• Anderson, J. R., , E. E. Hardy, , J. T. Roach, , and R. E. Witmer, 1976: A land use and land cover classification system for use with remote sensor data. U.S. Geological Survey Professional Paper 964, 36 pp.

  • Beljaars, A. C. M., 1995: The parameterization of surface fluxes in large-scale models under free convection. Quart. J. Roy. Meteor. Soc., 121 , 255270.

    • Search Google Scholar
    • Export Citation
  • Bornstein, R., , and Q. Lin, 2000: Urban heat islands and summertime convective thunderstorms in Atlanta: Three case studies. Atmos. Environ., 34 , 507516.

    • Search Google Scholar
    • Export Citation
  • Bright, D. R., , and S. L. Mullen, 2002: The sensitivity of the numerical simulation of the southwest monsoon boundary layer to the choice of PBL turbulence parameterization in MM5. Wea. Forecasting, 17 , 99114.

    • Search Google Scholar
    • Export Citation
  • Burian, S. J., , and J. M. Shepherd, 2005: Effect of urbanization on the diurnal rainfall pattern in Houston. Hydrol. Processes, 19 , 10891103.

    • Search Google Scholar
    • Export Citation
  • Cotton, W. R., , and R. A. Pielke, 1995: Human Impacts on Climate. Cambridge University Press, 288 pp.

  • Deardorff, J., 1972: Theoretical expression for the countergradient vertical heat flux. J. Geophys. Res., 77 , 59005904.

  • Diem, J. E., , and D. P. Brown, 2003: Anthropogenic impacts on summer precipitation in central Arizona, U.S.A. Prof. Geogr., 55 , 343355.

    • Search Google Scholar
    • Export Citation
  • Doran, J. C., , C. M. Berkowitz, , R. L. Coulter, , W. J. Shaw, , and C. W. Spicer, 2003: The 2001 Phoenix Sunrise experiment: Vertical mixing and chemistry during the morning transition in Phoenix. Atmos. Environ., 37 , 23652377.

    • Search Google Scholar
    • Export Citation
  • Dudhia, J., 1996: A multi-layer soil temperature model for MM5. Preprints. Sixth PSU/NCAR Mesoscale Model Users’ Workshop, Boulder, CO, National Center for Atmospheric Research, 49–50.

    • Search Google Scholar
    • Export Citation
  • Fast, J. D., , J. C. Doran, , W. J. Shaw, , R. L. Coulter, , and T. J. Martin, 2000: The evolution of the boundary layer and its effect on air chemistry in the Phoenix area. J. Geophys. Res., 105D , 2283322848.

    • Search Google Scholar
    • Export Citation
  • Fernando, H. J. S., , S. M. Lee, , J. Anderson, , M. Princevac, , E. Pardyjak, , and S. Grossman-Clarke, 2001: Urban fluid mechanics: Air circulation and contaminant dispersion in cities. Environ. Fluid Mech., 1 , 107164.

    • Search Google Scholar
    • Export Citation
  • Fisher, B. E. A., , S. Joffre, , J. Kukkonen, , M. Piringer, , M. Rotach, , and M. Schatzmann, 2005: Meteorology applied to urban air pollution problems. European Commission COST Action Final Rep. 715, 282 pp.

  • GP2100, 2003: Greater Phoenix regional atlas: A preview of the region’s 50-year future. Arizona State University, 70 pp. [Available online at http://www.asu.edu/copp/morrison/Atlas-5.pdf.].

  • Grell, A. G., , J. Dudhia, , and D. R. Stauffer, 1995: A description of the fifth-generation Penn State/NCAR Mesoscale Model (MM5). NCAR Tech. Note NCAR/TN-398+STR, 117 pp.

  • Grossman-Clarke, S., , J. A. Zehnder, , W. L. Stefanov, , Y. Liu, , and M. A. Zoldak, 2005: Urban modifications in a mesoscale meteorological model and the effects on near-surface variables in an arid metropolitan region. J. Appl. Meteor., 44 , 12811297.

    • Search Google Scholar
    • Export Citation
  • Guo, Y. R., , and S. Chen, 1994: Terrain and land use for the fifth-generation Penn State/NCAR Mesoscale Modeling System (MM5). NCAR Tech. Note NCAR/TN-397+IA, 114 pp.

  • Hong, S-Y., , and H-L. Pan, 1996: Nonlocal boundary layer vertical diffusion in a medium-range forecast model. Mon. Wea. Rev., 124 , 23222339.

    • Search Google Scholar
    • Export Citation
  • Lee, S-M., , W. Giori, , M. Princevac, , and H. J. S. Fernando, 2006: Implementation of a stable PBL turbulence parameterization for the mesoscale model MM5: Nocturnal flow in complex terrain. Bound.-Layer Meteor., 119 , 109134.

    • Search Google Scholar
    • Export Citation
  • Liu, Y., , F. Chen, , T. Warner, , and J. Basara, 2006: Verification of a mesoscale data-assimilation and forecasting system for the Oklahoma City area during the Joint Urban 2003 field project. J. Appl. Meteor. Climatol., 45 , 912929.

    • Search Google Scholar
    • Export Citation
  • Rife, D. L., , C. A. Davis, , Y. Liu, , and T. T. Warner, 2004: Predictability of low-level winds by mesoscale meteorological models. Mon. Wea. Rev., 132 , 25532569.

    • Search Google Scholar
    • Export Citation
  • Rozoff, C. M., , W. R. Cotton, , and J. O. Adegoke, 2003: Simulation of St. Louis, Missouri, land use impacts on thunderstorms. J. Appl. Meteor., 42 , 716738.

    • Search Google Scholar
    • Export Citation
  • Sailor, D. J., , and L. Lu, 2004: A top-down methodology for developing diurnal and seasonal anthropogenic heating profiles for urban areas. Atmos. Environ., 38 , 27372748.

    • Search Google Scholar
    • Export Citation
  • Seibert, P., , F. Beyrich, , S. E. Gryning, , S. Joffre, , A. Rasmussen, , and P. Tercier, 1998: Mixing height determination for dispersion modelling. Harmonisation of the pre-processing of meteorological data for atmospheric dispersion models, B. E. A. Fisher et al., Eds., European Commission COST Action Final Rep. 710, 121 pp. [Available online at http://www.wau.boku.ac.at/fileadmin/_/H81/H814/Downloads/cost710-2.pdf.].

  • Seibert, P., , F. Beyrich, , S. E. Gryning, , S. Joffre, , A. Rasmussen, , and P. Tercier, 2000: Review and intercomparsion of operational methods for the determination of the mixing height. Atmos. Environ., 34, 1001–1027.

    • Search Google Scholar
    • Export Citation
  • Shaw, W. J., , J. C. Doran, , and R. L. Coulter, 2005: Boundary-layer evolution over Phoenix, Arizona and the premature mixing of pollutants in the early morning. Atmos. Environ., 39 , 773786.

    • Search Google Scholar
    • Export Citation
  • Shepherd, J. M., 2006: Evidence of urban-induced precipitation variability in arid climate regimes. J. Arid Environ., 67 , 607628.

  • Stefanov, W. L., , M. S. Ramsey, , and P. R. Christensen, 2001: Monitoring urban land cover change: An expert system approach to land cover classification of semiarid to arid urban centers. Remote Sens. Environ., 77 , 173185.

    • Search Google Scholar
    • Export Citation
  • Stull, R. B., 1988: An Introduction to Boundary Layer Meteorology. Kluwer Academic Publishers, 665 pp.

  • Troen, I., , and L. Mahrt, 1986: A simple model of the atmospheric boundary layer: Sensitivity to surface evaporation. Bound.-Layer Meteor., 37 , 129148.

    • Search Google Scholar
    • Export Citation
  • Zhang, D-L., , and W-Z. Zheng, 2004: Diurnal cycles of surface winds and temperatures as simulated by five boundary layer parameterizations. J. Appl. Meteor., 43 , 157169.

    • Search Google Scholar
    • Export Citation
  • Zhong, S., , and J. D. Fast, 2003: An evaluation of the MM5, RAMS, and Meso Eta models at subkilometer resolution using VTMX field campaign data in the Salt Lake Valley. Mon. Wea. Rev., 131 , 13011322.

