Developing a global database for the CLM urban model

As part of an ongoing effort to improve the simulation of human impacts on climate in the National Center for Atmospheric Research (NCAR) Community Climate System Model (CCSM) we are incorporating an urban canyon model within the Community Land Model (CLM). The urban model is intended to serve a number of purposes with the context of CCSM. Besides simulating urban climate, the model will also become an interface for improving emissions releases to the atmosphere, and will have to account for human energy consumption and releases to the atmosphere. One of the long-term goals for this model is to use it as a specific source model for various anthropogenically produced biogeochemical constituents.

As sub-model in a GCM, the model will have to have global applicability. This means that it must be able to model different types of urban systems. Practically this means we wish to have one model that can be parameterized to simulate urban systems that range from low-density suburban housing to large metropolitan areas. Furthermore, it must have the flexibility to simulate urban systems of different physical characteristics related to geographically varying city and building properties. While there are some unique aspects to the development of this model (discussed elsewhere in this conference) this paper will focus on the development of datasets to be used to drive this model on a global basis.

Our model needs several basic input parameters for simulating urban climates. These can be roughly divided into physical properties of the urban system (e.g. canyon height to width ratios, building height, building plan area, vegetated plan area etc.), properties related to building materials (wall and road albedo, conductivity, heat capacity, etc.) and an assessment of human behavior that affect energy consumption/releases into the system (thermostat settings for heating and cooling, vehicle energy releases and other sources of energy releases to the system). Since many of these statistics are not readily available on a global basis, we propose to use a number of data sources and surrogate data to construct the appropriate statistics over time and space.

Our primary source of information will be based on current global estimates of human population density. Using the LandScan 30 arc minute resolution population database (http://www.ornl.gov/sci/gist/landscan/) we will divide the world into three urban classes (detached housing, medium density row housing, and central business district/high density city housing). Each type of urban class is assigned a number of physical properties reflecting typical class properties. Initially this will be held constant over space, but the intent is to vary these properties based on national level statistics related to housing types and typical urban architectures. While there are known drawbacks to linking urban characteristics to population density, this method was chosen in part because it can be related to historical and future projections of national level population data to simulate population density through time.

Estimating properties of various urban surfaces has long been a problem in urban canyon modeling efforts. For our purposes we believe we can assign fairly generic properties to each urban surface by urban class. However, traditional building materials vary widely around the world, so we are collecting information from the world housing encyclopedia (http://www.world-housing.net/) on typical building materials and house properties. From this information we will use look up tables to assign building material properties over time and space.

Finally we wish to resolve the amount of anthropogenic heat production for typical urban environments. This will vary significantly over space and time due to different building properties (e.g. insulation), standard of living differences, and based on such things as the number of vehicles per capita. One way we propose to solve these geographical differences is to calculate the amount of heat production based on how interior temperature regimes vary over space. We have included in our model a capability that allows the interior building temperature to vary between two preset values. Thus a minimum interior temperature setting determines the amount of heating used in the building. Similar the maximum temperature determines the energy consumed for cooling. When these values are set equal this is equivalent to maintaining a constant interior temperature, while allowing for a wide interior temperature range simulates minimal use of heating and cooling. Based on the interior building temperature the model calculateds the energy consumed to maintain a particular interior temperature regime. Assuming a 50% efficiency of typical appliance, we double this energy consumption statistic and release this energy source to the urban canyon as anthropogenic heat flux. Within this 50% assumption we also include energy created from electrical appliances and other human heat sources. Eventually with more data we can calibrate this efficiency rating over time and space. We also propose to include heat sources from traffic based on typical traffic patterns and vehicle density statistics, which will be tied to national level per capita vehicle data and standard of living indices.

Building these datasets will be a multistage process, and is currently in the early stages of development. From the outset one important criterion for this project is that we be able to allow the data to vary through time. This largely determines the data sources we have elected to use and the level of information that is available.