stomatal,internal, and root resistance. Soil type remains constantfor a particular location, but can vary substantially overa grid element. Avissar and Pielke (1989) and Wetzeland Chang (1988) have addressed these problems byproposing subgrid-scale heterogeneous surface forcingparameterizations. The expense of extensive observational networks andthe relatively small-scale circulations involved havemade observing NCMCs very difficult. Currently, it iseasier to simulate the potential impacts of NCMCs withnumerical models. The partition of energy between thesensible and latent heat fluxes at the surface is of primeimportance in achieving accurate simulations ofNCMCs in numerical models. Although mesoscale,synoptic, and climate models have been employingmore complex surface energy budgets, the particularparameterization of the surface energy budget may significantly effect the magnitude of smaller-scale circulations, such as NCMCs. Avissar and Pielke (1989)and Segal et al. (1988) have shown that a more complexrepresentation of the soil layer and vegetation can produce significantly different results from simpler parameterizations. Additional research is necessary todetermine how complex surface forcings need to beparameterized in order to adequately simulate the effects of NCMCs. It is apparent that more simultaneousobservations of meteorological and biophysical parameters are needed to understand the complex relationships at the earth/atmosphere interface and to verify two- and three-dimensional models. This was themotivation for the recent FIFE (Sellers et al. 1988)and HAPEX-MOBILHY (Andre et al. 1986) experiments. Some of the initial data from HAPEX-MOBILHY can be found in Andre et al. (1988) and Pintyet al. (1989). Segal et al. (1989) presented a set ofobservations taken over irrigated areas in northeastColorado along with a few three-dimensional simulations of the flow field. Despite the lack of soil-moisturedata, the possible effects of NCMCs on relatively largerscale circulations can be determined with hypotheticalsoil-moisture distributions as in Lanicci et al. (1987)and McCorcle (1988). This research will attempt toevaluate the intensity and the horizontal and verticalextent of NCMCs resulting from soil-moisture and soiltype distributions in the central United States. Thisinvestigation will differ from previous studies by determining whether a specific NCMC can significantlyeffect a baroclinic mesoscale circulation. By comparingsimulations with, and without, any horizontally inhomogeneous land properties, an estimate of the effectof NCMCs on baroclinic circulations can be obtained.The thermal and moisture interaction of NCMCs inthe boundary layer with circulations in the free atmosphere will also be evaluated. The diurnal boundarylayer may be altered enough to effect the structure oflarger-scale weather patterns such as low-level jets orfronts. Moisture-divergence fields may be altered byNCMCs to change the spatial distribution and intensityof convective precipitation. The coupled earth-atmospheric numerical modeldescribed by McCorcle (1988) has been modified toincorporate baroclinic initial conditions. Section 2outlines the development of the present mesoscalemodel used to study NCMC's embedded in barocliniccirculations. An observed summer baroclinic circulation of afrontal passage in the central United States is used toinitialize the numerical model. This front movedthrough the central United States during 21-23 June1989 and is described in section 3. This system produced three regions of scattered showers in the GreatPlains, with a few stations reporting moderate rainfall.The surface and upper-level characteristics of thermal,moisture, and momentum fields are presented. The role of soil-moisture and soil-type parameterizations on boundary-layer and mesoscale circulationsare examined by performing several control and sensitivity experiments. The results are presented in section4. Several simulations are performed with no synopticflow imposed to examine isolated NCMCs produced2142 MONTHLY WEATHER REVIEW VOLUME 119by various soil-moisture and soil-type distributions.Then synoptic flow is imposed to examine potentialeffects of NCMCs on the baroclinic circulation. Difference fields are calculated for several variables bysubtracting the results of the control simulations fromthe sensitivity simulations. Section 5 presents the conclusions of this study.2. Numerical model A hydrostatic, coupled earth-atmosphere numericalmodel described by McCorcle (1988) and Fast andMcCorcle (1990) has been modified to simulate baroclinic mesoscale phenomena. The current version ofthe model assimilates observed surface and upper-airdata to the three-dimensional numerical grid for theinitial conditions.a. Governing equations The vertical coordinate for the atmospheric governing equations has been transformed from an orthogonalto a nonorthogonal, terrain-following vertical coordinate. The functional form of the vertical coordinate ais defined by s( Z - zoI o- = , (1) \s - zo/where z is the Cartesian vertical coordinate, s is theconstant height of the model top, and zo is the elevationof the terrain. This type of vertical coordinate has beenused by several mesoscale models reported in the literature (Pielke 1984). In addition, the model employs a lower layer ofnodes in the domain that are logarithmically spaced.The governing equations are transformed in this layerto a new vertical coordinate ~ defined as ),where a is a constant and Zo is the roughness length ofthe surface. This transformation is retained from theboundary-layer model formulation as described inPaegle and McLawhorn (1983). The atmospheric portion of the coupled earth-atmosphere model is governed by an anelastic, hydrostatic system of equations. The governing equationsare transformed into the nonorthogonal grid systemfor the atmospheric portion of the model by a procedure similar to Pielke (1984) and are listed in the Appendix.: To more precisely predict surface forcings, the modelincorporates forecasts of both moisture and heat fluxeswithin the soil by using a soil-moisture forecast methodsimilar to that employed by the Air Force GeophysicsLaboratory Soil Hydrology Model as described byMahrt and Pan (1984) and Pan and Mahrt (1987). Aprognostic equation for the volumetric soil water content n is used that contains terms for hydraulic conductivity, hydraulic diffusivity, evaporation, transpiration, and dewfall. A soil heat-flux equation is used todetermine the soil temperature T~oil. Soil temperatureforecasts are dependent on the thermal conductivity,which is.highly dependent on soil moisture. The present numerical model does not contain parameterizations for grid-scale condensation, precipitation, latent heat release, or cumulus convection.Evaporated soil moisture is simply advected by thewind field and there are no feedback processes thatcould reduce this atmospheric moisture. The effect ofheterogeneous soil-moisture distributions on atmospheric circulations can still be examined, while neglecting these feedback processes as shown by Avissarand Pielke (1989), Mahfouf et al. (1987), Ookouchiet al. (1984), and Segal et al. (1988). The prognostic and diagnostic equations are solvedby a combination of finite-difference and finite-elementtechniques. The advection terms are approximated bya fourth-order scheme as described in Tremback et al.