    • Search Google Scholar
    • Export Citation
  • View in gallery

    Location of the four MM5 modeling domains as used in this study. Grid spacings are 54 km for domain 1 (3294 km east–west by 2700 km north–south), 18 km for domain 2 (1350 km east–west by 1080 km north–south), 6 km for domain 3 (594 km east–west by 414 km north–south), and 2 km for domain 4 (212 km east–west by 132 km north–south).

  • View in gallery

    Land use and terrain (m) as produced by MM5’s TERRAIN preprocessor for the inner modeling domain (2 km × 2 km) based on Landsat Thematic Mapper 1998 image-derived land use/cover data and three urban land use/cover classes: urban built-up, urban mesic residential (irrigated), and urban xeric residential (arid, partly irrigated). Other land use/cover categories were the same as in the standard MM5 24-category USGS dataset. Sites labels for radar wind and radiosonde profilers are 1 for Sky Harbor Airport, 2 for McDowell Mountain Park, and 3 for Waddell. Coordinates are given as universal transverse Mercator zone 12 (UTM12) easting and northing.

  • View in gallery

    Bias, RMSE, and SDE for 2-m air temperature, 10-m wind speed, and 10-m wind direction for each hour of the day for the 10 May–10 Jun 1998 simulation period for the NWS station at Sky Harbor Airport (Fig. 2; site 1). Plots represent standard MM5 (dashed line) and MM5 with modified MRF scheme (solid line) according to Liu et al. (2006), updated land use data, and urban energy balance scheme (Grossman-Clarke et al. 2005).

  • View in gallery

    As in Fig. 3, but for the 10–30 Jun 2001 simulation period.

  • View in gallery

    Measured and simulated composite profiles of (top) potential temperature (K) and (bottom) specific humidity (g kg−1) for the period of 10–31 May 1998 at Sky Harbor Airport (Fig. 2; site 1—7 radiosonde launch days for each period) at 0800, 1000, 1200, 1400, and 1700 LST. Plots represent observed (solid line), simulated using the modified MM5 (filled circles), simulated using the original MRF scheme with modified land use characteristics (solid triangles), and simulated using the standard MM5 (open triangles). Note that the vertical scale differs among panels.

  • View in gallery

    As in Fig. 5, but for the period of 1–10 Jun 1998.

  • View in gallery

    Measured and simulated profiles of potential temperature at 1700 LST for (a) 21 May and (b) 1 Jun 1998. Plots represent measured at Sky Harbor Airport (Fig. 2; site 1) (solid line), routine NWS radiosonde measurement at Flagstaff (solid circles), routine NWS radiosonde measurement at Tucson (open triangles), NCEP Eta Model simulated (plus signs), and MM5 simulated using the modified MRF scheme by Liu et al. (2006) (open circles).

  • View in gallery

    As in Fig. 5, but for the period of 10–30 Jun 2001 (12 radiosonde launch days) for 0500, 0600, 0800, 0900, and 1000 LST. Plots represent observed (solid line), simulated using the modified MM5 (filled circles), and simulated using the original MRF scheme with modified land use characteristics (solid triangles).

  • View in gallery

    Measured and simulated composite PBL heights above ground level for the period of 10 May–10 Jun 1998 at (a) Sky Harbor Airport and (b) McDowell Mountain Park (Fig. 2; sites 1 and 2, respectively). Plots represent measured (times signs), simulated using the modified MM5 (solid circles), simulated using the original MRF scheme with modified urban land use/cover and urban energy balance (solid triangles), and simulated using the standard MM5 (open triangles).

  • View in gallery

    As in Fig. 9, but for the period of 10–30 Jun 2001 at (a) Sky Harbor Airport and (b) Waddell (Fig. 2; sites 1 and 3, respectively). Plots represent measured (times signs), simulated using the modified MM5 (solid circles), and simulated using the original MRF scheme with modified urban land use/cover and urban energy balance (solid triangles).

  • View in gallery

    Vertical composite wind profiles for the period of 10 May–10 Jun 1998 near Sky Harbor Airport (Fig. 2; site 1). For clarity, the data for the lowest 300 m of the atmosphere are given as insets with a higher resolution. Plots represent measured (times signs), simulated using the modified MM5 (solid circles), simulated using the original MRF scheme with modified urban land use/cover and urban energy balance (solid triangles), and simulated using the standard MM5 (open triangles).

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 18 18 5
PDF Downloads 18 18 5

Simulations of the Urban Planetary Boundary Layer in an Arid Metropolitan Area

View More View Less
  • 1 Global Institute of Sustainability, Arizona State University, Tempe, Arizona
  • 2 Research Applications Laboratory, National Center for Atmospheric Research,* Boulder, Colorado
  • 3 Department of Atmospheric Sciences, Creighton University, Omaha, Nebraska
  • 4 Pacific Northwest National Laboratory, Richland, Washington
© Get Permissions
Full access

Abstract

A modified version of the fifth-generation Pennsylvania State University–National Center for Atmospheric Research Mesoscale Model (MM5) was applied to the arid Phoenix, Arizona, metropolitan region. The ability of the model to simulate characteristics of the summertime urban planetary boundary layer (PBL) was tested by comparing model results with observations from two field campaigns conducted in May/June 1998 and June 2001. The modified MM5 included a refined land use/cover classification and updated land use data for Phoenix and bulk approaches of characteristics of the urban surface energy balance. PBL processes were simulated by a version of MM5’s Medium-Range Forecast Model (MRF) scheme that was enhanced by new surface flux and nonlocal mixing approaches. Simulated potential temperature profiles were tested against radiosonde data, indicating that the modified MRF scheme was able to simulate vertical mixing and the evolution and height of the PBL with good accuracy and better than the original MRF scheme except in the late afternoon. During both simulation periods, it is demonstrated that the modified MM5 simulated near-surface air temperatures and wind speeds in the urban area consistently and considerably better than the standard MM5 and that wind direction simulations were improved slightly.

* The National Center for Atmospheric Research is sponsored by the National Science Foundation

Corresponding author address: Susanne Grossman-Clarke, Global Institute of Sustainability, Arizona State University, P.O. Box 873211, Tempe, AZ 85287-3211. Email: sg.clarke@asu.edu

Abstract

A modified version of the fifth-generation Pennsylvania State University–National Center for Atmospheric Research Mesoscale Model (MM5) was applied to the arid Phoenix, Arizona, metropolitan region. The ability of the model to simulate characteristics of the summertime urban planetary boundary layer (PBL) was tested by comparing model results with observations from two field campaigns conducted in May/June 1998 and June 2001. The modified MM5 included a refined land use/cover classification and updated land use data for Phoenix and bulk approaches of characteristics of the urban surface energy balance. PBL processes were simulated by a version of MM5’s Medium-Range Forecast Model (MRF) scheme that was enhanced by new surface flux and nonlocal mixing approaches. Simulated potential temperature profiles were tested against radiosonde data, indicating that the modified MRF scheme was able to simulate vertical mixing and the evolution and height of the PBL with good accuracy and better than the original MRF scheme except in the late afternoon. During both simulation periods, it is demonstrated that the modified MM5 simulated near-surface air temperatures and wind speeds in the urban area consistently and considerably better than the standard MM5 and that wind direction simulations were improved slightly.

* The National Center for Atmospheric Research is sponsored by the National Science Foundation

Corresponding author address: Susanne Grossman-Clarke, Global Institute of Sustainability, Arizona State University, P.O. Box 873211, Tempe, AZ 85287-3211. Email: sg.clarke@asu.edu

1. Introduction

Phoenix, Arizona, is the second-fastest-growing major city in the United States and is shifting from a midsized regional center to one of the nation’s largest metropolitan areas. Average annual population growth in the past 10 yr amounted to 4.1%, which represents a change in population from about 2.5 to 3.5 million people in the greater Phoenix region. The current size of the built-up urban area is about 3500 km2 (GP2100).

Urbanization significantly affects regional near-surface air temperatures, wind fields, and the evolution of the planetary boundary layer (PBL), modifies local circulations, and subsequently influences air quality, human comfort, and health (Cotton and Pielke 1995; Fernando et al. 2001). Mesoscale atmospheric models are increasingly employed to improve the understanding of these processes that are strongly influenced by the energy and momentum exchange between the atmosphere and the underlying surface. The quality of the simulations depends on characterizing the urban and rural land cover accurately and characterizing the physical approaches of processes related to the urban surface energy balance and PBL.

