Conceptual database system for urban model development &

applications

Jason ChingARL/NOAA –NERL/USEPA

Research Triangle Park, NC ching.jason@epa.gov

COST 728 WorkshopExeter, England May 3,4, 2007

Presentation outlineCAVEAT: mostly generic but contains some USA perspectives

• Background: “Setting the stage”

• Rationale for a “database” focus

• Database content/concepts• Prototype implementation of “NUDAPT”

(National Urban Database and Access Portal Tools)

Guiding Principles

• Each model application is unique. • The scale resolution of the modeling must be

appropriate to the task (Use the right tools!) • Current and future urban applications may

require tools/techniques not yet available or under development; e.g., urban to local scale

• Hierarchical (nesting or adaptable gridding) approaches usually necessary.

• Appropriate data needed for modeling - both development and operational.

Addressing societal issues with urban modeling focus

• Air Quality –Health – Exposure assessments– Policy and Controls– Acute to chronic time scales

• Homeland security – Transport on episodic bases

• Urban impact on climate change– Growth – Urban heat island and its mitigation

Multi-scale AQ field (CMAQ) NOx (36km, 12km 4km and 1.3 km grid sizes)

Multiscale Ozone (see previous slide)

Background: The “Problem”• To predict and characterize transport and concentration fields in urban

areas – Models must be commensurate with scale of the transport and

concentration gradients- Scale hierarchy– Current situation: Urban modeling too coarse a scale, boundary layer closure

schemes, model descriptions were overly coarse and simplistic for real cities– Urban land use, land classification schemes limited, also overly simplistic– Urban complexities such as street canyons control exposures in urban areas– Urban areas evolve, some very rapidly

• Advanced modeling tools are needed for future urban application requirements; a myriad of problems remains unsatisfactorily addressable without adequate and or appropriate tools.

– Data sets are not homogeneous (geospatial, activity, population, …)– Variety of model urbanization approaches have recently been

implemented into mesoscale models. Testing, evaluation will lead to improved model description of physics and thermodynamics and model engineering and reutilization of database.

– Operational urban models need to balance the overly simplistic to highly sophisticated parameterizations of urban features.

• Situation provides intriguing opportunities!

METHODOLOGYMETHODOLOGY: Meso-urban scale modeling: Meso-urban scale modeling

Modeler Needs: Modeler Needs: To capture the area-To capture the area-average effect of the average effect of the urban area in mesoscale urban area in mesoscale atmospheric modelsatmospheric models

Solution: Solution: Modelers have Modelers have implemented urban implemented urban canopy parameterizations canopy parameterizations into their models (e.g., into their models (e.g., MM5, WRF, HOTMAC, MM5, WRF, HOTMAC, RAMS, COAMPS…)RAMS, COAMPS…)

turbulence production

drag

radiationtrapping

radiation attenuation

canopy heating &cooling

urban thermalproperties

anthropogenicheating

Urban Canopy Effects

Salt Lake City, UT (Don Green Photography)

9

Rural Urban

Rural

Roughness Sub-Layer

Neighborhood scale

Local scale

1 km.

Meso scale

NEIGHBORHOOD SCALESUrban canopy details can not be represented: Parameterize the urban surface effects. Majority of pollutants emitted inside roughness sub-layer: Necessitates good precision on meteorological fields.Ground conditions in mesoscale model not satisfactory at neighborhood scale: apply drag-force and land use features at urban scales

Mesoscale: Model produces single meteorology profile applicable to grid cell Results influenced by the presence and aggregated effects of buildings.

Building scale: Intra-cell flow fields will be highly variable (horizontally and vertically), influenced by the individual buildings.

Buildings distributed in 1 km grid.

ISSUE: Relating meso-urban to building scale

T int

Q wall

Ts roof

Drainage outside the system

Sensible heat flux

Latent heat flux

Net radiation

Storage heat flux

Anthropogenic heat flux

Precipitation

Roughness approach

Root zone layer

Infiltration

Diffusion

Deep soil layer

Drainage

Drainage network

natural soil

roof

water

Paved surface

bare soil

Surface layer

Drag-Force approach

Rn pav Hsens pav LEpav

Gs pav Ts pav

An implementation: DA-SM2U in MM5 (Gayno-Seaman sub-system)

o Urbanization introduced at grid sizes of ~1km using drag approach (DA) o Land surface model (SM2-U) o Additional, within canopy layers

Wind tunnel and urban experiments provide guidance

(Kastner-Klein and Rotach, 2001)

Urban canopy parameterization 13

1

z

1

z

z

1

z

1

z

Roof area

density

Vegetation area

density

Building plan area density

Vegetation plan area density

Building frontal area

density

The knowledge of the vertical and horizontal distribution of the different urban land cover modes is necessary.