(1987). The vertical diffusion terms are discretized bya finite-element technique based upon Galerkin approximations. A Crank-Nicholson scheme is used tosolve the time-dependent terms. The transformed vertical velocity is determined diagnostically by integratingthe anelastic continuity equation [ Eq. (A8)] from theroughness height to the model top. The vertical velocityw is then determined from Eq. (A9). The hydrostaticequation, Eq. (A3), is integrated from the model topto the surface to determine the deviation pressure p'.b. Boundary conditions Radiative heating and cooling prescribed at theearth-atmosphere interface is one of the physical forcings in the model. The thermodynamic energy equationand soil heat-flux equation are coupled with the surfaceheat-balance equation at the soil roughness height Zoto obtain O0 0 TsoilG- Fn + t~CpKh ~zz- X, O--~- pwLvE = O (3)where F, is the longwave radiative flux, Xs the soil thermal conductivity, pw the density of water, Lv the latentheat of vaporization, and E the evaporation rate inmeters per second. The longwave (terrestrial) radiativeflux F, and the atmospheric flux divergence Q thatappear in Eq. (A5) are computed as functions of thewater-vapor pathlength integrated through the atmosphere (Paegle and McLawhorn 1983). Equation (3)is a balance of the solar radiative flux, the longwaveradiative flux, the sensible heat flux, the soil heat flux,and the evaporative flux. Solar radiation G, which appears in the surface energy budget equation, is calculated fromSEPTEMBER I991 JEROME D. FAST AND MICHAEL D. MCCORCLE 2143G = 1353 W m-2 X ( 1 - A)[sin4~ sinb + cosb cos4~ sin(~rt/12)]r (4)where A is the albedo, 4~ is the latitude, b is the declination, t is time in hours (that varies in longitude),and v is the transmittance. The declination is a functionof Julian day. Clear-sky conditions are assumed in thesimulations described in this paper; therefore, r is setto 1.0. For most simulations in this study, albedo variesaccording to summertime datasets taken from Matthews (1985). In addition, albedo may be determinedas a function of soil moisture as described by Idso etal. ( 1975 ). This parametedzation is valid only for loamsoil and is given by the following relation: A = 0.31 - 0.34n/,ls ~/~s ~< 0.5 A = 0.14 ~/~s > 0.5 (5)Equation (5) is used in several sensitivity simulationsto examine its effect on boundary-layer circulations. Temperature continuity is assumed at the roughnessheight such that rair = rsoil at z = Zo. (6)A similar continuity relation exists for the moistureflux across the interface so that Wait = W~oil at z = Zo (7) Wair = pgq~ (8) Wsoit = awe (9)where Wair and Wsoil are the vertical moisture fluxes inthe atmosphere and soil, respectively. At the bottomof the soil layer, the temperature is held to its initialvalue. The boundary-layer model described by McCorcle(1988) assumed zero-gradient lateral boundary conditions for all of the prognostic variables. This boundarycondition sets the derivative of a prognostic variablenormal to a lateral boundary equal to zero. The currentmesoscale model makes this assumption for the simulations that neglect synoptic flow. Time-varying lateral boundary conditions based onobjectively analyzed observation fields are used in thebaroclinic numerical simulations in this study. Whenthis lateral boundary condition is employed, there maybe unwanted numerical instabilities near the lateralboundaries. Several methods have been proposed toremove this numerical noise, such as additional horizontal diffusion near the lateral boundaries; however,the present model used a simple low-pass filter (Pielke1984) on the three outermost nodes near the lateralboundaries. Time-varying boundary conditions are used at the model top for all of the prognostic variables in the atmospheric portion of the earth-atmosphere model. Forthe simulations that neglect synoptic forcing, geopotential height gradients are assumed to be zero at themodel top so that no horizontal wind is forced. Potential temperature and specific humidity are held constantin time. For the baroclinic simulations, the prognosticvariables are allowed to change in time. At the model'top, spline interpolation of observed fields is used toupdate the prognostic variables and pressure in time.The horizontal wind field at the model top is determined from the geostrophic relationship.c. Initial conditions The initial basic-state temperature T~ is specified byassuming a vertical lapse rate of 6.5-C km-~ with asea4evel temperature of 298-C. The Poisson equationis used to determine the basic-state pressure. The basicstate density fields are then determined from the equation of state. For the simulations with no imposed synoptic flow,barotropic initial conditions are used in the mesoscalemodel. The initial deviation pressure and wind fieldsare set to zero. The specific humidity is determined byemploying the Clausius-Clapeyron relationship andassuming a 75% relative humidity throughout the entiredomain. Deviation temperature is then diagnosed froma combination of the equation of state and the virtualtemperature relationship. Potential temperature is determined from the Poisson equation. Since there is nosynoptic flow in the barotropic simulations, the onlyforcing will come from the diurnally varying surfaceenergy budget. Variable surface characteristics willFIG. 1. Model topography, contour interval of 150 m.2144 MONTHLY WEATHER REVIEW VOLUME 11910281020101: 1012161020a20June 2110201016101610201008 1016b '1012I;"IG. 2. The sea-level pressure (mb) and surface temperature (-C) fields on(a) 1200 UTC 21 June 1989 and (b) 1200 UTC 22 June 1989.cause horizontally inhomogeneous thermal fields todevelop which, in turn, will induce circulations. Initialization techniques using barotropic initialconditions may be adequate when simulating idealizedatmospheric circulations; however, baroclinic initialconditions are needed to more accurately simulate realistic events. For the baroclinic simulations, observed surface andupper-air potential temperature and specific humidityare objectively analyzed to the three-dimensionalSEPTEMBER 1991 JEROME D. FAST AND MICHAEL D. MCCORCLE 2145model grid. At the model top, pressure is determinedby the hydrostatic relationship from the gridded analysis of the observed 300-mb height field. The interiorpressure is obtained by the integration of the hydrostaticequation (A3). The initial winds are geostrophic, except below 324 m, where the winds are forced to logarithmically approach zero at the roughness height. Inthe baroclinic simulations, the synoptic field imposedat the top and lateral boundaries will force the model,in addition to the diurnally varying surface energya16 1618 18June 21 112b10 16l:lo. 3. The surface specific humidity (g kg-~) on (a) 1200 UTC21 June 1989 and (b) 1200 UTC 22 June 1989.2146 MONTHLY WEATHER REVIEW VOLUME 119budget. The thermally induced circulations due to thevariable surface characteristics will be modified by thesynoptic flow. An ageostrophic wind field could have been used forthe initial conditions; however, a dynamic initializationtechnique would have been necessary to balance themass and momentum fields. The model uses a Newtonian nudging technique described by Anthes (1974)and Hoke and Anthes (1976), but a preforecast adjustment period greatly increases the computationala21 1bJune 22, 1FIG. 4. The 300-mb height field ( t0~ m) and selected wind barbs for(a) 1200 UTC 21 June 1989 and (b) 1200 UTC 22 June 1989.SEPTEMBER 1991JEROMED. FAST AND MICHAEL D. McCORCLE 12 UTC ~tl~ E 22, 1989 ~u2147 FIG. 5. Observed 24-h precipitation on 1200 UTC 22 June 1989. The open circles denotestations reporting a trace to 0.5 in and the filled circles denote stations reporting more than 0.5in.FIG. 6. Crop moisture index for 24 June 1989.2148 MONTHLY WEATHER REVIEW VOLUME II9TABLE 1. Summary of the numerical simulations.Case Synoptic Soil moisture Soil type Albedo NS 1 no none ST 1 aNS2 no SM 1 c ST 1NS3 no SM2d STINS4 no SM3 e ST 1NS5 no SM 1 ST 1NS6 no SM2 STINS7 no SM3 ST1NS8 no SM 1 ST2 sNS9 no SM2 ST2NSI0 no SM3 ST2SI yes none ST1S2 yes SM 1 ST 1S3 yes SM2 ST1S4 yes SM3 ST1S5 yes SM 1 ST 1S6 ues SM2 ST1S7 yes SM3 STIS8 yes SM 1 ST2S9 yes SM2 ST2S 10 yes SM3 ST2 a ST1 = loan in entire domain b AI = albedo from summertime datasets (Matthews 1985) c SM 1 = distribution shown in Fig. 7a d SM2 = distribution shown in Fig. 7b c SM3 = 0.284 in entire soil layer fA2 = a function of soil moisture (Idso et al. 1975) 8 ST2 = distribution shown in Fig. 8 h A3 = a function of soil color (Wilson and Henderson-Sellers1985) time necessary for a single simulation. This technique requires an additional term added to the prognostic equations [Eqs. (AI), (A2), and (A5)] that nudges the numerical results toward the objective analysis of^lb the observed data to bring the dynamic and thermalA1A I fields into balance as much as possible. During this^l adjustment process, the diurnal forcings are removed^2f and the synoptic forcing at the boundaries are heldA2 constant in time.^2A3h Sensitivity tests with the Newtonian nudging techA3 nique showed that a preforecast period of 12 h wasA3 necessary to balance the mass and momentum fields.