Transcript Framework for Inclusion of Urbanized Landscapes in a Climate Model

Framework for Inclusion of
Urbanized Landscapes in a
Climate Model
AGU 12 Dec., 2003
Robert E. Dickinson
Georgia Institute of Technology
What is the Issue?
• Science of Climate Change needs to be more
directed at issues of human welfare
• Approximately half of humanity lives in cities
• Hence need to scale projections of global
climate models to urban areas
• Need to include potential feedbacks to
atmosphere and hydrological cycle
Specific Issues for Modeling of Urban Surfaces
• Large fraction of impervious (mostly black)
surfaces that impact surface temperatures and
hydrology
• Urban plants – either more or less than natural
background -may be irrigated
• Turbulent transports generated by buildings
and vegetation at multiple heights
Modeling Issues Common to Other Surfaces
• Radiative energy balance – exchanges both
within elements and with overlying atmosphere
– Albedo
– Emissivity
– Geometry of objects
ratio –heat storage with a diurnal cycle
interacting with overlying boundary layer
• Bowen
•Conditioning of air for moist convection
thunderstorms
Data Issues
• What is urban?
– Usual definition of lights seen from space not helpful
to modeling
– Natural is to define in terms of a sufficiently large
fraction of urban concentrated surfaces such as road
density
•Fractional areas needed?
Roads & and other impervious flat surfaces
Vegetation cover – trees or grass, etc.
buildings
Data Issues (continued)
• Leaf and other plant areas (LAI & SAI)
• Spatial scales of vegetation – homogeneous or
mixed?
• Albedos (visible and Near –IR broadband)
• Water fraction (lakes and pools)
• Energy generated by electricity and fossil fuel
combustion
• All above over seasonal cycle
Much of above available from MODIS and
other NASA instruments
Focus on turbulent transport from
multiple surfaces
• Transport of heat, moisture, and other
constituents between surface and overlying
atmosphere
• Need to do separate moisture and energy
budgets for different surfaces
• These separate budgets require separate
formulations of transfer processes – usual
climate modeling assumption of separate
homogeneous surfaces probably fails
Mixing of transport from smooth and
rough surfaces
• Assume an atmospheric mixed layer above the
canopy = the tallest trees or buildings
• Transfer to mixed layer requires a z0 =
roughness length and h = displacement height
(centroid of momentum absorption).
• Also need to formulate the transfer from the
(relatively) flat surfaces to canopy air space –
same issue for grass and trees (savannah ) or
sparse vegetation in semi-arid regions
Complex Surface Parameters -Some
Background Literature
• Earliest study – U Wisc. MS thesis of John Kutzback,
1961– studies effect on surface roughness of density of
objects that were bushel baskets placed on frozen Lake
Mendota
• Mike Raupach1992,1994 formulates theoretically the
issue of partitioning of stress between smooth and such
roughness objects- relates roughness length and
displacement height to fractional area of obstacles
normal to wind direction
• Linroth, 1993, identifies LAI as contributing to tree effect
• MacDonald et al, 1998, formulates modified version for
buildings
Key Elements
• Uh = wind at the top of the canopy
• Cd = drag coefficient ,such that the overlying
atmosphere loses t = l Cd Uh 2 of momentum to
the complex canopy
• where l = effective frontal area density
• But sqrt(l Cd ) = 0.4 / log(( h – d)/ z0 ) from the
simple relationship between logarithmic wind
profile and momentum transfer, which is inverted
to get z0 / (h –d )
What’s not so easy!
• Formulation of l Cd ? Need vertical integral of
drag to scale as known coefficient only applies
at top of canopy – McDonald for buildings uses
a scaling factor of b (1-d/h) where b = 0.55 from
calibration
– For vegetation, the dependence on LAI + SAI
- this appears to be a scaling factor of
exp( - g (LAI + SAI) )
d to z0 and l: Raupauch works
for trees, not buildings, McDonald for
buildings not trees.
• Relating
Resolution of Roughness Issues
• McDonald scaling likely ok if modified to include
reduction by leaf area factor for vegetation
• Raupach expression for d/h should be modified
to allow for eddy sheltering (i.e. wake vortices of
building only interact over the area not covered
by building so that d should become h as
structures become a single building - however
for closed canopy, this is between 0.7 and 0.85 h
depending on crown-aspect ratio
Partitioning of fluxes between surfaces
• Raupach primarily interested in partioning of momentum
– important for wind erosion issues.
• All fluxes derivable through modeling of momentum
absorption and wind profile within the canopy
• Derive from wind profile, the low-flat surface fluxes – that
from the trees or buildings then obtained by subtracting
this from the total
• Has been shown to be a critical issue for semi-arid
vegetation; initial version of CLM2 released by NCAR
had skin temperatures over 10K too large. e.g.in Arizona,
because of an inappropriate treatment.
Directions
• Incorporation into Community Land Model (NCAR)
– Each atmospheric grid square currently is subdivided into
different plant functional types with prescribed properties
and with lakes a separate surface
– Need to extend these data descriptions to include a global
data base for impervious and building fractional areas –
improve lake sub-area data for urban locations
– Estimate if most or all includes surfaces are mixed on a
scale of a few km or less, and if so, improve the energy
exchange formulations as outlined above.
Conclusions
• Inclusion of urban surface types in global climate
models will facilitate assessment of climate
change of the environment in which lives half the
worlds population
• Doing this requires the development of new data
sets for existing climate models
• Doing this also requires further development of
the climate model treatments of complex
surfaces