The goal of this study was to test the ability of a modified version of the fifth-generation Pennsylvania State University–National Center for Atmospheric Research (NCAR) Mesoscale Model (MM5) to simulate characteristics of the urban PBL for the arid Phoenix metropolitan region under typical summer conditions of the desert Southwest, which are characterized predominantly by high pressure systems with weak synoptic-scale forcing.

The modified version of MM5 used in this study included a refined land cover classification and updated land use/cover data for the Phoenix region as well as bulk approaches for characteristics of the urban surface energy balance such as increased heat storage flux, the production of anthropogenic heat, and radiation trapping (Grossman-Clarke et al. 2005). The nonlocal closure Medium-Range Forecast Model (MRF) boundary layer scheme (Hong and Pan 1996) was replaced by the version of the MRF scheme described by Liu et al. (2006). MM5 with the original MRF scheme (Hong and Pan 1996) has been shown to simulate the atmospheric boundary layer in the southwestern United States better than do other available PBL schemes in MM5 (Bright and Mullen 2002). However, there is a tendency for the MRF scheme to underestimate near-surface wind speeds and to overestimate sensible heat fluxes and boundary layer heights under free-convection conditions (Zhang and Zheng 2004; Liu et al. 2006). The model’s behavior is caused by the empirical description of the convective velocity and was discussed in detail in Liu et al. (2006). To resolve problems of poor surface momentum flux representation in the original MRF scheme, Liu et al. (2006) introduced a formulation of the convective velocity by Deardorff (1972) and Beljaars (1995) and replaced the bulk Richardson number that is used in the PBL height calculations by a local Richardson number.

In a previous study (Grossman-Clarke et al. 2005), it was shown for a 3-day simulation period that these changes to MM5 considerably improved the regional response of the near-surface meteorological variables. However, the influence of the modifications to MM5 on the simulated evolution of the urban PBL and the model performance for near-surface variables over a longer time period were not tested. To use the modified version of MM5 in future studies to analyze the effect of urbanization on meteorological processes in the region with higher confidence, its performance was evaluated by comparing the models’ outputs with surface meteorological data, upper-air radiosonde data, and radar wind profiler data from two field campaigns conducted in May/June of 1998 (Fast et al. 2000) and June of 2001 (Doran et al. 2003) in the Phoenix metropolitan area, spanning an extended time period of 53 days.

2. Data, model, and analysis methods

a. Design of numerical simulations

MM5, version 3.7, was employed for the two simulation periods of the two field experiments: 10 May 1998–10 June 1998 and 10–30 June 2001. The simulations were carried out in 72-h increments, allowing 24 overlapping hours for model spinup. The runs were started at 0000 UTC [1700 local standard time (LST)]. Nested simulations with four domains, with horizontal grid spacings of 54 km (3294 km east–west by 2700 km north–south), 18 km (1350 km east–west by 1080 km north–south), 6 km (594 km east–west by 414 km north–south), and 2 km (212 km east–west by 132 km north–south), respectively, and 51 vertical layers were performed with MM5 (Fig. 1). The lowest prognostic level was approximately 7 m above ground level. The innermost domain included the Phoenix metropolitan area, the surrounding desert, and agricultural land. Initial and boundary conditions were provided by the National Centers for Environmental Prediction (NCEP) Eta grid 212 (40-km spacing) analysis and, if not available (as was the case for three days in June 1998), by the NCEP–NCAR reanalysis, including assimilation of upper-air observations. Every 6 h, the lateral boundary conditions were updated from the Eta analysis and NCEP–NCAR reanalysis. PBL processes were included by using the original and modified versions of the nonlocal closure MRF scheme (Hong and Pan 1996; Liu et al. 2006). The 5-layer soil model (Dudhia 1996) was enhanced by bulk approaches for characteristics of the urban energy budget as described in section 2b. Four-dimensional data assimilation (FDDA) was applied to the grid with a 54-km spacing over the 72-h duration of the 1998 simulations to improve MM5’s ability to describe the observed trend in the maximum 2-m air temperature T2m over the simulation period. It was shown in Grossman-Clarke et al. (2005) that FDDA did not change the differences in the maximum simulated T2m between the land cover scenarios by more than 1 K and that therefore it could not compensate for deficits in the model’s surface energy balance as mediated through land use/cover, which led to considerably larger differences of about 4 K in the simulated T2m. FDDA was used in MM5 through three-dimensional analysis nudging for wind and temperature; that is, forcing functions were added to the governing equations to nudge the model state gradually toward the observations (Grell et al. 1995).

b. Model description

The spatial extent, nature, and heterogeneity of urban land cover across the rapidly urbanizing arid Phoenix metropolitan region is not represented well in the 24-category U.S. Geological Survey (USGS) Land Use/Land Cover System (Anderson et al. 1976) as provided with MM5, which includes only one urban land use/cover type (Guo and Chen 1994) and reflects the region as of the 1960s. Therefore, as described in Grossman-Clarke et al. (2005), a new land cover classification was developed using land cover data derived from 1998 Landsat Thematic Mapper satellite images (Stefanov et al. 2001). Two additional urban land cover classes were introduced into the existing 24-category USGS land cover classification used in MM5 to give three urban categories: urban built-up, urban mesic residential, and urban xeric residential, which were distinguished by the type of vegetation and irrigation (no vegetation, well-watered flood or sprinkle irrigated, and drought-adapted vegetation with drip irrigation, respectively). The land cover distributions and contours of terrain for the inner domain (2 km × 2 km grid spacing) as generated by MM5’s “TERRAIN” preprocessor, based on the USGS 30-s land use data, are given in Fig. 2. The urban/suburban core is surrounded by irrigated agricultural land and a dry, sparsely vegetated natural desert that is assigned as shrubland category 8 in the USGS classification.

Characteristics of the urban energy balance that contribute to increased nighttime temperatures are anthropogenic heat production, sky-view factor in the longwave radiation balance, increased surface volumetric heat storage capacity, and thermal conductivity. A detailed description of physical parameter choices for the urban land use classes was given in Grossman-Clarke et al. (2005). A value of 0.85 was determined for the sky-view factor of the urban land use categories. We adopted the approach of Sailor and Lu (2004) to calculate hourly profiles of anthropogenic heat based on resident and working population density data, electricity consumption, and vehicle miles traveled. Urban-induced modifications on the momentum exchange between the atmosphere and the surface were included through adjustments in the aerodynamic roughness length.

To evaluate the performance of the modified MRF scheme in this study for the Phoenix area, PBL processes were included in the simulations through use of the original and modified MRF schemes. In both versions of the MRF scheme, the nonlocal-K approach by Troen and Mahrt (1986) is used, which represents large-eddy turbulence within the well-mixed boundary layer by incorporating a correction to the local gradient of the prognostic variables in the turbulence diffusion equation.

In the original MRF scheme by Hong and Pan (1996), the PBL height h is calculated as a function of a critical bulk Richardson number, Ribcr, the horizontal wind speed at h U(h), and the virtual potential temperatures at the lowest prognostic level (θυa), at the PBL height [θυ(h)], and at the surface (θs = θυa + θT, with θT being the virtual temperature access near the surface):
i1558-8432-47-3-752-e1

Problems arise under convective conditions from an overestimation of the friction velocity and therefore very weak surface winds (Zhang and Zheng 2004) and an overestimation of PBL heights (Liu et al. 2006) and subsequent vertical mixing. This situation is due to the fact that h appears in the profile function used to calculate diffusivity coefficients under convective conditions (Hong and Pan 1996).

To improve the surface momentum flux representation in the MRF scheme, Liu et al. (2006) introduced a formulation of the convective velocity w* that was described by Deardorff (1972) and further studied by Beljaars (1995). The convective velocity is added to the magnitude of the mean wind vector in the surface flux calculations to consider extra eddy mixing induced by surface-layer free-convective instability. Furthermore, in the modified MRF scheme, the PBL height is determined by using a local bulk Richardson number Ribl that is computed for each model layer based on wind shear and thermal stability between two neighboring model levels. Instability searching from the bottom up is applied by comparing Ribl with a critical Richardson number, Ric = 0.25; if Ribl > Ric is found for a model layer, the PBL height is set to the height of the top of the layer plus an adjustment that is obtained by upward extrapolation based on the difference between Ribl and Ric.

c. Experimental data

Phoenix is located in the center of the Salt River valley at a mean elevation of about 335 m above sea level. There is typically little synoptic forcing in the desert Southwest during the spring, early and late summer, and autumn months, which are generally dry. The focus of this study was on the latter conditions when thermal flows due to horizontal pressure gradients caused by complex terrain as well as potential mesoscale circulations due to land use/cover variation are the dominant climatological features and determine the local winds.