^l As in Anthes et al. (1982), initialization with unbalA1 anced temperatures and winds produced no discernible^1 increase in noise when compared to simulations thatAI employed balance fields. Since the objective of this^2 study is to qualitatively simulate baroclinic circulations,A2A2 not forecast observed events, dynamic initialization wasA3 not used.A3 The baroclinic initial conditions of the mesoscaleA3 model are based on observations obtained from the Unidata SDM (Scientific Data Management) system (Sherretz and Fulker 1988). A procedure has been de veloped to create an objective analysis of the horizontal wind components, specific humidity, potential tem perature, and height fields from radiosonde data for arbitrary horizontal grids for every standard observa tion level. This procedure incorporated a single-pass Barnes (Barnes 1964) objective analysis technique withaDRYbDRY FIG. 7. Initial volumetric soil-moisture distributions representing the relatively wet and dry regions indicated by the crop moisture indexwhere (a) distribution SM 1 comprises of a gradual horizontal soil-moisture gradient and (b) distribution SM2 comprises of a sharp horizontalsoil-moisture gradient.SEPTEMBER 1991 JEROME D. FAST AND MICHAEL D. MCCORCLE 2149exponential weights. The objective analysis fields usedevery available radiosonde station in North Americawith an average grid station spacing of 300-400 km.The objectively analyzed variables are then interpolatedto the vertical levels of the mesoscale model. Surfaceand upper-air data are received continually by the SDMsystem from a satellite link, so that near real-time simulations of mesoscale circulations can be made. Thesedata are continually archived so that simulations ofpast events also can be made.d. Model domain The domain used in this study is the central UnitedStates as depicted in Fig. 1. The model domain employs25 nodes in both horizontal directions with a grid spacing of 104 km. A time step of 360 s is used. There are 27 nodes in the vertical in the atmosphericportion of the model, and 10 of those nodes are logarithmically spaced below 324 m. A grid spacing of 557m is employed between 324 m and the model top at9793 m. Most mesoscale models set the level of themodel top higher in the atmosphere; however, no convective effects are included in this model. The verticalgrid in the soil portion of the model consists of 15 soiltemperature computation levels, spaced equally 0.04m apart, extending from the roughness height z0 at0.04 to 0.52 m below the surface. The soil hydrologyconsists of a two-layer method to update soil-moisturecontent. The upper layer is 0.08 m deep and the lowerlayer extends 0.96 m below the surface. Because ternLOAM FIG. 8. Soil-type distribution ST2 for the central United Statesbased on a general soil-type map depicted in Foth and Schafer (1980). moximum vector = 2.83 m s-l~ab FIG. 9. Numerical results from dry soil, no-synoptic-flow simulation NS1 2 m above the surface predicted for 1800 LST 21 June.(a) Wind and temperature fields, contour interval of 2-C and (b)specific humidity field, contour interval of 1 g kg-~.perature forecasts depend on soil-moisture content tocalculate soil thermal conductivity and heat capacity,updated volumetric soil-moisture values are interpolated to match the soil-temperature forecast levels.2150 MONTHLY WEATHER REVIEW VOLUME I19maximum vector -- 0.73 m s-l~ maximum vector = 1.04 m s-l~b d FIG. 10. Predicted difference fields 2 m above the surface for 1800 LST 21 June. (a) Wind and temperature difference fields (simulationNS2 - NS1 ), contour interval of 0.5-C, (b) specific-humidity difference field (simulation NS2 - NS1 ), contour interval of 1 g kg-I, (c)as in (a), except for simulation NS3 - NS1. (d) as in (b), except for simulation NS3 - NS1, (e) as in (a), except for simulation NS9 NSI, (f) as in (b), except for simulation NS9 - NS1.3. Case description Model performance and sensitivity to soil-moisturedistributions were examined by using a typical summercold-front passage. The model is initialized with observed surface and upper-air data from 1200 UTC 21June 1989. The lateral and top boundary conditionsincorporate synoptic data every 12 h from 1200 UTCSEPTEMBER 1991 JEROME D. FAST AND MICHAEL D. MCCORCLE 2151emQxirnurn vector =01.17 m s-l~FIG. 10. (Continued)21 June to 1200 UTC 23 June 1989. Numerical resultswill focus on the first 24 h period; therefore, only thedetails of the synoptic situation during this period aredescribed here. Surface fronts, pressure, temperature, and winds for1200 UTC 21 and 22 June are shown in Figs. 2a,b. At1200 UTC 21 June a weak low pressure system insouthern South Dakota and northern Nebraska is located on a cold front that extended from northeasternNorth Dakota to southern Colorado. The front wasstationary from Colorado to central California. Thestrongest southerly surface winds were 8.8 m s-~ innorthern Texas, and the strongest northerly surfacewinds behind the front were 6.0 m s-~ in central SouthDakota. A relatively uniform warm air mass existedahead of the front with a surface temperature between18 - and 23 -C. The coldest air temperature of 5 -C waslocated well behind the front in Wyoming and Idaho. During the next 24 h, the front slowly advanced andweakened as it progressed southeastward across thecentral United States. At 0000 UTC 22 June daytimeheating produced temperatures in excess of 30-C fromthe desert southwest to southeastern Missouri aheadof the front (not shown). At 1200 UTC 22 June thefront extended from northern Minnesota throughsouthwest Missouri to western Texas, and the temperature gradient across the front was greatly reduced. Thefront continued to move southeastward, but virtuallydissipated by 1200 UTC 23 June (not shown). The evolution of the specific humidity fields at thesurface for this frontal passage is shown in Figs. 3a,b.At 1200 UTC 21 June a strong moisture gradient existed just behind the cold front from North Dakota tonorthern Texas. At 0000 UTC the moisture gradientin the northern plains increased due to advection (notshown). Moist air was advected from the Gulf of Mexico by the southerly winds between the surface and the700-rob level ahead of the front. Between 0000 and1200 UTC 22 June the 850-rob flow changed from thesouth to the southwest, cutting offthe moisture advection to the north-central United States. The 300-mb heights and wind fields, depicted in Figs.4a,b, reveal a nearly stationary trough over the westernUnited States. During the 24-h period, the troughmoved southeastward and deepened slightly. Thehighest wind speeds occurred over North Dakota andincreased from 40 m s-~ on 1200 UTC 21 June to 60m s-~ on 1200 UTC 22 June. Lack of a rapidly propagating upper-air system was responsible for the relatively slow movement of the surface front. This synoptic system produced some isolated thundershowers along the front in northeast Minnesota,eastern Nebraska and Kansas, and the panhandle ofTexas. The observed 24-h precipitation for 22 June1989 is plotted in Fig. 5. Cloudy conditions existedalong the central and northern portions of the frontand in the northern plains during most of the day.Clouds are important modulators of both solar andlongwave radiation; however, the present mesoscalemodel cannot simulate their potential effects on theboundary-layer circulations because the model doesnot utilize a parameterization for clouds. This particular case is chosen, not only for the frontal2152 MONTHLY WEATHER REVIEW VOLUME II9~>0a1000100~10 1200 LST .: : : : ,, :O~.0: . A [30 ~ [] O ~l [2]0 I[ [30 II []O [~ o o~ .... ! ~l:~290 300 310 1800 LST :::::~:~, a~5 a~ ~ []o ~ I] /[ []o [I [30 II []o290 300 310 0000 LST : : : : I :n~:/-:- P El) II [2I) /~ [3O /I [] O /I [30 // []O I/ rl 290 300 310 0600 LST ', ~ : : ', :O:EI: : O/ cg II [2D /X []O II [] O Xl []o I~ [3o II [3o ,,,!~/ ....290 300 310100010010b 1200 LST O[2 O[2 II O[3 O [3 O []5 10 15 20 1800 LST o[3 o [] o 2 o []5 10 15 20 0000 LST [~ o[3 o [] O [2 II o [2 I/ o[2 II o[25 10 15 20 0600 LST o[3 o [] o [] o [] o [] o []5 10 15 20 FIG. 11. Evolution of the simulated boundary layer in eastern Oklahoma from 1200 LST 21June to 0600 LST 22 June. (a) Potential temperature profiles (K), where O denote results fromdry-soil simulation NS1 and [] denote results from moist-soil simulation NS3, (b) as in (a), butfor specific humidity (g kg-~ ).passage, but also for the soil-moisture conditions present. During most of June 1989, the southeastern UnitedStates and the Great Lakes region experienced abundant precipitation. As shown in Fig. 6, these regionswere reporting relatively wet soil conditions. Most ofthe other areas in the central United States were reporting abnormally dry conditions, with excessive dryness reported in Nebraska and southern Texas. Forthis reason, this case seemed appropriate to examinethe potential effects of a soil-moisture distribution onthe forecast varifibles of a baroclinic circulation.4. Numerical results This section summarizes some of the simulationsthat are performed to isolate the mechanisms by whichNCMCs could effect larger-scale circulations. Thisstudy consists of several control and sensitivity experiments to demonstrate that inhomogeneities