Simulations of the Urban Boundary Layer in Phoenix, Arizona Susanne Grossman-Clarke Arizona State University Global Institute of Sustainability 17 January.

Simulations of the Urban Boundary Layer in Phoenix,

Arizona

Susanne Grossman-ClarkeArizona State University

Global Institute of Sustainability

17 January 2008

Yubao LiuNCAR, Research Applications Laboratory

Joseph A. Zehnder Creighton University

Introduction

• Extent of Phoenix ~ 4000 km2 & population 3.7 Million: Potentially large enough to influence mesoscale meteorological processes.

• Investigation of Phoenix’ influence on weather (NSF ATM-0710631 and NSF DEB-0423704 CAP LTER):

• Wind, temperature and moisture fields.

• Mesoscale circulations generated by urban–rural thermal differences.

• Orographic circulations.

• Convective activity.

• Applications

• Weather forecasting, air quality simulations, urban heat island, human comfort and heat related illness studies.

Introduction

• Urban roughness.• Increased heat storage

and heat conductivity in built materials.

• Anthropogenic heating (electricity consumption and traffic).

• Long-wave radiation trapping due to urban form.

• Urban vegetation.

Physical characteristics of cities affect momentum, turbulent heat transport & surface energy balance.

Introduction

Urban Canopy Models (UCM)

Describe area average effect of cities on drag, turbulence production, heating, and surface

energy balance.

State variables within the urban canopy are of interest.

Differences in physical approach and detail.Roughness and drag approach.

Application of UCM depends on PBL scheme.

Introduction

• MM5 modifications (bulk roughness scheme)– Urban land use.– Surface energy balance.– Turbulent transport.– Medium Range Forecast PBL scheme.

• Testing of original and modified MM5– Surface and upper air data from two

extended field campaigns in Phoenix.

Grossman-Clarke et al. 2005, JAMGrossman-Clarke et al. 2007, JAMC

Urban Land Use derived from ASTER Satellite Data

Stefanov et al. 2001, Remote Sens. Environ.

MM5 Model Description

• Based on Landsat Thematic Mapper satellite images (visible, shortwave infrared & vegetation index).

• Post-classification in expert system using additional data sets.

• Derive land cover data with 30 m resolution.

Stefanov et al. 2001, Remote Sens. Environ.


• Convert data for use in MM5 or WRF.• Re-projecting data to the geographic

projection parameters of 30-second USGS data set.

• Mapping categories to 24 USGS categories.

• Land cover class with highest fraction of cover assigned to 30sec grid cell.


• Additional urban land use/cover classes:– urban built-up (no vegetation)– mesic residential (well-watered)– xeric residential (drought-

adapted vegetation)


24-category USGS classification and two additional urban classes.


Standard land use Improved


Surface Energy Balance Equation

)( gTfEGHnRtgT

gC

Tg … Ground temperature (K)

cg … Heat capacity of the ground (J m-2 K-1)

Rn … Net radiation balance (W m-2)

H … Sensible heat flux (W m-2)

G … Soil heat flux (W m-2)

E … Latent heat flux (W m-2)


Latent Heat Flux

)/(ln

)(

0Lz

z

z

qTqkuME

aha

vagvsa

M … Moisture availability factor [-]

z0 … Roughness length [m]

Yh … Stability function [-]

qvs … Saturation specific humidity [-]

qva … Specific humidity at za[-]


• Heat storage in man-made materials; modified heat capacity and thermal conductivity (Liu et al. 2004).

• Sky view factor in the in the long wave radiation balance (Noilhan 1981):

whwhroad /1/5.02

w – Road width

h – Building height

4ggskylong TLR


• Anthropogenic heating Qa from traffic and electricity consumption (Sailor & Lu 2004).

3600/, EVDVDhFhhQ ctipop

iva

3600/, ceipop

iea EhFhhQ

Qa,v Qa,e Anthropogenic heat from traffic and electricity

pop Avg. population density for urban LU classes

h Hour of day

Ft Fe Fractional traffic profiles and electricity consumption

DVDc Avg. daily vehicle distance traveled per person in Phoenix

EV Energy release per vehicle per meter of travel

Ec Daily per capita electricity consumption


0

5

10

15

20

25

30

35

40

0 2 4 6 8 10 12 14 16 18 20 22

Hour (LST)

Qa (

W m

-2)

□ Urban built-up

Xeric residential

x Mesic residential

Anthropogenic Heat


z

Q

cD

T

c

Qgwp

t

p

cT

t

T a

paT

ppa

11

0

00

''

VV

Temperature tendency equation at first prognostic level:

Q Heating rate resulting from diabatic processes

Dt Horizontal and vertical diffusion

() Adiabatic warming

Surface Parameters

Urban built-up

Urban mesicresidential

Urban xericresidential

Fraction vegetative cover 0 0.23 0.1

Moisture availability 0.005 0.12 0.02

Roughness length (m) 0.8 0.5 0.5

Heat capacity (106 J m-3 K-1)

3.0 2.4 2.7

Thermal conductivity (W m-1 K-1)

3.24 2.4 2.6

Sky view factor 0.85 0.85 0.85

MRF Scheme Nonlocal-K Approach for Turbulent Diffusion within

the Mixed Layer

Hong and Pan 1996, Monthly Weather ReviewTroen and Mahrt 1986, Boundary Layer Met.