A meteorological and atmospheric-chemistry field study was carried out in the Phoenix metropolitan area from 10 May to 10 June 1998 to study the convective PBL. The details of the experimental sites and instrumentation were described in Fast et al. (2000). The relevant sites for this study for the 915-MHz radar wind profiler are shown in Fig. 2 and included 1) Sky Harbor Airport (elevation 350 m) and 2) the rural McDowell Mountain Park (elevation 616 m). Each profiler was configured to operate in a high-resolution mode with 60-m range gates and a lower-resolution mode with 100-m range gates. The data were continuously collected and were aggregated to hourly values of wind speed and wind direction. At the Sky Harbor Airport site, radiosondes were released on 14 days at 0800, 1000, 1200, 1400, and 1700 LST to measure profiles of potential temperatures. At the McDowell Mountain Park sites vertical profiles of potential temperature θ were measured on 11 days at 0800, 1000, 1200, and 1700 LST and on 8 days at 1400 LST. During the 1998 experimental period, winds above 2000 m were generally from the south or southwest at all times of the day and were associated with a series of upper-level troughs during this period (Fast et al. 2000).

The 2001 Phoenix Sunrise experiment was conducted from 10 to 30 June 2001 to obtain information on the evolving structure of the PBL during the morning transition period in Phoenix (Doran et al. 2003). Upper-air information (half-hour averages of wind data) was obtained on 16 days from 915-MHz radar wind profilers near Sky Harbor Airport (Fig. 2; site 1) and at Waddell, which is also in the vicinity of Phoenix (Fig. 2; site 3, elevation 428 m). Radiosondes were released at the Sky Harbor Airport site on 12 days at 0000, 0200, 0500, 0600, 0800, 0900, and 1000 LST. At the Waddell site, radiosondes were launched on 12 days at 0500, 0700, 0800, 0900, and 1000 LST. For most days of the 2001 experimental period, characteristic early-summer conditions occurred, with light winds and very low dewpoints. Data for evaluating the ability of mesoscale models to describe the evolution of the morning PBL are generally scarce, and hence there is a high value in the data from the 2001 field experiment (Doran et al. 2003).

3. Comparison of model results with observations

a. Surface temperature and winds

MM5’s performance was evaluated by means of comparing simulated 2-m air temperatures T2m, 10-m wind speeds V10m, and wind direction with data from the National Weather Service (NWS) station at Sky Harbor Airport. In MM5, T2m and V10m are determined diagnostically by means of the Monin–Obukhov similarity theory under consideration of the atmospheric stability from the simulated temperatures at ground level and simulated temperatures and components of horizontal wind speed at the lowest prognostic level.

The statistical measures of bias, root-mean-square error (RMSE), and standard deviation (SDE) were calculated for each hour of the day for the two simulation periods from the standard MM5 (original MRF scheme; USGS land use data) and the modified version as described in section 2b [MRF scheme according to Liu et al. (2006), and updated land use data and improved representation of the urban energy balance according to Grossman-Clarke et al. (2005)]. In the statistical calculations, observed data were subtracted from the simulated data, giving positive model biases when mean errors were positive. The results are shown for the two simulation periods 10 May–10 June 1998 and 10–30 June 2001 in Figs. 3 and 4, respectively.

In the simulations for 10 May–10 June 1998 with the standard MM5 version, a strong cool bias for T2m at all times of the day was found that was particularly large at night (∼5 K) because physical processes in the surface energy balance that contribute to the nocturnal urban heat island were not included in the model (Fig. 3). A cold bias of 2–3 K during daytime was caused by the high moisture availability and subsequent evapotranspiration for the single urban category in the standard MM5. RMSE values were particularly high for the standard MM5 during night (up to 5 K).

The model surface temperature simulation was consistently improved when applying the modified MM5 version (Fig. 3), with the cold bias and RMSE reduced at most hours of the day to within ±1 K and ∼2 K, respectively. An analysis of the influence of land use and surface energy balance characteristics on the T2m was presented in Grossman-Clarke et al. (2005), with the conclusion that during daytime hours the T2m were mostly influenced by the moisture availability for evaporation and transpiration in the urban area and also by the surrounding desert whereas nighttime temperatures were influenced highly by characteristics of the urban energy balance such as anthropogenic heating, limited sky view, and heat storage. Turbulent vertical diffusion was the dominant effect contributing to near-surface air temperatures in MM5 under typical early-summer conditions in Phoenix. Therefore, the improved results for T2m with the modified version of MM5 reflect the improved land use characterization as well as the physical representation of urban energy balance. On the other hand, the choice of the PBL scheme (original vs modified MRF) had no significant effect on T2m when land use characteristics were the same in the simulations (results not shown).

The largest bias for the modified MM5 of about 2 K occurred in the early-morning hours around 0500 LST (Fig. 3). Early-morning temperatures were overestimated by the model by up to 5 K on four days during the 1998 period. Because the Phoenix metropolitan area is situated in complex terrain under weak synoptic conditions, the diurnal pattern of the airflow consists of widespread predominantly westerly anabatic winds in the afternoon that develop in conjunction with the daily insolation. In the hours from early morning to around noon, drainage winds from the southeast, east, and north to northwest occur (Fernando et al. 2001). Using data from the 2001 Phoenix field campaign, Shaw et al. (2005) identified density currents that arrived in the Phoenix valley in the early-morning hours and that were usually a few hundred meters deep and were associated with cold advection.

On two of the four days, when the modified MM5 greatly overestimated early morning T2m, the simulated arrival of the easterly downslope flow was delayed by several hours. For the other two days, the simulated direction of the flow never changed from westerly to a flow with an easterly component, which result was contrary to the measurements. MM5 did not capture the associated sudden drop in the measured T2m before sunrise on those days. Under the latter conditions, a relatively strong westerly synoptic flow suppressed the development of the terrain-induced flow. It is common for mesoscale models not to simulate the onset of flow reversal in a timely manner (Zhong and Fast 2003; Lee et al. 2006) because the dynamic and thermodynamic effects of the complex terrain on the ambient flow and regional pressure gradients are not captured adequately. Problems might also result from motions that are not deterministically predictable because they occur on subdiurnal time scales that are not represented in the observations used to determine MM5’s initial conditions (Rife et al. 2004).

Unlike the simulated T2m, the wind speed at 10-m height V10m was influenced only slightly by the land use characteristics, but rather was influenced mainly by the choice of the PBL scheme. The modified version of the MRF scheme led to a much-improved agreement between simulated and observed V10m during daytime (Fig. 3) in comparison with the V10m obtained with the original version of the MRF scheme, which were greatly underestimated, thereby confirming results by Liu et al. (2006) and Zhang and Zheng (2004). The original MRF’s behavior originated from the high values of the convective velocity that resulted in an overestimation of the friction velocity and vertical momentum transport. The slow bias of the original MRF scheme during daytime hours was reduced from up to −3 m s−1 (RMSE ∼ 3–4 m s−1) to less than 0.5 m s−1 (RMSE ∼ 1–2 m s−1), and wind speed during nighttime hours was relatively unchanged. The nighttime wind speed bias was in the range of ±1 m s−1 for both model versions. As for T2m, the SDE values for wind speed were similar for the two model versions but were slightly larger during nighttime for the standard MM5, with values approximately between 1 and 2 m s−1.

The modifications to the MRF scheme improved the simulation results for wind direction slightly. The bias was in the range of ±50° for both model versions. The RMSE was improved for the modified MM5 to values between 50° and 70° for the hours from 1000 LST to midnight as compared with RMSE values up to ∼100° obtained with the standard MM5.

The highest RMSE values of 80°–90° were calculated for the modified MM5 during the early-morning hours and were also higher than for the standard MM5 (60°–80°). As was discussed before, during that time of the day the direction of the airflow changed from westerly to a flow with an easterly component. The simulated change in wind direction near the surface was not always captured accurately by MM5, with a tendency to lag by several hours.

