neighborhood scale modeling - us epa · neighborhood scale modeling develop air quality grid models...

Neighborhood Scale Modeling

Tanya L. Otte1, Avraham Lacser2, Jason Ching1, and Sylvain Dupont1

1NOAA/ARL & U.S. EPA/NERL, RTP, NC2Israel Institute for Biological Research,

Ness Ziona, Israel

Objectives for Neighborhood Scale ModelingDevelop air quality grid models for risk assessments at neighborhood scales

Develop urban canopy parameterizationsfor air quality grid modeling with CMAQ

Create sub-grid pollutant concentration variability fields using PDFs

Link to human population exposure models

Design Issues

Scale-dependent model parameterizations in momentum, TKE, surface energy balance, vertical mixing for air quality simulations

Many area, point, and line sources within urban canopy

Near-source transport, and horizontal and vertical dispersion constrained by urban canopy characteristics

Why use a UCP?

Want to improve urban simulations for ~1 kmParameterized roughness may not adequately account for heterogeneities in urban areasUCP allows for more specific treatment of urban contributions to dynamics and thermodynamics

Ultimately want to run CMAQ to simulate photochemical pollutant species on that scale

About the UCP

Based on Brown and Williams “drag approach”Applied in 1.3-km MM5 simulationsDirectly impacts grid cells with non-zero urbanDrag and TKE effects due to urban structuresAnthropogenic heat as time-varying functionExtinction of radiation in city canyonsRoof top contribution under development

Implementing the UCP in MM5

40 layers: 12 in lowest 100 mTypical setup has 30 layers, 3 in lowest 100 m

Updates U, V, TKE in Gayno-Seaman PBLEnergy modifications in RRTM, solve, slabUses fractional land use categoriesAdded new “urban zones” definitions

Pseudo-Morphology for Philadelphia

1

23

4

5

6

7

Preliminary Results

Four experiments here…Initialized 00 UTC 14 July 1995 from 4-km, 30-layer

“nocan30”: 30-layer, no UCP“nocan40”: 40-layer, no UCP“sens7”: 7 urban categories (e.g., morphology)“sens8”: 1 urban category (wgt’d avg morphology)

Surface Observations

NXX

TTN

WRI

ACY

PNE

PHL

MIV

ILG

NEL

PHL Temperature — 14 July 1995

20

25

30

35

40

0 4 8 12 16 20 24

T OBST nocan30T nocan40T sens7T sens8

T (°

C)

time (h)

MIV Temperature — 14 July 1995

20

25

30

35

0 4 8 12 16 20 24

T OBST nocan30T nocan40T sens7T sens8

T (°

C)

time (h)

RMSE — Temperature

0.51

1.52

2.53

3.54

0 4 8 12 16 20 24

nocan30nocan40sens7sens8

RM

SE (°

C)

time (h)

nocan30: 2.17nocan40: 2.41

sens8: 1.93sens7: 2.01

Index of Agreement — Temperature

0

0.2

0.4

0.6

0.8

1

0 4 8 12 16 20 24

nocan30nocan40sens7sens8

Inde

x of

Agr

eem

ent

time (h)

nocan30: 0.50nocan40: 0.48

sens8: 0.58sens7: 0.57

RMSE — Moisture

11.5

22.5

33.5

4

0 4 8 12 16 20 24

nocan30nocan40sens7sens8

RM

SE (g

/kg)

time (h)

nocan40: 2.16 nocan30: 2.58sens7: 2.11 sens8: 2.21

WRI Wind Speed — 14 July 1995

1234567

0 4 8 12 16 20 24

S OBSS nocan30S nocan40S sens7S sens8

win

d sp

eed

(m/s

)

time (h)

WRI Wind Direction — 14 July 1995

180190200210220230240250260

0 4 8 12 16 20 24

D OBSD nocan30D nocan40D sens7D sens8

win

d di

rect

ion

(deg

)

time (h)

RMSE — Wind Direction

0102030405060

0 4 8 12 16 20 24

nocan30nocan40sens7sens8

RM

SE (d

eg)

time (h)

All Expts: 24.3 - 24.8

RMSE — Wind Speed

0.81

1.21.41.61.8

22.22.4

0 4 8 12 16 20 24

nocan30nocan40sens7sens8

RM

SE (m

/s)

time (h)

nocan30: 1.39nocan40: 1.52sens7: 1.32

sens8: 1.56

Bias — Wind Speed

-2-1.5

-1-0.5

00.5

11.5

2

0 4 8 12 16 20 24

nocan30nocan40sens7sens8

Bia

s

time (h)

nocan30: +0.75nocan40: -0.91sens7: -0.31

sens8: -0.96

Summary of UCP

UCP at 1.3-km tends to produce desired effectsChanges to wind, TKE, temperature

UCP tends to be superior to 40-layer run w/o UCP and 30-layer run w/o UCP (standard set-up)UCP with morphology tends to be superior to UCP with homogeneous city…esp. wind speed

…must evaluate more days before concluding!

Advanced UCP

Extend drag approach to all roughness elements inside canopy (buildings and vegetation)

Couple drag approach to urban soil model (SM2-U from French SUB-MESO)

Use actual urban morphology database

Advanced UCP ConsiderationsMomentum sources:

Horizontal and vertical surfaces of buildingsVegetation

TKE sources:Horizontal and vertical surfaces of buildingsVegetationSensible heat fluxes

Heat and humidity sources:Buildings and vegetative surfacesAnthropogenic contributions

Urban Canopy Parameters

Mean and std dev of building, vegetation heightRoof and vegetation area densityBuilding and vegetation plan area densityFrontal area density of buildingsSurface area of wallsPlan area fraction of vegetation, roof tops, water, and paved surfaces

Urban Canopy Parameters (cont.)Ratio of building height to widthMean orientation of streetsSky view factorRoughness lengthDisplacement heightMaterial type of building surfacesPercent impervious area directly connected to draining system

Obtaining Urban Morphology from Lidar Mapping

Very high data density: 100,000’s points/km2

High accuracy: 15-30 cm RMSE in open areasFlexible: independent of sun angle, cloud cover“Multi-Return” allows mapping in canopy gapsRapid collection of elevation data

Typical Flight with LIDAR

�

30 km radiusGPS Ground Station

915 m AGL

210-240 kph

Swath width = 625 m

10-30% overlap

3 m spacing111,000 points/sq.km.

Multiple Returns

1st return

2nd return

3rd return

4th return

�Source laser beam

Raster Building Data Set

Defining Morphology for MM5Where full data coverage…

GIS processing of raster and vector data

Distinguish roof tops from vegetation canopy

In data-void areas…Extrapolate based on co-relationships between building histogram and land use

Next StepsComplete development and evaluation of UCP for Philadelphia for additional daysEvaluate 1.3-km CMAQ…modifications?Continue development of advanced UCP and apply to Houston with morphology databaseExplore sub-grid variability with PDFsExplore linkages to human population exposure models