Introduction of canopy concepts and urban morphology parameters make possible improved modeling

0

50

100

150

200

250

300

0.000 0.001 0.002 0.003 0.004 0.005 0.006 0.007 0.008 0.009

af(z)

He

igh

t (m

)

North

West

Northwest

Frontal Area Index as a Frontal Area Index as a Function of HeightFunction of Height

We have technology and means for obtaining building data at high resolution; such data and ancillary data are becoming increasingly moreavailable for our major cities

High resolution urban morphological data from lidar mapping and photogrammetric techniques

ALTMS Normal Operating Parameters

30 km radius

GPS Ground Station

915m AGL

210-240kph

Swath width = 625m

10-30% overlap

3 m spacing111,000 points/sq.km.

Profiling

* Record Longest Return* Normally Rotary Wing* Continuous Ground Coverage

CANOPY UCPs BUILDING UCPs

VEGETATION, OTHER UCPs

Mean vegetation height

Mean canopy height Mean Height Vegetation plan area density

Canopy plan area density Std Dev of heights Vegetation top area density

Canopy top area density Height histogram Vegetation frontal area density

Canopy frontal area density Wall-to Plan area ratio

Roughness Length* Height to width ratio Mean Orientation of Streets

Displacement height* Plan area density Plan area fraction surface covers

Sky View Factor Rooftop area density % connected impervious areas

Frontal area density Building material fraction

Gridded (1 km) Urban Canopy Parameters (UCP) from high resolution data for urbanized MM5

*Parameters used in RA formulations Height dependent UCP

Selected Urban Canopy Parameters per 1 km2 cells for Harris County, TXNOTE! Each grid cell has unique combination of UCPs

MM5 Sensible Heat Flux (w/UCP) MM5 PBL w/UCP

MM5 Sensible Heat Flux (RA) MM5 PBL (RA)

Sensitivity study: Comparison of results using DA-SM2U (UCP version) Standard MM5 (RA)

Air Quality Model (CMAQ) at fine scales

• Pollutant model simulations are sensitive to(and dependent on) grid resolution

• AQ simulations depend on outputs of meteorological models which in turn depend on model descriptions of physics and thermodynamics.

Ozone (1 km gridded CMAQ simulations) @ 2100 GMT

UCP noUCP Difference (UCP-noUCP)

• Significant differences in the spatial patterns shown between UCP and noUCP runs (titration effect occurs in both sets)

• Flow, thermodynamics & turbulent fields differ between the UCP and noUCP simulations & contribute to differences

Fundamental urban model engineering design requirements

• Urban morphological structures:– Form and pressure drag from obstacles– Reflective, radiative and thermodynamic properties of

buildings, roofs, paved surface areas, street canyons• Land surfaces:

– Soils– Vegetative canopy– Degree of imperviousness to moisture– Surface propertiesThermal resistivities, ground storage

• Land cover classes:– Apt model descriptors and classifications– Grid coverage: Dominant vs fractional area methodology

• Anthropogenic heating

Database Concepts

• Bases for implementing descriptive parameterizations at appropriate scales– High resolution building and other urban morphological features– Sets of urban canopy parameterizations for various advanced meteorological

modeling– Tools to generate UCPs for generalized gridding and reference systems– Community based, flexible and encouraging of collaborative studies to improving

urban scale modeling and facilitating their scientific acceptance– Supports hierarchical (nesting) approaches from mesoscale (regional) to urban

scale to applications requiring CFD type approaches– Provide for evaluation- at appropriate scales

• Facilitates advanced, scale dependent applications – Transboundary to regional to urban to neighborhood– Forecast WX and air quality in urban areas– Air quality (Transboundary-regional-urban–to local)– Dispersion (vectors to agents) – Exposure (personal, population, air pollutants, agents)– Urban planning (mitigating intensity of heat islands)