in soilmoisture, and soil type can significantly modify typicalmesoscale circulations. A brief description of these experiments is summarized in Table 1. One set of control experiments uses the initial conditions described in section 3 for 1200 UTC 21 June1989. The control simulations of the frontal passageare made with dry, bare soil under clear-sky conditions.The sensitivity experiments are similar to the controlsimulations, except that soil moisture, soil type, andalbedo characteristics are modified. Another set of control experiments simulate the circulations that develop over the domain without anysynoptic forcing. These simulations are made with solarradiation attributes from 21 June. Numerical studies,such as Ookouchi et al. (1984), have shown thatSEPTEMBER 1991 JEROME D. FAST AND MICHAEL D. MCCORCLE 2153NCMCs may be as significant as the sea-breeze phenomena in the absence of synoptic forcing with a horizontal grid spacing of 10 km. The purpose of this setof experiments is to demonstrate the potential magnitude of NCMCs without the complicating effects ofsynoptic flow patterns using a horizontal grid spacingof 104 km. By comparing the sensitivity and control experiments, the effect of the simulated NCMCs can be evaluated in detail. This is accomplished by subtracting theresults from the control experiments from the resultsfor the sensitivity experiments. Most of the figures inthis section depict these difference fields to demonstratethe structure and extent of the secondary circulationscaused by surface inhomogeneities. As indicated in Table 1, three initial soil-moisturedistributions are used. Distributions SM 1 and SM2 aredepicted in Figs. 7a,b. Soil-moisture distribution SM 1is representative of the relatively wet and dry regionsfrom the 24 June crop moisture index (CMI) (Fig. 6).Wet CMI values are arbitrarily assigned a value of ~= 0.35 (about 80% of the saturation value for loamsoil) in SM 1. Excessively dry CMI values are assigneda value of~ = 0.05. This distribution then incorporatesa gradual horizontal gradient in soil-moisture contentbetween the wet and dry regions. The CMI is availableduring the growing season every two weeks. The 24June field is used because it remained relatively constant during June and early July. In SM2, two uniformhorizontal initial soil moisture regions of ~ = 0.284(about 65% the saturation value for loam soil) and n= 0 are assumed. The relatively wet and dry regionsare located in the same areas as in Fig. 7a; however, asharp horizontal gradient in soil moisture is used. Soilmoisture distribution SM3 sets n -- 0.284 in the soillayer in the entire domain. Estimating initial soil-moisture distributions formesoscale models can lead to large uncertainties because of large spatial irregularities in the domain ofinterest and the transience of contrast lines. Only theoretical or plausible soil-moisture distributions can beincorporated into the present mesoscale model. Moreresearch is needed to develop routine procedures thatassimilate quantitatively the effects of soil moisture intoshort-range forecasts, such as derived soil-moisturevalues from satellite data (Wetzel and Chang 1988).Observations of daily variations of soil moisturethroughout the United States are needed to verify results from these forecasts. It is possible to use a subgrid-scale weighting technique similar to Avissar and Pielke (1989) to determinesoil type; however, this would require an accurate dataset of soil type for the entire United States [griddedsoil texture and albedo data are available from NCARdata archives as discussed in Wilson and HendersonSellers (1985)]. Most of the simulations in this paperuse loam soil throughout the domain (distributionST1 ) since the principle objective is to examine theeffect of soil moisture. In several simulations, soil typeis allowed to vary horizontally as shown in Fig. 8, butit is homogeneous within a grid cell. This distributionis based on a general soil-type map depicted in Fothand Schafer (1980). When this soil-type distributionis used, albedo is allowed to vary according to soil coloras described by Wilson and Henderson-Sellers ( 1985 ).a. No-synoptic-flow experiments Each of the simulations listed in Table 1 for no synoptic flow are integrated for a period of 48 h, althoughall of the figures present results of the 12-h forecast.The resulting circulations for the second day were verysimilar to those from the first 24-h period because soilmoisture did not sufficiently dry out during the firstday. Neumann lateral boundary conditions were usedas described in section 2.1 ) DRY-SOIL SIMULATIONS Figures 9a,b depict the wind, temperature, and specific humidity fields for the 12-h forecast valid for 1800LST 21 June, 2 m above the surface for control experiment NS 1. At this time, upslope wind speeds inexcess of 2.0 m s-~ were predicted in the west-centralGreat Plains near the surface. The model predictedupslope flow for most of the day and a maximum upslope wind speed of 3.7 m s-~ occurred at 1800 LST,43 to 119 m above the surface in northwest Texas. Thewind direction veered during the evening due to Coriolis forcing to produce a nocturnal southerly jet of5.1 m s- ~, 119 m above the surface in northwest Texasbetween 2100 LST and midnight (not shown). By 0600LST 22 June, a downslope westerly wind was evident.The specific-humidity distribution shown in Fig. 9bdid not change considerably during the simulation because the winds were relatively light over most of theperiod. No significant moisture advection occurredfrom the Gulf of Mexico. Holton (1967) demonstrated that the diurnally oscillating slope flow was an important mechanism ofthe Great Plains low-level jet. Even the relatively largehorizontal scales and gentle slopes used in the presentstudy can produce significant slope flows (Fast andMcCorcle 1990). The slope flow predicted by the modelresembled the type of flow that can occur over smallerterrain features with much steeper slopes.2) EFFECT OF HETEROGENEOUS SOIL MOISTURE AND TYPE The addition of soil moisture can produce seabreeze-type circulations when simulated by numericalmodels using horizontal scales between 5 and 10 kmas shown by Avissar and Pielke (1989), Mahfouf et al.2154 MONTHLY WEATHER REVIEW VOLUME 119maximum vector :9.51 m s-l~a c ' max[mum vector = 7.04 m s-l~ d F~G. 12. Numerical results from dry soil, frontal-passage simulation SI 2 m above the surface. (a) Wind and temperature fields for 1800LST 21 June, contour interval of 2-C, (b) as in (a), but for 0600 LST 22 June, (c) specific-humidity fields for 1800 LST 21 June, contourinterval 1 g kg-l, (d) as in (c), but for 0600 LST 22 June and (e) MFC field for 1800 LST 21 June where positive values indicate moistureconvergence and negative values indicate moisture divergence, contour interval 25 x 109 s-l, (f) as in (e), but for 0600 LST 22 June.(1987), and Ookouchi et al. (1984). Evaporation ofsoil moisture also effected circulations with a horizontalscale of 140 km in the boundary-layer model ofMcCorcle ( 1988 ) and in global climate models (Dickinson et al. 1986; Meehl and Washington 1988; Wilsonet al. 1987). The differences in the forecast variablesdue to the various soil-moisture and soil-type distributions in this study are shown in Figs. 10a-f.SEPTEMBER 1991 JEROME D. FAST AND MICHAEL D. MCCORCLE 2155f FIG. 12. (Continued) The most significant changes in the forecast variablesoccurred for the sharp moisture gradient of distributionSM2 as seen in Figs. 10c,d. Surface temperatures decreased by as much as 3.0-C in northern Arkansas dueto evaporative cooling. Evaporation from the soil layerin the moist regions increased the specific humidity byas much as 5.6 g kg-~ in western Arkansas. The reduction in temperature had the effect of producing amesohigh over the regions of moist soil. A weak seabreeze-like circulation developed near the boundaryof the warmer, dry-soil areas and the cooler, moist-soilareas. Wind speeds at 1800 LST differed from the drysoil simulation by as much as 1.5 m s-~ in northwestTexas 119 m above the terrain. The response of themodel to soil distribution SM 1 as seen in Figs. 10a,bwas similar to SM2, except that the resulting NCMCwas weaker due to the smaller gradients in soil moisture. The maximum difference in wind speed betweenthe dry-soil simulation was 1.0 m s-l in northeast Arkansas from 43 to 119 m above the terrain. Evaporation rates are highly dependent upon soiltype. Because the water holding and retention properties of soils vary by more than 300% with soil type(Taylor and Ashoroft 1972), the soil properties can besignificant