The results for the time period 10–30 June 2001 show characteristics similar to those for the 1998 period. A great improvement in the simulated T2m for the modified versus the standard MM5 version for most times of the day (Fig. 4) was achieved. The cold bias of the standard MM5 version was particularly large during afternoon and the evening until midnight (up to −5 K). With the modified MM5, the cold bias was reduced to biases of less than 2 K while the RMSE decreased from values of 4–6 K to values of 1–2 K. As for the 1998 experimental period, the largest bias of ∼2 K occurred for the time from 0400 to 0700 LST when using the modified MM5, the main reason being the timing of the simulated shift versus the measured shift in wind direction from westerly to flow with an easterly component and the lack of the associated cold advection. SDE was similar for both model versions.

The statistical measures of bias, RMSE, and SDE for the simulation results for surface wind speeds V10m and wind direction for 10–30 June 2001 are given in Fig. 4. The modifications applied to MM5 greatly enhanced the model’s performance for wind speed simulation during the daytime hours. The maximum slow bias of −3 m s−1 of the standard MM5 was reduced considerably. With daytime values of ∼1.5 m s−1 for the modified MM5, the RMSE was reduced from the ∼3 m s−1 of the standard MM5 version. During nighttime, the quality of the model simulations of the standard and modified MM5 versions differed only slightly. The models were in better agreement with the measurements during the first part of the night (bias ≈ 1 m s−1 and RMSE of 1–2 m s−1), than for the time from late afternoon to midnight for which a relatively large slow bias of up to −3 m s−1 and an RMSE of 2–3 m s−1 were calculated.

With values of ±50° for bias and 40°–80° for RMSE and SDE for most hours of the day, the results for the wind direction simulations were similar for both the modified and standard MM5. As for the 1998 period, the deviations between simulated and measured values were largest during early-morning hours (RMSE ≈ 100°) when the change of direction in airflow that results from the location of Phoenix in complex terrain occurred.

b. Potential temperature, specific humidity, and wind profiles

The measured and simulated vertical profiles of potential temperature θ and specific humidity qs were averaged for the hours during which radiosondes were launched for the periods of 10–31 May 1998 (7 launch days) and 1–10 June 1998 (7 launch days), resulting in composite soundings for the site near Sky Harbor Airport (Fig. 2; station 1) and the rural McDowell Mountain Park site (Fig. 2; site 2). We differentiated between the two time periods because the simulated vertical profiles of θ were generally in better agreement with the measurements for June than for May of 1998. The results are given in Figs. 5 and 6 for Sky Harbor Airport for the modified MM5 and the MM5 version using the original MRF in conjunction with 26-category land use and improved representation of the urban energy balance (called original MRF–modified LS). Also included were results for the standard MM5, to assess the successive effect of changes to the surface energy balance and land use and the PBL scheme on the simulation results.

For the 7 days of the June period (Fig. 6), the sequence of simulated composite soundings agreed to within ∼1 K with the measurements at all times and heights when using the modified MM5, whereas the simulated vertical profiles of potential temperatures for the 7 days in May 1998 were on most days much cooler within the PBL than were the measured profiles. Therefore the composite soundings for the May period exhibited deviations between measured and simulated values of θ of 2–3 K for all hours (Fig. 5). The deviations between the measured and simulated composites of θ near the surface were larger than the bias in T2m at the same hours in Fig. 3. The reason is that the data in Fig. 3 were averages of the 32 days of the 1998 period for which on many days a good agreement between measurements and simulations existed. In addition, there was a tendency for the measured T2m at the NWS station at Sky Harbor Airport to be lower than at the radiosonde launch site. For the radiosonde launch days in June, the “control” of the larger-scale weather as initialized and imposed through boundary conditions from the Eta model to MM5 was more accurately characterized than for most of the launch days in May. To illustrate characteristic differences in the simulations between the two time periods, θ profiles at Sky Harbor Airport as simulated by MM5 and NCEP Eta and as measured are shown in Figs. 7a and 7b for 1700 LST on 21 May 1998 and 1 June 1998, respectively. Also added to the plots were 1700 LST θ profile measurements from the nearest routine NWS radiosondes in Flagstaff and Tucson, Arizona, located 235 and 193 km away from the center of Phoenix at elevations of 2179 and 788 m, respectively. The 0500 and 1700 LST data of the latter sites were used in the Eta Model reanalysis runs. The MM5 simulations were in good agreement with the Eta Model output on both days and also agreed well with the measured θ profiles at Sky Harbor Airport for 1 June 1998 but not for 21 May 1998 on which day the simulated profiles were much cooler.

Despite the larger deviations between measured and simulated vertical profiles of θ for the seven days of the May 1998 period, the evolution of the convective PBL was reflected in the composite θ profiles obtained from the simulations with a similar quality as for the June period (Figs. 5 and 6). At around 0800 LST the measured vertical profiles of θ showed a shallow unstable layer and mixing to ∼600 m, with the modified MM5 simulating mixing to heights of ∼700 m. The original MRF–modified LS overpredicted mixing to heights of ∼1000 m, with θ values close to the modified MM5 in the lower PBL. Because of the higher amounts of available moisture for evaporation in the urban area, the simulated θ were cooler when using the standard MM5 and less mixing occurred than with the original MRF–modified LS model.

Mixing reached to about 1200 m at 1000 LST, and by 1200 LST the mixed layer had typically grown rapidly to about 2000 m. In the early afternoon (1400 LST) the measured vertical profiles of θ showed mixing up to ∼2300 m. At the three times, the results for the three model versions showed the same qualitative behavior, with a good agreement of mixing heights for the modified MM5, overestimation of mixing by the original MRF–modified LS by about 500 m, and comparatively lesser mixing when using the standard MM5. Late in the afternoon (1700 LST), the measured profiles of θ indicated further mixing to ∼2500 m that was captured well by the original MRF–modified LS whereas the simulations exhibited a decline in the mixing at that time of the day to heights of ∼2000 m when using the modified MM5.

The profiles of specific moisture in Figs. 5 and 6 confirm the results for vertical mixing as obtained from the θ profiles. However, the magnitude of qs was greatly overestimated by all model versions near the ground during the morning hours and was underestimated above a height of about 3000 m. The reason could be that the initial profiles from the Eta Model were too moist. Close to the surface, qs was higher for the modified MM5 than for the original MRF–modified LS, which can be attributed to mixing to higher layers in the atmosphere with the latter model version. Above the PBL the model behavior was reversed, with lower values for the modified MM5.

The radiosonde launch site near Sky Harbor Airport was located in the center of the city, and the data were therefore assumed to be representative for the convective PBL over the urban area (Fast et al. 2000). The results of the vertical profiles of θ for the rural McDowell Mountain Park site (Fig. 2; site 2) located northeast of Phoenix and adjacent to a low ridge were similar to those of the site near Sky Harbor Airport (data not shown). According to Fast et al. (2000), there was a tendency for the vertical potential temperatures to be ∼1 K warmer than those found at the Phoenix site in the afternoon, which was confirmed by the simulations.

The measured and simulated (for the modified MM5 and original MRF–modified LS) composite soundings (at 0500, 0600, 0800, 0900, and 1000 LST) for the period 10–30 June 2001 for the site near Sky Harbor Airport are given in Fig. 8. On 10 days, a strong stable stratification in the lowest 100–300 m of the atmosphere developed that persisted through 0600 LST followed by rapid deepening of the mixed layer after sunrise (Doran et al. 2003).

The simulated nighttime and early-morning composite vertical profiles of θ were about 1 K cooler than measured up to about 3000 m except in the first tens of meters near the ground (Fig. 8). Under conditions of very strong stability, the simulated near-surface temperature inversion at 0000, 0200, and 0500 LST was weaker than measured. This was true also for the Waddell site at 0500 LST (Fig. 2; site 3—results not shown). After sunrise, the simulated θ profiles near the surface improved (0800 LST) to a good agreement between simulated and measured data at 0900 and 1000 LST, though vertical mixing was overestimated by both versions of the MRF scheme and was more pronounced with the original MRF (Fig. 8). As for the 1998 simulations, the θ profiles obtained with the standard MM5 were much cooler than for the other two model versions (results not shown).

