The primary objective of WUDAPT is to gather data on the world’s cities suited to urban climate studies. These data should be gathered in a consistent manner so as to facilitate comparison between cities and include information on urban form (surface cover, materials and building geometry) and function (human activities that drive water and energy use) that can be used for modelling (Table 1).
Table 1: Urban characteristics and parameters
|Cover||Vegetative, building, impervious surface cover|
|Material||Wall type, roof type, window type, road materials, window fraction on the wall, colour/albedo|
|Geometry||Building height, width of streets, contiguous or isolated buildings, roof geometry|
|Function||Building use, irrigation, road type, temperature setting, occupancy, air conditioning, shutters or shading, window opening, building age, building renovation post 1990.|
WUDAPT has a hierarchic approach to gathering data:
- Level 0: Cities are mapped using the Local Climate Zone (LCZ) scheme (Stewart and Oke, 2012), which categorizes landscapes into 10 urban and 7 natural surface cover types. Each LCZ type is described in terms of the typical appearance of each in ground-based and aerial photographs and is linked to some urban parameter values.
- Level 1: The LCZ maps are used to sample the urban landscape to provide information on more aspects of form and function in greater detail.
- Level 2: This is the highest level and refers to urban data gathered at a specified spatial scale (e.g. 250 m) across the entire urban area (‘wall-to-wall’ coverage).
Level 0 data is the coarsest level of data gathering but should provide comprehensive and consistent coverage. These data describe the urban landscape in terms of neighbourhood-scale (>1km2) spatial units using the LCZ scheme. The classification of an urban area into LCZ types is based on semi-automated process (Bechtel and Daneke, 2012, Bechtel et al. 2015) using available multi-spectral satellite imagery (Landsat8) and free SAGA (Conrad et al., 2015) and Google Earth software. For this process to function, a sample of training areas that identify LCZ types across a selected city is needed. These training areas are used to identify the statistical characteristics of LCZ pixel values within the available multi-spectral images; this information is used then to develop a model that categorises the entire images into LCZ types.