in surface energy exchange. Evaporationrates in the present soil-hydrology model were highlydependent on soil type as shown in Fast and McCorcle(1990). This was due, in part, to the initialization ofsoil-moisture content. For instance, relatively largeevaporation rates were produced over loam soil when~ = 0.284 in soil-moisture distribution SM2. Evaporation rates were much smaller over clay soil with n= 0.284 since this soil-moisture content is just abovethe wilting point (V = 0.208 ). Clay soils have very finepore sizes and consequently more energy was requiredto evaporate the water from this soil. The NCMC produced in simulation NS9 using soilmoisture distribution SM2, soil-type distribution ST2,and albedo A2 also demonstrated that evaporation canproceed at different rates for different soil types. Asdepicted in Figs. 10e,f, evaporation proceeded readilyover north Texas, Oklahoma, and Kansas, while lessevaporation occurred over the nonloam regions. Thehorizontal extent of the NCMC was much less thansimulation NS3 with uniform loam soil, although thecirculation was just as intense. Soil-moisture distribution SM3 profoundly alteredthe 'flow field (not shown) because it resulted in temperature reductions of 1.7- to 3.1 -C everywhere in thedomain. This damped the magnitude of the slope flowby as much as 3.8 m s-l. The reduction in the upslopecomponent is qualitatively similar to the results of theterrain simulations reported in Ookouchi et al. (1984). The simulations that represent albedo by Eq. (5)according to soil moisture (Idso et al. 1975) in experiments NS5-NS7 did alter the albedo somewhat; however, the horizontal potential-temperature field determined indirectly by Eq. (4) did not differ significantlyfrom experiments NS2-NS4. The forecasted variablesproduced difference fields similar to those shown inFigs. 10a-f, which indicated that soil-moisture differences were more important than albedo differences, The time evolution of the potential temperatureprofile for experiments NSI and NS3 located in eastern2156 MONTHLY WEATHER REVIEW VOLUME ll9Oklahoma is shown in Fig. 11 a. At midday, the evaporation of soil moisture reduced the temperature of themixed layer by 2-C. This reduction in temperaturestabilized the boundary layer somewhat and diminished the vertical mixing to lower the boundary-layerheight by 500 m at 1200 LST. During the evening, astable layer 20 m in depth developed due to radiationalcooling in the dry-soil simulation. The addition of soilmoisture reduced the radiational cooling at night. Thisresulted in warmer temperatures near the surface andamaximum vector :1.66 m s-1 ~maximum vector0.99 m s-1 ~b d lqG. 13. Predicted difference fields (simulation S2 - S1 ) 2 m above the surface. (a) Wind and temperature difference fields for 1800LST 21 June, contour interval of0.5-C, (b) as in (a), but for 0600 LST 22 June, (c) specific-humidity difference field for 1800 LST 21June, contour interval of 1 g kg-~, (d) as in (c), but for 0600 LST 22 June.SEPTEMBER 1991 JEROME D. FAST AND MICHAEL D. MCCORCLE 2157reduced the lapse rate in the lowest 100 m when compared to the dry-soil simulation. The time evolutionof the corresponding specific humidity profiles for thesame location in Fig. 1 lb show the increase in moistureevaporated from the soil. At 1200 LST, the dry-soilsimulation was slightly moister 880 m above the surfacebecause of the greater strength of the daytime boundarylayer that mixed moisture upward. Since the boundarylayer depth was suppressed somewhat in the moist-soilsimulations, moisture accumulated near the surfaceinitially, and was not transported above the 324-mlevel. These simulations with no synoptic flow indicatethe general structure and magnitude of the NCMCsbmaximum vector -maximum vector : ) ./2.58 m s-1- __ ~, ~ i! .F1.24 m s-1 ~d14. As in Fig. 13, except for simulation S3 - S1.2158 MONTHLY WEATHER REVIEW VOLUME 119that could develop with the soil-moisture and soil-typedistributions used in this study. Now the effect of thesecirculations on synoptic flows can be evaluated.b. Synoptic-flow experiments Each of the simulations listed in Table I incorporating synoptic flow were initialized with observed datataken from 1200 UTC 21 June 1989 and were integrated for a period of 48 h. Most of the figures depictresults from the first 24-h period when the model wasmost sensitive to surface inhomogeneities. Preliminary simulations of this frontal passage whereperformed with zero-gradient, time-varying, and radiation lateral boundary conditions. Results indicateda ~'O'maximum vector =2.08 m s-1 ~ o)Cb i\~maximum vector =0.45 m s-14dMG. 15. As in Fig. 13, except for simulation S9 - S1.SEPTEMBER 1991 JEROME D. FAST AND MICHAEL D. M-CORCLE 2159that time-varying conditions quickly effected the resultsin the model interior and contaminated all of the features of the observed front. Both time-varying and radiation lateral boundary conditions, as described insection 2, retained the observed frontal features duringthis case; however, the simulated position of the frontwas slightly superior when time-varying conditionswere used. Time-varying lateral boundary conditionsare used for all the prognostic variables, except for potential temperature, which used zero-gradient boundaryconditions. 1 ) DRY-SOIL SIMULATIONS The results for the dry-soil simulation S 1 indicatedthat the numerical model was able to qualitativelysimulate the thermal and dynamic fields associated withthis particular front. Figures 12a-f depict the wind, temperature, specifichumidity, and moisture-convergence fields for the 12and 24-h forecasts 2 m above the ground. The front insimulation S 1 has moved to northern Minnesota, central Iowa, and on into central Oklahoma and northernTexas by 1800 LST 21 June as seen in Fig. 12a. Thesimulated front advanced 100-200 km farther to thesoutheast than the observed front on 0000 UTC 22June (not shown). A warm pocket of air in excess of30-C stretched from southern Texas to southern Kansas ahead of the front. The coldest air was located wellbehind the front in Wyoming. A sharp moisture gradient was evident near the frontal boundary in the central United States in Fig. 12c. The large gradients atthe southeast and northern boundaries resulted fromthe lateral boundary conditions employed by the mesoscale model. At this time, the model predicted significantly lower humidities near the boundaries thanthe observed values. Moisture-flux convergence (MFC) is a useful diagnostic quantity because it can be used to identify thelocations of potential thunderstorm development(Waldstreicher 1989). Here, MFC is defined by -qV. V- V.Vq = -V.(qV), (10)where the first term is mass divergence and the secondterm is moisture advection. A large positive value ofMFC does not guarantee thunderstorm developmentbecause strong capping inversions may be present. Asdescribed in Waldsteicher (1989), convection oftendevelops downwind of a MFC maxima, where MFCincreases rapidly in time, and where the gradient ofMFC is increasing. Figure 12e depicts convergence of moisture alongthe southern frontal boundary, with a local maximumin central Texas. There is a divergence of moisturealong the northern portions of the front in Minnesota.Most of the convergence or divergence of moisture inthis simulation is due to the first term in Eq. (10). During the following 12 h the simulated front weakened considerably and moved slightly southeastwardas seen in Fig. 12b. The position of the simulated frontagrees quite well with the observed front on 1200 UTCb FIG. 16. Predicted MFC difference fields (simulation S3 - S1 ) 2m above the ground, contour interval 20 x 109 s-I for (a) 1800 LST21 June and (b) 0600 LST 22 June. Positive values indicate greaterconvergence in simulation S3 and negative values indicate greaterdivergence in simulation S3.2160 MONTHLY WEATHER REVIEW VOLUME II922 June as shown in Fig. 2b. The moisture gradientremained relatively strong in the northern portions ofthe front, with the driest air located just behind thefront in the western plains (Fig. 12d). In Fig. 12f, theconvergence of moisture has weakened significantlyalong the southern portions of the front.2) EFFECT OF HETEROGENEOUS SOIL MOISTURE AND TYPE The addition of soil moisture in the sensitivity simulation with a gradual gradient in soil moisture (SM 1 )cooled the boundary layer over northwest ArkansasdFIG. 