The absolute values of simulated composite vertical profiles of specific moisture deviated significantly from the measurements by 2–3 g kg−1 (Fig. 8). However, the shape of the profiles indicated the same qualitative model behavior as for the profiles of θ, that is, a more rapid of the breakup of the nocturnal inversion by the original MRF–modified LS scheme in comparison with the modified MM5.

The results from this study are in agreement with a comprehensive PBL scheme comparison by Zhong and Fast (2003), who showed that all of the tested schemes tended to produce weaker nocturnal inversions than were measured because of the parameterization of eddy diffusivities under stable conditions and subsequent overestimation of vertical mixing. This was also confirmed by Lee et al. (2006), who introduced to the MRF scheme a stability-dependent turbulent Prandtl number that allowed momentum to be transported by internal waves in the model, while heat diffusion was impeded by the stratification under stable conditions, thereby improving MM5’s performance for nocturnal near-surface air temperatures.

The reason for the overestimation of vertical mixing in the original MRF scheme is, in part, the formulation of the convective velocity, as discussed in section 2b. In addition, different PBL height diagnostics in the two versions of the MRF scheme led to much higher values for h for the original MRF–modified LS than for the modified MM5 with larger differences than are recognizable in the vertical profiles of θ. There is a positive feedback of h on the vertical mixing through the profile function used to calculate diffusivity coefficients (Hong and Pan 1996), which further intensifies vertical mixing in the original MRF.

The evolution of the composite measured and simulated PBL heights between 10 May and 10 June 1998 and between 10 and 30 June 2001 are shown in Figs. 9a and 10a, respectively, for Sky Harbor Airport, as well as for the rural McDowell Mountain Park (Fig. 9b) and Wadell site (Fig. 10b).

Under convective conditions the measured PBL heights were obtained by means of two methods: the bulk Richardson method (Seibert et al. 2000) and by subjectively evaluating the profile measurements of potential temperature. In the latter case, the measured PBL height was assumed to be at one-half of the depth of the elevated inversion or stable layer, as suggested by Stull (1988). Both methods yielded similar PBL heights, with the Rib method giving higher values for the late afternoon for days on which the elevated inversion was not very pronounced. When mechanical turbulence production was significant, two different regions of the vertical θ profile could be distinguished near the ground: a layer with a strong nearly linear increase of the potential temperature followed by a layer with a weaker decrease in θ as a result of exclusively radiative cooling. The PBL height was assumed to be at the usually sharp transition between the two layers (Seibert et al. 1998). The simulated PBL heights were calculated as described in section 2b.

Even though mixing was still overestimated with the modified MM5 in the early-morning hours, a much better agreement between simulated and measured PBL heights was achieved than with the original MRF–modified LS for the morning and early afternoon for both years and all sites. However, because of the diagnostic methods for the determination of PBL heights, the differences in h as simulated with the two MRF schemes were larger than the θ profiles indicated. The values for the standard MM5 fell between the results of the other two model versions (Figs. 9a,b). Because of the decrease in vertical mixing at around 1600 LST, the modified MM5 underestimated the PBL heights late in the afternoon at 1700 LST (Figs. 9a,b). The results for the rural sites as obtained with the two PBL schemes were in qualitative agreement with the launch site in the urban area.

For the 1998 period, the simulations gave, on average, slightly (∼100 m) higher maximum values of the simulated PBL height for the urban site in comparison with the rural McDowell Mountain site. According to the measurements and simulations of the 2001 period, the urban PBL at Sky Harbor Airport developed slightly faster between 0700 and 0900 LST in comparison with the rural Waddell site and was about 200 m higher from 1400 to 1600 LST.

The results are in agreement with Liu et al. (2006) for Oklahoma City, for which the PBL height calculated with the original MRF approach developed too early and became too deep during the daytime. The modified MRF scheme led to a consistent improvement in the simulation of the convective PBL at all times of the day except in the late afternoon, when vertical mixing was underpredicted.

To evaluate MM5’s performance for the observed wind speeds and flow pattern, the measured and simulated wind profiles for the 1998 and 2001 experimental periods were averaged for undisturbed days (weak synoptic conditions) for the site near Sky Harbor Airport to obtain composite wind profiles. The profiler data indicated that the typical flow pattern developed on 27 of the 33 days of the 1998 field experiment and on 13 days of the 16 days of the 2001 Phoenix Sunrise experiment. The westerly winds, when fully developed, had wind speeds of ∼5 m s−1 in the lowest 1–2 km of the atmosphere and reached the synoptic-scale flow. For the 2001 period, the predominantly westerly flow occurred on average earlier in the day (∼1200 LST) than during the 1998 experimental period (∼1700 LST). The easterly flow developed after midnight, extending to a maximum depth of about 1–1.5 km above ground, and was on most days more pronounced during the 2001 experimental period in comparison with the 1998 experimental period in terms of wind direction and wind speed. For both years there was a tendency for the simulated onset of the easterly and westerly flow to be delayed by several hours and to affect deeper layers in the atmosphere earlier than was measured, with no consistent differences between the results for the two versions of the MRF scheme.

The 1998 vertical profiles of measured and simulated (with three model versions: standard MM5, modified MM5, and original MRF–modified LS) composite wind speeds for 0800, 1000, 1200, 1400, and 1700 LST are given in Fig. 11. Sodar data were available for the 1998 experimental period that provided measurements of wind speed for the lowest ∼100 m in the atmosphere and were included as insets in the graphs of Fig. 11 for each hour. In the early-morning hours, winds deviated only slightly among the three model versions. For most daytime hours, the PBL winds were underpredicted by the standard MM5 except for a minimum in wind speed at about 700 m. The modifications to the surface energy balance and land use improved the representation of the PBL winds with both the modified MM5 and the original MRF–modified LS. However, as is recognizable from the comparison with the sodar data, the original MRF–modified LS model considerably underestimated near-surface winds for 1000, 1200, and 1400 LST (which was also true for other daytime hours; results not shown) whereas the agreement was good for the modified MM5. Those results confirm the simulations for surface winds as discussed in section 3a. There were no sodar data available for the 2001 period, but the behavior for PBL winds was qualitatively similar to that for the 1998 period (results not shown).

4. Conclusions

In this study, a version of MM5 was applied that included a refined land cover classification and updated land use/cover data for the arid Phoenix metropolitan area, as well as bulk approaches of characteristics of the urban surface energy balance, while PBL processes were simulated by a modified version of the MRF scheme. The results of the study show a consistently improved model performance for near-surface and upper-air temperatures and wind speeds when applying the modified MM5. In particular, a cold bias in T2m and θ in the standard MM5 during day and nighttime hours was reduced at all times of the day because of the improved land use characterization and physical representation of the urban energy balance. An analysis of the influence of land use and surface energy balance characteristics on the simulated T2m and V10m (Grossman-Clarke et al. 2005) showed that, during daytime hours, the T2m were mostly influenced by the moisture availability for evapotranspiration in the urban area and also by the surrounding desert, whereas nighttime temperatures were influenced highly by characteristics of the urban energy balance (anthropogenic heating, limited sky view, and heat storage). Those results were confirmed in this study by the vertical profiles of θ that showed a close agreement for the two MRF schemes in the lower PBL when surface characteristics were the same but deviations in the upper PBL that are due to the more vigorous mixing for the original MRF. The modified MRF scheme described vertical mixing more accurately than the original MRF scheme except during the late afternoon, when mixing was underpredicted.

Near-surface wind speeds were influenced mainly by the choice of the PBL scheme, with a much-improved agreement between simulated and observed V10m during daytime for the modified MRF in comparison with the original MRF scheme. The modifications to the surface characteristics improved the representation of the PBL winds for both versions of the MRF scheme, whereas the PBL winds were underpredicted by the standard MM5 during daytime.

There is a possibility that errors in surface flux simulations compensate or exaggerate deficits in the PBL schemes. Surface flux data were not available from the two field campaigns, and further analysis is necessary to evaluate the quality of the model simulations with respect to the components of the surface energy balance. As was shown in Grossman-Clarke et al. (2005) under the weak synoptic conditions that were dominant during the two field experiments in Phoenix, the surface T2m are strongly influenced by vertical mixing and therefore surface sensible heat fluxes. The improved simulations for T2m indirectly support the quality of the surface flux simulations. Furthermore the results of this study support results from previous studies by Zhong and Fast (2003), Zhang and Zheng (2004), and Liu et al. (2006).