Turbulence diffusion equation for potential temperature within the mixed-layer:

z

Kzt z

z

wb

''

… Potential temperature (K)w … Vertical velocity (ms-1)K … Eddy diffusivity (m2s-1)b … Empirical parameter (-)

Correction to local gradient to represent large eddy turbulence:

MRF SchemeMixed Layer Turbulent Diffusivity Coefficients

Kzm for Momentum

1 ms uw

k ... von Karman constant (-)ws ... Mixed layer velocity scale (ms-1)z ... Height (m)h ... PBL height (m)p ... Profile shape exponent (p=2)u* ... Friction velocity (ms-1)m ... Wind profile function at top of surface layer (-)Cm ... Drag coefficient for momentum (-)Uc ... Horizontal wind speed under convective conditions(ms-1)

p

szm h

zzkwK

1

22* cmUCu

with

MRF Scheme PBL Height for Mixed Layer

gv

vacr hg

hURibh

2

Ribcr … Critical bulk Richardson number (0.5)va … Virtual potential temperature at first prognostic levelv … Virtual potential temperature at z=hg … Virtual potential temperature at ground level z=0U(h) … Wind speed at z=h

MRF Scheme

5.0* vavgCw

agChUCw ''

Ch ... Drag coefficient for heat (-)g ... Potential temperature at ground (K)a ... Potential temperature at first prognostic level (K)Uc ... Wind speed under convective conditions (ms-1)U ... Mean horizontal wind speed (ms-1)w* ... Convective velocity (ms-1)C, ... Empirical constantsvg ... Virtual potential temperature at ground (K)va ... Virtual potential temperature at first prognostic level (K)

Under convective conditions w* is added to U in surface flux calculations to consider extra eddy mixing induced by surface-layer instability:

2*

22 wUUc with

MRF Scheme

Zhang and Zheng 2004, JAMLiu et al. 2006, JAM

• Underestimate near-surface wind speed.

• Overestimate sensible heat fluxes.

• Overestimate PBL heights.

Because:

• w* function of height of the lowest prognostic level.

• Virtual surface temperature depends on the choice of surface model.

• High values of w* result in overestimation u* weak surface winds, high surface sensible heat fluxes, high PBL heights.

Under free-convection conditions, tendency to:

MRF SchemeBeljaars’ Approach for Convective Velocity

3/1''

*

vw

ghw

Beljaars 1995, Quart. J. Roy. Meteor. Soc.

Liu et al. 2006, JAM

• w* directly linked to surface heat flux and PBL height. Both related to strength of convective turbulence.

• No tuning parameter.

• Parameter in Uc calibrated with LES (0.8 – 1.3).

Comparison of Model Behavior with Field Observations

• Meteorological and atmospheric chemistry field study in Phoenix 10 May to 10 June of 1998 (Fast et al. 2000):

– To study the convective boundary layer.

• 915-MHz radar wind profiler near Sky Harbor Airport to give hourly values of wind speed and wind direction.

• Radiosondes near Sky Harbor Airport on 14 days at 0800, 1000, 1200, 1400, 1700 LST.

• “Phoenix Sunrise Experiment” 10 – 30 June 2001 (Doran et al. 2003):

– To study the evolving structure of the PBL during the morning transition.

• 915-MHz radar wind profiler near Sky Harbor Airport.

• Radiosondes near Sky Harbor Airport site on 12 days at 0000, 0200, 0500, 0800, 0900 and 1000 LST.

Design of Numerical Simulations

• Fifth Generation PSU/NCAR Mesoscale Model (MM5).

• Initial and Boundary Conditions from NCEP/ETA grid 212 (40 km grid spacing).

• 10 May – 10 June 1998 & 10 – 30 June 2001.

• Nested Run of MM5: 54 km 18 km 6 km 2 km.

• 51 vertical layers.• Original and modified MRF PBL

scheme (Liu et al. 2006) and 5 layer soil model.

• Urban surface energy balance (Grossman-Clarke et al. 2005).

Comparison of MM5 Simulations with Field Observations

• Original MM5.

• Original MRF scheme and surface modifications.

• Modified MRF scheme and surface modifications.

Comparison of MM5 Simulations with Field Observations

8 June 1998 at Sky Harbor Airport

Correcting land use improvesdaytime temperatures.