17. Predicted specific-humidity difference fields (simulation S3 - S1 ) 880 m above the ground for (a) 1200 LST 21 June (b) 1800 LST 21 June, (c) 0000 LST 22 June and (d) 0600 LST 22 June, contour interval 1 kg-~.SEPTEMBER 1991 JEROME D. FAST AND MICHAEL D. M-CORCLE 2161and northwest Mississippi by 4.5-C (Fig. 13a). Thewind field in the moist simulation differed from thedry-soft simulation by as much as 1.7 m s-~ at 12 hand 1.0 m s-~ at 24 h. As with the simulations havingno synoptic flow, the effect of soil moisture was to produce a weak mesohigh over the moist regions. Specifichumidity increased by as much as 5 g kg-~ in Oklahoma and 6 g kg-~ in western Tennessee (Fig. 13c)after 12 h. During the evening most of the additionalmoisture evaporated during the day was advected tothe front where it converged over Oklahoma, as seenin Fig. 13d. The results of the soil-moisture distribution SM2in Figs. 14a-d are similar to those in Figs. 13a-d, exceptthat the simulated NCMC is significantly stronger. Thesurface temperature was reduced by as much as 5.5 -Cin southern Oklahoma. The effects of the NCMC inthe boundary layer extend as far as 200-300 km northof the soil-moisture gradient into northern Missouri.A larger mesohigh produced wind-speed differencesnear the surface of 2.6 m s-~ at 12 h in Fig. 14a and1.2 m s-~ after 24 h in Fig. 14b. The position of thefront was not altered by the addition of soil moisture,but the wind-speed modifications resulted in a weakened front near the surface. In the sensitivity simulation, the largest increase in specific humidity was 8 gkg-1 that occurred at 1200 LST 21 June over the moistsoil region southwestern Missouri. Moisture evaporatedover the wet region also was advected northward toproduce significantly higher humidities far from themoisture-transition region. Evaporated moisture converged to the front much sooner than in simulation S2as seen in Figs. 14c-d because of the greater moistureavailability in soil-moisture distribution SM2. The effects of horizontally varying soil-type and soilmoisture distribution of simulation S 12 are shown inFigs. 15a-d. As with the corresponding simulationshaving no synoptic flow (NS 12), the structure of theNCMC was significantly different from those cases thatused uniform loam soft. Even though the area of intenseevaporation was smaller, significant specific humiditydifferences developed after 12 h as seen in Figs. 15cd. Once again the evaporated moisture converged intoOklahoma ahead of the front, but in slightly less quantities than for the other soil-moisture distributions. The NCMCs resulting from the soil-moisture gradients in Figs. 13-15 are significantly stronger to thoseproduced in the absence of synoptic forcing (Fig. 10).This was due, in part, because of the specification ofthe initial conditions in the two sets of experiments.The synoptic-flow simulations incorporated a more realistic initial temperature distribution. During the afternoon periods of the model integration the predictedsurface temperatures in the southern plains were asmuch as 7-C warmer than the no-synoptic-flow simulations; therefore, evaporation occurred at a muchhigher rate in the synoptic-flow experiments. Thiseventually forced stronger NCMCs to develop. Thehorizontal distribution of the soil moisture in the sensitivity simulations also had an effect on the strengthof the NCMCs. Drier, warmer air was advected intothe southern plains. This situation leads to an intensification of the horizontal pressure gradients that intensifies the nonclassical circulation. These latter mechanics were also illustrated in the synoptic flow simulation in Avissar and Pielke (1989). The modification of the specific humidity and windfields by the presence of soil moisture also altered theMFC fields in the central United States for case SM2(sharp moisture gradient) as seen in Figs. 16a,b. Theinteraction of the synoptic circulation and the NCMCenhanced the moisture convergence just behind thefront in Kansas and western Iowa, and ahead of thefront in Iowa, Wisctnsin, and northern Illinois after12 h. The enhanced divergence in the southern statesdemonstrates that moisture evaporated in those regionsadvected north towards the front. The enhanced convergence is the same order of magnitude as the moistureconvergence in the dry-soil simulation S I. An important difference from the dry-soil simulations is thata significantly greater portion of the MFC predicted byEq. (10) is now due to the moisture-advection term.Observations have shown that moisture advection cansignificantly contribute to the development and subsequent intensification of storms (Bothwell 1986). The time evolution of the specific humidity 880 mabove the surface is shown in Figs. 17a-d. After 6 h,the specific humidity in the moist-soft simulation isless than the dry-soil simulation at this level becausethe cooler boundary layer reduced the mixing as inFig. 1 l a. By 12 h, daytime heating has allowed theboundary layer to grow above 880 m so that the additional moisture is transported from the surface. Themaximum value is 3.5 g kg-~, about a 30% increaseat this level, in northeastern Oklahoma and southeastern Kansas. During the next 12 h, this additionalmoisture was advected northeastward ahead of thefront into northern Wisconsin. While the most profound modifications in the boundary-layer structure occurred near the surface, the NCMCs caused by the soil-moisture distribution were noticeable up to 2500 m above the terrain. The vertical cross-section plots in Figs. 18a-d demonstrate a specific humidity increase up to I km above the terrain after 12 h. During the evening, vertical mixing was reduced, but some of the additional moisture appears to have reached 2500 m above the surface. This was probably due to the passage of the front and synoptic-scale ver tical motions ahead of the front that transport this moisture. Near the surface, the evaporated moisture was advected 300 km north of the soil-moisture-con trast zone as seen in Fig. 18c. The vertical cross-section plots in Figs. 18e, fclearly depict the reductions in temperature in the boundary2162 5.04.0 3.0km 2.0aMONTHLY WEATHER REVIEWVOLUME 1 191.05.0 I ~ I i \ ~, ,' '..~,, , 0\ I -,~-'~% \ i1~ ~x ~ zxx . ~ ~ "~,...-K x --~L__ ~ :~:~ :~ ..... .~-z<-,',l~.-, / ~.~ ~ , ,, ) ~ ~ [ ~z-.-2',,~, ~ ~-~ll ~ ~ ~1 r~~::~ ~104.25 ~9.75 05.25 ~0.75 8~.254.0 3.0km 2.01.00.0 0.0108.75 27.5Iongifude ::::::::::::::::::::::::::: o,: . ____ _ , . .:-:> :,:,>,: 33.0 36.5 41.0 45.5 50.0latitude5.0 5.04.0 3.0km 2.01.04.0 i , i i /! /'\\ ~ ~ 0 /~ pi /~ '~-- / , ',/o,, ~-) ',: ~ / ',,;/:x'~'.: )~- \ ~..-----~ :-"11 ~XXY, ,' 104.25 0~.75 ~5.25 00.75 8~.25 3.0km 2.01.00.0 ' O.C108.75 27.5Iongifudeb d: L 2 33.0 0. ' '------~o ' ~-. x~ ~ ~ / \ xx /~ x iIw~.~/~ \ N i x t I ~ ,I ~ I I \ I I 11 xx I Ill x I ill X ,,,,,,~0~ / Ill I .~~/// L.36.5 41.0 45.5 50.1latitude 1~0. 18. Vertical cross-section difference fields (simulation S3 - SI ). (a) Specific-humidity difference field on 1800 LST 21 June corresponding to line AA' in Fig. 17b, contour interval of 0.5 g kg-~, (b) as in (a), but for 0600 LST 22 June, (c) specific humidity differencefield on 1800 LST 21 June corresponding to line BB' in Fig. 17b, contour interval of 0.5 g kg-l, (d) as in (c), but for 0600 I_ST 22 June,(e) temperature difference field 1800 LST 21 June corresponding to line AA' in Fig. 17b, contour interval of 0.5-C, (f) as in (e), but for0600 LST 22 June.layer. The changes in the temperature profile are notconstant even though the initial soil-moisture regionfor this case was uniform (SM2). As in the no-synoptic-flow simulations, albedo calculated by Eq. (5) in experiments S5-S7 was somewhataltered by soil moisture; however, the horizontal potential temperature field did not differ significantlyfrom experiments S2-S4. The forecasted variablesproduced difference fields similar to those shown inFigs. 10a-fiSEPTEMBER 1991 JEROME D. FAST AND MICHAEL D. MCCORCLE 21635.04.0 3.0km 2.0 o O --. ..........,.0~~iyT,, _ .... ~ ~ ~ ~ x /-~-~ ~ ~x ~ ~ ~ ~ xx x ~ ~ \ x \ )~:x / ',; .,__,., ...'