The analysis of differences between the urban and rural PBL in the region was complicated by the influence of complex terrain on the airflow, and only slight differences between the simulated urban and rural PBL were detected. Therefore, to study the urban effect, we conducted additional case studies for which the urban thermodynamic and dynamic effects were removed from the simulations and were displaced by physical characteristics of the desert land use category. The results indicated that, because of the special urban properties, the urban effect on the PBL in the arid Phoenix metropolitan area differs from that found in cities in midlatitude environments. For the latter, the urban PBL is characterized by greatly enhanced mixing in comparison with the rural PBL, resulting from large surface roughness and increased surface heating (Fisher et al. 2005), whereas according to the simulations for Phoenix the urban PBL is slightly lower during afternoon hours than it would be for desert land use because of irrigated landscaping and subsequent enhanced evapotranspiration in the urban areas versus the natural desert areas. Mesoscale meteorological model simulations for cities in midlatitude environments showed that that the deeper and drier urban PBLs caused low-level convergence that is favorable for producing deep precipitating cumulus clouds downwind of the city (Cotton and Pielke 1995; Bornstein and Lin 2000; Rozoff et al. 2003; Burian and Shepherd 2005). Therefore, the Phoenix metropolitan area might potentially influence precipitation differently than cities located in humid midlatitude environments. Combining historical and remote sensing–based precipitation data, Diem and Brown (2003) and Shepherd (2006) showed a statistically significant increase in mean precipitation of 12%–14% from the preurban to the posturban period for locations in northeastern suburbs of Phoenix. The authors assume that in this region the outflow boundaries from storms can most effectively interact with the convergence and flux forcing on the edge of the city. To study mechanisms by which Phoenix might potentially influence monsoon convective activity using mesoscale meteorological models, it is necessary to resolve the evolution of the urban PBL and the resulting thermal mesoscale circulations that interact with the thunderstorm outflow propagating from the elevated terrain to the northeast and north (Bright and Mullen 2002). A comprehensively tested mesoscale model in terms of the evolution of the urban PBL is a significant step in enabling the investigation of those individual contributions of dynamic forcing mechanisms that potentially affect precipitation.

Acknowledgments

This material is based upon work supported by the National Science Foundation under Grant DEB-9714833, Central Arizona–Phoenix Long-Term Ecological Research (CAP LTER). Any opinions, findings, and conclusions or recommendation expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation (NSF). Partial financial support was also provided by the Salt River Project (SRP). The measurements used in this study were provided by Chris Doran (Pacific Northwest National Laboratory), Rich Coulter (Argonne National Laboratory), and Peter Hyde (Arizona Department of Environmental Quality). The 1998 and 2001 experiments were supported by the U.S. DOE Atmospheric Chemistry Program through the Office of Biological and Environmental Research.

REFERENCES

  • Anderson, J. R., , E. E. Hardy, , J. T. Roach, , and R. E. Witmer, 1976: A land use and land cover classification system for use with remote sensor data. U.S. Geological Survey Professional Paper 964, 36 pp.

  • Beljaars, A. C. M., 1995: The parameterization of surface fluxes in large-scale models under free convection. Quart. J. Roy. Meteor. Soc., 121 , 255270.

    • Search Google Scholar
    • Export Citation
  • Bornstein, R., , and Q. Lin, 2000: Urban heat islands and summertime convective thunderstorms in Atlanta: Three case studies. Atmos. Environ., 34 , 507516.

    • Search Google Scholar
    • Export Citation
  • Bright, D. R., , and S. L. Mullen, 2002: The sensitivity of the numerical simulation of the southwest monsoon boundary layer to the choice of PBL turbulence parameterization in MM5. Wea. Forecasting, 17 , 99114.

    • Search Google Scholar
    • Export Citation
  • Burian, S. J., , and J. M. Shepherd, 2005: Effect of urbanization on the diurnal rainfall pattern in Houston. Hydrol. Processes, 19 , 10891103.

    • Search Google Scholar
    • Export Citation
  • Cotton, W. R., , and R. A. Pielke, 1995: Human Impacts on Climate. Cambridge University Press, 288 pp.

  • Deardorff, J., 1972: Theoretical expression for the countergradient vertical heat flux. J. Geophys. Res., 77 , 59005904.

  • Diem, J. E., , and D. P. Brown, 2003: Anthropogenic impacts on summer precipitation in central Arizona, U.S.A. Prof. Geogr., 55 , 343355.

    • Search Google Scholar
    • Export Citation
  • Doran, J. C., , C. M. Berkowitz, , R. L. Coulter, , W. J. Shaw, , and C. W. Spicer, 2003: The 2001 Phoenix Sunrise experiment: Vertical mixing and chemistry during the morning transition in Phoenix. Atmos. Environ., 37 , 23652377.

    • Search Google Scholar
    • Export Citation
  • Dudhia, J., 1996: A multi-layer soil temperature model for MM5. Preprints. Sixth PSU/NCAR Mesoscale Model Users’ Workshop, Boulder, CO, National Center for Atmospheric Research, 49–50.

    • Search Google Scholar
    • Export Citation
  • Fast, J. D., , J. C. Doran, , W. J. Shaw, , R. L. Coulter, , and T. J. Martin, 2000: The evolution of the boundary layer and its effect on air chemistry in the Phoenix area. J. Geophys. Res., 105D , 2283322848.

    • Search Google Scholar
    • Export Citation
  • Fernando, H. J. S., , S. M. Lee, , J. Anderson, , M. Princevac, , E. Pardyjak, , and S. Grossman-Clarke, 2001: Urban fluid mechanics: Air circulation and contaminant dispersion in cities. Environ. Fluid Mech., 1 , 107164.

    • Search Google Scholar
    • Export Citation
  • Fisher, B. E. A., , S. Joffre, , J. Kukkonen, , M. Piringer, , M. Rotach, , and M. Schatzmann, 2005: Meteorology applied to urban air pollution problems. European Commission COST Action Final Rep. 715, 282 pp.

  • GP2100, 2003: Greater Phoenix regional atlas: A preview of the region’s 50-year future. Arizona State University, 70 pp. [Available online at http://www.asu.edu/copp/morrison/Atlas-5.pdf.].

  • Grell, A. G., , J. Dudhia, , and D. R. Stauffer, 1995: A description of the fifth-generation Penn State/NCAR Mesoscale Model (MM5). NCAR Tech. Note NCAR/TN-398+STR, 117 pp.

  • Grossman-Clarke, S., , J. A. Zehnder, , W. L. Stefanov, , Y. Liu, , and M. A. Zoldak, 2005: Urban modifications in a mesoscale meteorological model and the effects on near-surface variables in an arid metropolitan region. J. Appl. Meteor., 44 , 12811297.

    • Search Google Scholar
    • Export Citation
  • Guo, Y. R., , and S. Chen, 1994: Terrain and land use for the fifth-generation Penn State/NCAR Mesoscale Modeling System (MM5). NCAR Tech. Note NCAR/TN-397+IA, 114 pp.

  • Hong, S-Y., , and H-L. Pan, 1996: Nonlocal boundary layer vertical diffusion in a medium-range forecast model. Mon. Wea. Rev., 124 , 23222339.

    • Search Google Scholar
    • Export Citation
  • Lee, S-M., , W. Giori, , M. Princevac, , and H. J. S. Fernando, 2006: Implementation of a stable PBL turbulence parameterization for the mesoscale model MM5: Nocturnal flow in complex terrain. Bound.-Layer Meteor., 119 , 109134.

    • Search Google Scholar
    • Export Citation
  • Liu, Y., , F. Chen, , T. Warner, , and J. Basara, 2006: Verification of a mesoscale data-assimilation and forecasting system for the Oklahoma City area during the Joint Urban 2003 field project. J. Appl. Meteor. Climatol., 45 , 912929.

    • Search Google Scholar
    • Export Citation
  • Rife, D. L., , C. A. Davis, , Y. Liu, , and T. T. Warner, 2004: Predictability of low-level winds by mesoscale meteorological models. Mon. Wea. Rev., 132 , 25532569.

    • Search Google Scholar
    • Export Citation
  • Rozoff, C. M., , W. R. Cotton, , and J. O. Adegoke, 2003: Simulation of St. Louis, Missouri, land use impacts on thunderstorms. J. Appl. Meteor., 42 , 716738.