Prototypic ImplementationThe “NUDAPT” Framework

• Urban modeling is its major focus• Adopts a community system paradigm-

– Encourages collaborations, accelerates model advancements with Portal technology

– Supports various meteorological modeling systems, others are possible– Broad user base (Model developers to users) – Extensible (to smaller scales, to current and future city structures, to revised sets

of UCPs) • Database consists of primary and derived parameters

– High resolution geospatial data: repository or links (133 cities in USA)– Appropriate and complete set of parameterizations at urban grid scale– Ancillary data (to facilitate applications)– Allowance for evaluation, operational utility

• Features include basic processing methodologies and tools• Selected cities serves as example prototypes to highlight capabilities

and features

NUDAPT Portal: Two systems, One Whole

• Quickplace– Powerful, flexible collaboration suite– Built-in security controls, file sharing ability– Leverages existing EPA Lotus Domino technology

• Data Download Portal– Delivers server-side data processing, minimizing or

eliminating the need for desktop GIS– Responsive data exploration map viewer– Relies on ESRI’s ArcGIS Server technology

Quickplace Welcome

Quickplace Summary

• Collaboration tool – what the group gets out depends on what the individual puts in

• Easy to share documents, model results, smaller datasets (less than 200MB), presentations, etc

• Available calendar/task management tools• Help build consensus on UCP methods

and strategies• Tool lets you manage the collaboration

Data Download Portal

• Map– AJAX for smooth dragging and zooming– Built-in identify, measure, and magnify tools– Dynamic table of contents

• Data repository– Quickly import data, add to map, publish to web– Tightly integrated with windows security– GIS tools allow fast, easy data pre-processing

Data Download Tool Inputs

• Input Raster or Basket of Rasters

• Clip Extent

• Output Coordinate Reference System

• Resampling Method

• Output Cell Size

• Output File Format

Clip Extent

• Draw extent directly on the map

• Tool uses bounding box envelope

• Envelope projected into spatial reference of raster and output

• Could investigate taking extent input in Lat/Long instead

Output Coordinate Reference System

• Initially contains only four systems, all NAD83– Geographic Latitude/Longitude– UTM Zone 15N– USGS Albers Equal Area– South Central Texas State Plane (Feet)

• ESRI Library contains hundreds, all could be added in a few minutes

• Also have option of leaving all rasters in source projection

Resampling Method

• Nearest Neighbor

• Bilinear Interpolation

• Cubic Convolution

Download Processing Flow

Input Raster

Input Raster

InputExtent

Polygon

OutputCoordinate

System

ResamplingMethod

Output Cell Size

OutputFile Format

GetFeature

Envelope

ClipExtent

Clip RasterClippedRaster

ClippedRaster

ProjectRaster

ProjectedRaster

ProjectedRaster

Clip Raster ClippedRaster

ClippedRaster

ConvertRaster to

Other FormatOutputRaster

OutputRaster

ZipOutputZip File

ProjectFeature

ProjectedEnvelope

GetFeature

Envelope

ProjectedClip Extent

Output Cell Size

• All rasters will be resampled to user-specified output cell size

• If no cell size specified, all rasters will remain at source resolution

• Regardless of input, no output will have smaller cell size than the minimum output resolution cutoff (15m)

• Minimum resolution determined by security restrictions

Output File Format• Available formats are:

– NetCDF– ASCII– Floating Point– Imagine Image– GeoTiff

• All output files (rasters, header files, metadata) are zipped for download

• Binary results “key” allows you to pick up output later

NUDAPT Tools

• Generalized methodology for alternative sets of UCPs

• Spatial allocation for (generalized regridding and grid geo-referencing capability

• Portal system and Internet collaboration

Extrapolating UCPs to Areas Without Data

• More than 90% grids in the modeling domain do not have UCP data.

• UCPs correlated to underlying land use in areas where base CPs correlated to underlying land use in areas where base data existed and then extrapolated to other areas by area-data existed and then extrapolated to other areas by area-weighted averaging (from Chingweighted averaging (from Ching, Burian).

• Building UCPs correlated to population (e.g., day, night, worker) Building UCPs correlated to population (e.g., day, night, worker) at 250-m and 1-km resolution (Burian).at 250-m and 1-km resolution (Burian).