~ ,-108.75 104.25 99.75 95.25 90.75 86.25elongitude5.04.0 3.0km 2.0f1.00.0108.7: i i i i104.25 99.75 95.25 90.75 86.25 IongifudeFIG 18. (Continued) The time evolution of the potential temperature andspecific-humidity profiles for simulations SI and S3are shown in Figs. 19a,b for a point in eastern Oklahoma. The effect of moisture on the temperature andspecific-humidity profiles over a moist surface weresimilar to the corresponding no-synoptic-flow simulations in Figs. 1 la,b, except that the effect was larger.The temperature in the daytime was reduced by asmuch as 4-C. The boundary-layer heights in the sensitivity simulation were as much as 500 m lower thanthe dry-soil simulation because the reduced temperature at the surface suppressed vertical mixing near thesurface. The time evolution of the potential temperature andthe specific humidity over northern Illinois, a dry region, are shown in Figs. 20a,b. Moisture was not adveered to that area until after 12 h. The most significantincrease in specific humidity occurred after 18 h. Thepotential temperature was not reduced significantlybecause daytime heating before the advection of moisture kept the mixed layer relatively warm.5. Conclusions Atmospheric processes are inherently connected toenergy exchanges at the ocean and earth surface. Theobservation and numerical prediction of NCMCs hasreceived growing attention in the research literaturebecause they may be as important as other more thoroughly examined mesoscale phenomena, such as seaand land breezes, mountain and valley winds, and urban circulations. The presence of soil moisture or vegetation is expected to modify the surface thermal fluxeswhen compared to a bare-soil surface under the sameenvironmental conditions. Two- and three-dimensional numerical studies have indicated that horizontaldiscontinuities in soil moisture or vegetation could induce significant discontinuities in surface thermalforcing and, consequently, mesoscale circulations.Most numerical studies have simulated the resultingmesoscale circulations, that are similar to sea breezes,with horizontal grid spacing of approximately 5-15 kmwith no imposed synoptic flow. Such circulations mayplay an important role in patterns related to local meteorology and climatology, cumulus convection, andair quality. A major task of this research has been to expand thecoupled earth-atmosphere model described by McCorcle (1988) in order to include dynamics above theboundary layer, baroclinic initial conditions, and various boundary conditions. These changes were necessary to examine the effect of surface inhomogeneitieson the thermal and momentum properties of barocliniccirculations. The mesoscale model is governed by ananelastic, hydrostatic system of equations that aretransformed to a nonorthogonal grid system. For thebaroclinic simulations in this study, the lateral boundary conditions varied in time and were based on theobjective analysis of observed data. The prognosticvariables at the model top also varied in time and weredetermined from an objective analysis of observed data,except for the horizontal wind components that wereset equal to their geostrophic value. Observations from 1200 UTC 21 June 1989 of afrontal passage were used to initialize the three-dimen2164 MONTHLY WEATHER REVIEW VOLUME II9a1000100 1200 LST ~th: [3 0 o o [] o m o [] 0! ,::', ~!~, I\::~290 300 310 320 1800 LST []0 II [] 0~0:300 310 320 0o00 LST [20 t/ // //2~0 300 310 0600 LST // // []c [302~0 ~00:310 3201000100.10.b 1200 LST o [] 0 [] c [] I I o 121 I 0 ~0 []It :~"J .... 1. 5 10 15 20 251800 LST I\ o [] 0 o o 12 o [] 0 [] 10 lfi 20 25 0000 LST 0 [] 0 [] o o [] I I o [],5 10 15 20 25 0600 LST O[] O[] 0 [] c m o [] o5 10 15 20 25 FIG. 19. Evolution of the simulated boundary layer in eastern Oklahoma from 1200 LST 21June to 0600 LST 22 June. (a) Potential temperature profiles (K), where O denote results fromdry-soil simulation SI and [] denote results from moist-soil simulation S3, (b) as in (a), but forspecific humidity (g kg-~).sional model. This particular case was chosen, not onlyfor the frontal passage, but also for the horizontal distribution of abnormally dry and wet soil-moisture conditions present. The sharp horizontal variations in soilmoisture indicated that surface inhomogeneities maysignificantly effect the thermal, moisture, and momentum fields associated with this front. Two sets of soil-moisture numerical experimentswere executed to determine the magnitude and structure of the simulated NCMCs. One set of experimentsconsisted of several soil-moisture and soil-type distributions with no imposed synoptic flow. The secondset of experiments used the same surface characteristics,except that baroclinic initial conditions were used. Numerical results from the no-synoptic-flow experiments showed that soil-moisture and soil-type distributions could significantly effect the boundary layereven for relatively large horizontal scales. Evaporationfrom the soil increased the specific humidity by as muchas 6.1 g kg-~ and cooled the surface by as much as3.0-C. The NCMC resembled a mesohigh wind fieldwith a magnitude of 1.0-2.0 m s-~. This altered thewind direction and speed of the slope flows over theterrain in the central United States. The effects ofevaporation on the thermal and moisture fields wereobserved up to 1 km above the terrain. The evaporation of soil moisture also effected theboundary layer structure embedded in the barocliniccirculation. Evaporation from the soil increased thespecific humidity by as much as 10 g kg-~ and loweredthe surface temperature by as much as 6-C. As in theno-synoptic-flow experiments, a mesohigh wind fieldwas produced by the altered thermal field with windspeeds between 1.5 and 3.0 m s-~ near the surface.Some studies have indicated that significant synopticflow patterns could mask or reduce the potential effectsof surface inhomogeneities. In this study, soil-moistureand soil-type distributions in the synoptic-flow experSEPTEMBER 1991 JEROME D. FAST AND MICHAEL D. McCORCLE 21650a1000100I0 1200 LST .... ~/ / [] 7 \ ~I> [l~ rl [~ o / []1 ::::l::~:l:::,,i2@0 300 310 320 1800 LST [33 ....290 200 310 220 0000 LST // [30 // [213 [1~ ..ll. ,,290 300 310 320 0600 LST //290 300 310 3200b1000.1001200 LST (3 C~ O 10 15 20 25 1800 LST O[3 0[3 O[3 O[35 10 15 20 25000o LST O [3 O [3 O [] O [3 O [3 0600 LST O[3 0[3 O[3 0[3 0[3 0[35 10 15 20 25FIG. 20. As in Fig. 19, but for boundary layer in western Illinois.iments were found to have an even greater effect thanin the no-synoptic-flow experiments; however, someof this effect was due to the different initial conditionsused in the two sets of experiments. While the mostsignificant effects occurred near the surface, evaporatedsoil moisture was advected horizontally far from itssource and transported vertically into the free atmosphere by nonlinear synoptic-scale circulations. Moisture ilux convergence was used in this study todemonstrate the potential impact of horizontally heterogeneous soil moisturc on the spatial distribution andintensity of precipitation. This could be examined inmore detail by mesoscale models that include grid-scalecondensation, precipitation, latent heat release, andcumulus convection parameterizations; nevertheless,the possible effects of soil-moisture distributions onmesoscale circulations can still be addressed using thepresent mesoscale model. It is important to note thatthe present mesoscale model does not contain a cumulus parameterization that might act as a feedbackmechanism for the evaporated soil moisture; therefore,the magnitude of the resulting NCMCs may be overpredicted. This research also demonstrates the need for routine,accurate observations of soil moisture content and distribution in the United States. These data are necessarybecause the parameterization of horizontal heterogeneous land characteristics in operational models maysignificantly influence short-range forecasts. Globalclimate models have shown considerable sensitivity todrastic changes in the formulation of soil evaporationand evapotranspiration; therefore, local climatologicalchanges may not be predicted correctly. More routinesoil-moisture observations on the horizontal scale areneeded to verify two- and three-dimensional simulations of NCMCs. The parameterization of the effectsof surface inhomogeneities in studies reported in theliterature vary in complexity, and it is not clear howdetailed a model needs to be to adequately simulatethe energy and moisture exchanges at the soil-atmosphere interface. The present mesoscale model does not incorporate2166 MONTHLY WEATHER REVIEW VOLUME I19a vegetation-layer parameterization. Evaporation fromthe bare-soil surface could be greatly reduced, depending on the type of vegetation, due to shading. In addition, evapotranspiration from a vegetation layer canincrease the moisture and decrease the temperaturenear the surface. Horizontally inhomogeneous distributions in plant types would greatly increase the complexity of energy and moisture exchanges between thesurface and the atmosphere. It is anticipated that the present mesoscale modelwill be used in the future to simulate mesoscale flowpatterns with observed atmospheric and soil-layer data.This would require executing the model with a muchsmaller spatial resolution so that data from experimentssuch as HAPEX-MOBILHY could be employed. Possible forecast errors due to initial conditions, boundaryconditions, grid resolution, and surface parameterizations could be evaluated in more detail. Acknowledgments. This research was supported bythe Iowa State University Agricultural and Home Economics Experiment Station under Project 2804. Thedata and some of the software for this study were madeavailable through the Unidata program, which is sponsored by the National Science Foundation. APPENDIX Governing Equations The governing equations described by McCorcle(1988) have been transformed into the nonorthogonalgrid system for the atmospheric portion of the modeland areDu Ou 1 Ou 10u Ou otD~- Ot + u-- + v + o~-- a cos qO OX ~ ~'~ O~ a = ~, -f acos- OX u+~ v + ~ ~a cos- 0X I Oz6(~)+V.