Heat storage, anthropogenic heat, sky view factor improves nighttime temperatures.

Results - Surface Temperature and Winds

• 10 May to 10 June 1998 simulation period for NWS station at Sky Harbor Airport.

Results - Surface Temperature and Winds

• 10 June to 30 June 2001 simulation period for NWS station at Sky Harbor Airport.

Composite Profiles of Potential Temperature (K)


Results – Potential Temperature



Results – Composite PBL Heights

Composite Winds (m s-1) 10 to 31 May 1998

Conclusions

• Bulk approaches for the urban surface energy balance enabled MM5 to consistently improve performance for near-surface meteorological variables.

• Modified MRF PBL scheme by Liu et al. (2006) led to improved:

− Profiles of potential temperature.

− PBL height determination

− Wind speed in the lower PBL

• MM5 can be applied in studies investigating the influence of urbanization on weather with higher confidence.

Work with WRF

• Ported model physics into WRF; results are applicable to WRF YSU scheme; UCM and LSM vegetation parameterization for Phoenix based on gas exchange measurements.

• Investigate the combined influence of global climate change and urbanization on near-surface air temperatures on human comfort and health (NSF Coupled Human Natural Systems Proposal).

• Investigate the influence of urbanization on weather in Phoenix (NSF ATM-0710631) – Co-PIs C.S.B. Grimmond, King’s College London & J.A. Zehnder, Creighton University in collaboration with F. Chen, National Center for Atmospheric Research

WRF Urban Canopy Model

za First prognostic level.

Ta, TS Air temperature at first prognostic level and street canyon.

TR, TW, TG Surface temperatures of roof, wall, ground.

H, HR, HW, HG, Ha Sensible heat fluxes.

• Consideration of more detailed characteristics of the urban surface (construction materials and urban form) and urban vegetation processes possible.

Work with WRF

Surface Energy Balance Tower

Part I - Evaluate the UCM simulated surface energy fluxes with:

• Comprehensive meteorological and energy flux data obtained from previous urban field experiments.

• Surface energy balance measurements in at least two typical Phoenix neighborhoods during a one year period beginning in summer 2008.

Part II - Apply the WRF/UCM system to the Phoenix metro area to investigate:

• How past and potential future land use changes influence near surface atmospheric state variables and characteristics of the planetary boundary layer?

• How mesoscale circulations due to the variability in urban and rural land use interact with the mesoscale thermal circulations due to complex terrain?

• If the increasing extent of the urban area affects the development and propagation of summer thunderstorms.

Potential Effects of Phoenix on Monsoon Convective Activity

• Increased surface roughness suppresses thunderstorm outflow and inhibits propagation into the region.

• Urban heat island effect.

• Surface roughness causes divergence of air flow around the urban area and convergence zone downwind.

• Evapotranspiration from irrigated vegetation and anthropogenic open water surfaces increases CAPE.

• Interaction of topograhically and physiographically forced circulations.

• Pollution aerosols.

Influence of Urbanization on Near-surface Air Temperature

• Model performance during extreme heat events for past and projected future land use/cover in the Phoenix metropolitan area.

• 12 - 17 July 2003 and 9 - 12 August 2003.

• Simulations of surface temperature and relative humidity.

WRF – Simulations for Heat Waves 12-17 July 2003 & 9-12 August 2003

Sky Harbor Airport, 12-17 July 2003

0

10

20

30

40

50

12 12 13 13 14 14 15 15 16 16 17 17

Date July 2003

Rel

ativ

e H

um

idit

y (%

)

Simulated Measured

Sky Harbor Airport, 12-17 July 2003

85

90

95

100

105

110

115

120

12 12 13 13 14 14 15 15 16 16 17 17

Date July 2003

2 m

air

tem

per

atu

re (

F)

Simulated MeasuredSky Harbor Airport, 8-12 August 2003

85

90

95

100

105

110

115

120

9 9 10 10 11 11 12 12

Date August 2003

2 m

air

tem

per

atu

re (

F)

Measured Simulated

Sky Harbor Airport, 9-12 August 2003

0

10

20

30

40

50

9 9 10 10 11 11 12 12

Date August 2003

Rel

ativ

e H

um

idit

y (%

)

Measured Simulated

Land Use/Cover Scenarios

MM5 – Simulated 2 m Air Temperatures 14 July 2003 0500 pm

MM5 – Simulated 2 m Air Temperatures 15 July 2003 0500 am

Simulated 2 m Air Temperatures 14 July 2003

Simulations of the Urban Boundary Layer in Phoenix, Arizona Susanne Grossman-Clarke Arizona State University Global Institute of Sustainability 17 January.

Documents

urban heat island

heat conductivity

urban vegetation

model descriptionconvert

lelatent heat flux w

hsensible heat flux

gsoil heat flux w

land cover data