• GIS Extrapolation Tool are programmed with models selected based on fit to data, testing results, and judgment to estimate UCPs based on population and land use (Burian).

NUDAPT ancillary data resources

• Anthropogenic heating (component of model thermodynamics)– Gridded (3-D)– Daily– Diurnal– Seasonal

• Population (Exposure applications)– Day– Night

• Advanced land use data, systems (Model evaluation, urban planning applications)– 100 City studies– Transims

UCPs for MM5 (see earlier) Mean and standard deviation of building and vegetation height Mean and standard deviation of building and vegetation height Plan-area weighted mean building and vegetation heightPlan-area weighted mean building and vegetation height Building height histogramsBuilding height histograms Plan area fraction and frontal area index at ground levelPlan area fraction and frontal area index at ground level Plan area density, top area density, and frontal area densityPlan area density, top area density, and frontal area density Complete aspect ratioComplete aspect ratio Building area ratioBuilding area ratio Building height-to-width ratioBuilding height-to-width ratio Sky view factor at ground level and as a function of heightSky view factor at ground level and as a function of height Aerodynamic roughness length and displacement height (Raupach, Aerodynamic roughness length and displacement height (Raupach,

Macdonald, Bottema, Coefficient)Macdonald, Bottema, Coefficient) Mean orientation of streetsMean orientation of streets Surface fraction of vegetation, roads, rooftops, and water and Surface fraction of vegetation, roads, rooftops, and water and

impervious area, directly connected impervious area, albedo and impervious area, directly connected impervious area, albedo and building material using remote sensingbuilding material using remote sensing

UCPs for urbanized WRF• Urban fraction• Building height, ZR• Roughness for momentum above the urban canopy layer, Z0C• Roughness for heat above the urban canopy layer Z0HC• Zero-displacement height above the urban canopy layer, ZDC• Percentage of urban canopy, PUC• Sky view factor, SVF• Building coverage ratio (roof area ratio), R• Normalized building height, HGT• Drag coefficient by buildings, CDS• Buildings volumetric parameter, AS• Anthropogenic heat, AH• Heat capacity of the roof, wall, and road• Heat conductivity of the roof, wall, and road• Albedo of the roof, wall, and road• Emissive of the roof, wall, and road• Roughness length for momentum of the roof, wall, and road• Roughness length for heat of the roof, wall, and road

Other model systems

• Canadian model based on TEB

• Global model with urban features

• COAMPS

• Advanced urbanized WRF with canopy-drag formulations

• Others?

Prototypes- by urban area• Houston

– High resolution building data base – DA-SM2U/MM5; uMM5, urbanized WRF, urban components in Global, urbanized

COAMPS, Canadian (TEB)– Coastal, bay breeze geo-climate– FDDA sea surface temperatures– Anthropogenic heating– Day-night population– Model evaluation databases TEXAS 2000, 2006– Model sensitivity studies to input of urbanized met model fields

• CMAQ AQ studies• Dispersion (HPAQ and HySplit) • Exposure assessments (AQ –hospital admissions study)

• Phoenix– Mountain valley flow regime – Rapid urbanization– Urbanized DA-SM2U/MM5– Urban heat island mitigation studies– Utilizes MODIS and ASTER data

• Atlanta– Either urbanized MM5 or WRF– Application of TRANSIMS – Exposure assessments

SUMMARY: Urban database conceptual design provides:

• Platform for advancing state of urban modeling- accomodates new modeling systems, new (sets of) parameterizations

• Community framework facilitates collaborations• Modeler’s focused system • Several tools including regrid and remap to different size & map

projections• Prototypes provide strategic means for extensibility of its capability

(copycat principle)• Is non stagnant (cities grow), can accommodate finer resolution data,

data refresh cycle.• Facilitates handover from model development to application

deployment

• EU sponsored Megacity study and its databases can be incorporated and accommodated as a special prototype

The EndThanks for your attention

Disclaimer: The research presented here was performed under the Memorandum of Understanding between the U.S. Environmental Protection Agency (EPA) and the U.S. Department of Commerce's National Oceanic and Atmospheric Administration (NOAA) and under agreement number DW13921548. This work constitutes a contribution to the NOAA Air Quality Program. Although it has been reviewed by EPA and NOAA and approved for publication, it does not necessarily reflect their policies or views.