(KaVu) (AI) + g a cos~ O~ mDv Ov 1 Ov 1 Ov Ova ~ ~oW = oW + u a cos, ox + v; ~ + ~ oe ~ [s~] ~ K~; ;-fu-p~O~ + g~ o~k s ] + V.(KaVv) (A2) ~ 7 = -~'g (~3)+ ~a/ j\s-zo/ Km3 (0l)'* + V. (KaVe)10X(A4)(AS)(A6)DX Ox I OX OX otD-~= O--J' + u- + v + -- a cos4~ OX ~ ~-~o~ O~ ~ = K~O~ .]a , v.(g~vx) (AT) L a cos ~ 0x + a ~ J + o~ ( u Ozow ;~voz-~ s-z~ acos~ o,. +-~/=- (aS) u Ozo a -- s a cos4 0X ~ OzG ~ - s s a~ s- ga p' - ~RT' ~' = <A~o~ RT~ r'=o~~*/%- rs +0.61qTs (All) XPo l P= Ps + P'. (AI2)The pfim~ variables are u, v, w, w, p, p', q, 0, e, X,T', and a'. Here, u and v are the hofizon~ velocitycomponents, ~ the transfo~ed veaical velocity component, w the veaicfl velociW component, p the to~SEPTEMBER 1991 JEROME D. FAST AND MICHAEL D. MCCORCLE 2167pressu,re, p' the deviation pressure, q the specific humidity, 0 the potential temperature, e the turbulentkinetic energy, x the particulate concentration, p' thedeviation density, and T' the deviation temperaturethat is adjusted for moisture. The constants used inthese equations include g the gravitational force, landf the Coriolis parameters, Q the diabatic heating, Rthe gas constant for dry air, Po the reference pressure,Cp the specific heat capacity for dry air, Ts the basicstate temperature, Ps the basic state pressure, and psthe basic state density. For numerical grids that incorporate terrain slopesless than 5- and have horizontal scales much largerthan the vertical scales, the hydrostatic approximation,Eq. (A3), is sufficiently accurate. Closure for the prognostic equations is based on Ktheory. The vertical exchange coefficient for momentum Km is determined from mixing-length theory andthe turbulence kinetic energy. The other vertical exchange coefficients, Kq, Kn, Ke, and Kx are a functionof Kin. The horizontal exchange coefficient Ka is calculated from the deformation rate. In Eq. (A6),/~ is aconstant and 1 is a length scale. Additional details ofthe turbulence parameterizations and closure methodsare described in Paegle and McLawhorn (1983) andMcCorcle (1988).REFERENCESAndre, J.-C., J.-P. Goutorbe and A. Perrier, 1986: HAPEX-MOB ILHY: A hydrologic atmospheric experiment for the study of water budget and evaporation flux at the climatic scale. Bull. Amer. Meteor. Soc., 67, 138-144. , ---, F. Becker, P. Bessemoulin and collaborators, 1988: Evaporation over land surfaces: First results for HAPEX MOBILHY special observation period. Annales Geophysicae, 6, 477-492.Anthes, R. A., 1974: Data assimilation and initialization of hurricane prediction models. J. Atrnos. Sci., 31, 702-719. .. , Y.-H. Kuo, S. G. Benjamin and Y.-F. Li, 1982: The evolution of the mesoseale environment of severe local storms: Preliminary modeling studies. Mon. Wea. Rev., 110, 1187-1213.Avissar, R., and R. A. Pielke, 1989: A parameterization of heterogeneous land surfaces for atmospheric numerical models andits impact on regional meteorology. Mon. Wea. Rev., 117, 21132136.Barnes, S. L., 1964: A technique for maximizing details in numerical weather map analysis. J. Appl. Meteor., 3, 396-409.Bothwell, P. D., 1986: The role of moisture convergence on initiating and sustaining convection. MESOS Data Analysis Note Number 4. Technical Attachment SR4-15-86, NOAA / NWS, Southern Region Headquarters, Fort Worth, Texas. 4 pp.Dickinson, R. E., A. Henderson-Sellers, P. J. Kennedy and M. F. Wilson, 1986: Biosphere-atmosphere transfer scheme (BATS) for the NCAR community climate model. NCAR Technical Note NCAR/TN 275+STR. 69 pp. [Available from National Center for Atmospheric Research, P. O. Box 3000, Boulder, CO 80307.]Fast, J. D., and M. D. McCorcle, 1990: A two-dimensional numerical sensitivity study of the Great Plains low-level jet. Mon. Wea. Rev., 118, 151-163.Foth, H. D., and J. W. Schafer, 1980: Soil geography and land use.John Wiley and Sons, Inc., New York., 484 pp.Hoke, J. E., and R. A. Anthes, 1976: The initialization of numerical models by a dynamic-initialization technique. Mon. Wea. Rev., 104, 1551-1556.Holton, J. R., 1967: The diurnal boundary layer wind oscillationabove sloping terrain. Tellus, 19, 119-205.Idso, S. B., R. D. Jackson, R. J. Reginato, B. A. Kimball and F. S. Nakayama, 1975: The dependency of bare soil albedo on soil water content. J. Appl. Meteor., 14, 109-113.Lanicci, J. M., T. N. Carlson and T. T. Warner, 1987: Sensitivity ofthe Great Plains severe-storm environment to soil-moisture distribution. Mon. Wea. Rev., 115, 2660-2673.Mahfouf, J., E. Richard and P. Mascart, 1987: The influence of soil and vegetation on the development of mesoscale circulations. J. Climate Appl. Meteor., 26, 1483-1495.Mahrt, L., and H. Pan, 1984: A two-layer model of soil hydrology. Bound. Layer Meteor., 29, 1-20.Mattbews, E., 1985: Atlas of archive vegetation, land-use and seasonal albedo data sets. NTIS No. N8524508, NASA Technical Mem orandum 86199, Goddard Space Flight Center, Institute for Space Studies, New York. 53 pp.McCorcle, M. D., 1988: Simulation of surface-moisture effects on the Great Plains low-level jet. Mon. Wea. Rev., 116, 1705- ~, 20.Meehl, G. A., and W. M. Washington, 1988: A comparison of soil moisture sensitivity in two global climate models. J. Atmos. Sci., 45, 1476-1492.Ookouchi, Y., M. Segal, R. C. Kessler and R. A. Pietke, 1984: Eval uation of soil moisture effects on the generation and modification of mesoscale circulations. Mon. Wea. Rev., 112, 2281-2292.Ofianski, I, 1976: A simple boundary condition for unbounded hy perbolic flows. J. Comput. Phys., 21,251-269.Paegle, J., and D. W. McLawhorn, 1983: Numerical modeling of diurnal convergence oscillations above sloping terrain. Mon. Wea. Rev., 111, 67-85.Pan, H.-L., and L. Mahrt, 1987: Interaction between soil hydrology and boundary-layer development. Bound.-Layer Meteor., 38, 185-202.Pielke, R. A., 1984: Mesoscale Meteorological Modeling. AcademicPress, New York., 612 pp.Pinty, J.-P., P. Mascart, E. Richard and R. Rosset, 1989: An inves tigation of mesoscale flows induced by vegetation inhomogene ities using an evapotranspiration model calibrated against HAPEX-MOBILHY data. J. Appl Meteor., 28, 976-992.Segal, M., R. Avissar, M. C. McCumber and R. A. Pielke, 1988: Evaluation of vegetation effects on the generation and modifi cation of mesoseale circulations. J. Atmos. Sci., 45, 2268-2291. , W. E. Schreiber, G. Kallos, J. R. Garratt, A. Rodi, J. Weaver and R. A. Pielke, 1989: The impact of crop areas in northeast Colorado in midsummer mesoscale thermal circulations. Mon. Wea. Rev., 117, 809-825.Sellers, P. J., F. G. Hall, G. Asrar, D. E. Strebel and R. E. Murphy, 1988: The first ISLSCP field experiment (FIFE). Bull Amer. Meteor. Soc., 69, 22-27.Sherretz, L. A., and D. W. Fulker, 1988: Unidata: Enabling univer sities to acquire and analyze scientific data. Bull Amer. Meteor. Soc., 69, 373-376.Taylor, S. A., and G. L. Ashoroft, 1972: Physical Edaphology. Free man Company, San Francisco., 533 pp.Tremback, C. J., J. Powell, W. R. Cotton and R. A. Pielke, 1987: The forward-in-time upstream advection scheme: Extensions to higher orders. Mon. Wea. Rev., 115, 540-555.Waldsteicher, J. S., 1989: A guide to utilizing moisture flux conver gence as a predictor of convection. Nat. Wea. Dig., 14, 20-35.Wetzel, P. J., and J.-T. Chang, 1988: Evapotranspiration from non uniform surfaces: A first approach for short-term numerical weather prediction. Mon. Wea. Rev., 116, 600-621.Wilson, M. F., and A. Henderson-Sellers, 1985: A global archive of land cover and soils data for use in general circulation climate models. J. Climatol., 5, 119-143. , R. E. Dickinson and P. J. Kennedy, 1987: Sensitivity of ~he biosphere-atmosphere transfer scheme (BATS) to the in clusion of variable soil characteristics. J. ClimateAppl. Meteor., 26, 341-362. Yan, H., and R. A. Anthes, 1988: The effect of variations in surface moisture on mesoscale circulations. Mon. Wea. Rev., 116, 192 208.