    • Search Google Scholar
    • Export Citation
  • Sailor, D. J., , and L. Lu, 2004: A top-down methodology for developing diurnal and seasonal anthropogenic heating profiles for urban areas. Atmos. Environ., 38 , 27372748.

    • Search Google Scholar
    • Export Citation
  • Seibert, P., , F. Beyrich, , S. E. Gryning, , S. Joffre, , A. Rasmussen, , and P. Tercier, 1998: Mixing height determination for dispersion modelling. Harmonisation of the pre-processing of meteorological data for atmospheric dispersion models, B. E. A. Fisher et al., Eds., European Commission COST Action Final Rep. 710, 121 pp. [Available online at http://www.wau.boku.ac.at/fileadmin/_/H81/H814/Downloads/cost710-2.pdf.].

  • Seibert, P., , F. Beyrich, , S. E. Gryning, , S. Joffre, , A. Rasmussen, , and P. Tercier, 2000: Review and intercomparsion of operational methods for the determination of the mixing height. Atmos. Environ., 34, 1001–1027.

    • Search Google Scholar
    • Export Citation
  • Shaw, W. J., , J. C. Doran, , and R. L. Coulter, 2005: Boundary-layer evolution over Phoenix, Arizona and the premature mixing of pollutants in the early morning. Atmos. Environ., 39 , 773786.

    • Search Google Scholar
    • Export Citation
  • Shepherd, J. M., 2006: Evidence of urban-induced precipitation variability in arid climate regimes. J. Arid Environ., 67 , 607628.

  • Stefanov, W. L., , M. S. Ramsey, , and P. R. Christensen, 2001: Monitoring urban land cover change: An expert system approach to land cover classification of semiarid to arid urban centers. Remote Sens. Environ., 77 , 173185.

    • Search Google Scholar
    • Export Citation
  • Stull, R. B., 1988: An Introduction to Boundary Layer Meteorology. Kluwer Academic Publishers, 665 pp.

  • Troen, I., , and L. Mahrt, 1986: A simple model of the atmospheric boundary layer: Sensitivity to surface evaporation. Bound.-Layer Meteor., 37 , 129148.

    • Search Google Scholar
    • Export Citation
  • Zhang, D-L., , and W-Z. Zheng, 2004: Diurnal cycles of surface winds and temperatures as simulated by five boundary layer parameterizations. J. Appl. Meteor., 43 , 157169.

    • Search Google Scholar
    • Export Citation
  • Zhong, S., , and J. D. Fast, 2003: An evaluation of the MM5, RAMS, and Meso Eta models at subkilometer resolution using VTMX field campaign data in the Salt Lake Valley. Mon. Wea. Rev., 131 , 13011322.

    • Search Google Scholar
    • Export Citation
Fig. 1.
Fig. 1.

Location of the four MM5 modeling domains as used in this study. Grid spacings are 54 km for domain 1 (3294 km east–west by 2700 km north–south), 18 km for domain 2 (1350 km east–west by 1080 km north–south), 6 km for domain 3 (594 km east–west by 414 km north–south), and 2 km for domain 4 (212 km east–west by 132 km north–south).

Citation: Journal of Applied Meteorology and Climatology 47, 3; 10.1175/2007JAMC1647.1

Fig. 2.
Fig. 2.

Land use and terrain (m) as produced by MM5’s TERRAIN preprocessor for the inner modeling domain (2 km × 2 km) based on Landsat Thematic Mapper 1998 image-derived land use/cover data and three urban land use/cover classes: urban built-up, urban mesic residential (irrigated), and urban xeric residential (arid, partly irrigated). Other land use/cover categories were the same as in the standard MM5 24-category USGS dataset. Sites labels for radar wind and radiosonde profilers are 1 for Sky Harbor Airport, 2 for McDowell Mountain Park, and 3 for Waddell. Coordinates are given as universal transverse Mercator zone 12 (UTM12) easting and northing.

Citation: Journal of Applied Meteorology and Climatology 47, 3; 10.1175/2007JAMC1647.1

Fig. 3.
Fig. 3.

Bias, RMSE, and SDE for 2-m air temperature, 10-m wind speed, and 10-m wind direction for each hour of the day for the 10 May–10 Jun 1998 simulation period for the NWS station at Sky Harbor Airport (Fig. 2; site 1). Plots represent standard MM5 (dashed line) and MM5 with modified MRF scheme (solid line) according to Liu et al. (2006), updated land use data, and urban energy balance scheme (Grossman-Clarke et al. 2005).

Citation: Journal of Applied Meteorology and Climatology 47, 3; 10.1175/2007JAMC1647.1

Fig. 4.
Fig. 4.

As in Fig. 3, but for the 10–30 Jun 2001 simulation period.

Citation: Journal of Applied Meteorology and Climatology 47, 3; 10.1175/2007JAMC1647.1

Fig. 5.
Fig. 5.

Measured and simulated composite profiles of (top) potential temperature (K) and (bottom) specific humidity (g kg−1) for the period of 10–31 May 1998 at Sky Harbor Airport (Fig. 2; site 1—7 radiosonde launch days for each period) at 0800, 1000, 1200, 1400, and 1700 LST. Plots represent observed (solid line), simulated using the modified MM5 (filled circles), simulated using the original MRF scheme with modified land use characteristics (solid triangles), and simulated using the standard MM5 (open triangles). Note that the vertical scale differs among panels.

Citation: Journal of Applied Meteorology and Climatology 47, 3; 10.1175/2007JAMC1647.1

Fig. 6.
Fig. 6.

As in Fig. 5, but for the period of 1–10 Jun 1998.

Citation: Journal of Applied Meteorology and Climatology 47, 3; 10.1175/2007JAMC1647.1

Fig. 7.
Fig. 7.

Measured and simulated profiles of potential temperature at 1700 LST for (a) 21 May and (b) 1 Jun 1998. Plots represent measured at Sky Harbor Airport (Fig. 2; site 1) (solid line), routine NWS radiosonde measurement at Flagstaff (solid circles), routine NWS radiosonde measurement at Tucson (open triangles), NCEP Eta Model simulated (plus signs), and MM5 simulated using the modified MRF scheme by Liu et al. (2006) (open circles).

Citation: Journal of Applied Meteorology and Climatology 47, 3; 10.1175/2007JAMC1647.1

Fig. 8.
Fig. 8.

As in Fig. 5, but for the period of 10–30 Jun 2001 (12 radiosonde launch days) for 0500, 0600, 0800, 0900, and 1000 LST. Plots represent observed (solid line), simulated using the modified MM5 (filled circles), and simulated using the original MRF scheme with modified land use characteristics (solid triangles).

Citation: Journal of Applied Meteorology and Climatology 47, 3; 10.1175/2007JAMC1647.1

Fig. 9.
Fig. 9.

Measured and simulated composite PBL heights above ground level for the period of 10 May–10 Jun 1998 at (a) Sky Harbor Airport and (b) McDowell Mountain Park (Fig. 2; sites 1 and 2, respectively). Plots represent measured (times signs), simulated using the modified MM5 (solid circles), simulated using the original MRF scheme with modified urban land use/cover and urban energy balance (solid triangles), and simulated using the standard MM5 (open triangles).

Citation: Journal of Applied Meteorology and Climatology 47, 3; 10.1175/2007JAMC1647.1

Fig. 10.
Fig. 10.

As in Fig. 9, but for the period of 10–30 Jun 2001 at (a) Sky Harbor Airport and (b) Waddell (Fig. 2; sites 1 and 3, respectively). Plots represent measured (times signs), simulated using the modified MM5 (solid circles), and simulated using the original MRF scheme with modified urban land use/cover and urban energy balance (solid triangles).

Citation: Journal of Applied Meteorology and Climatology 47, 3; 10.1175/2007JAMC1647.1

Fig. 11.
Fig. 11.

Vertical composite wind profiles for the period of 10 May–10 Jun 1998 near Sky Harbor Airport (Fig. 2; site 1). For clarity, the data for the lowest 300 m of the atmosphere are given as insets with a higher resolution. Plots represent measured (times signs), simulated using the modified MM5 (solid circles), simulated using the original MRF scheme with modified urban land use/cover and urban energy balance (solid triangles), and simulated using the standard MM5 (open triangles).

Citation: Journal of Applied Meteorology and Climatology 47, 3; 10.1175/2007JAMC1647.1

Save