Measuring urban growth, urban form and accessibility as indicators of urban sprawl in Hamilton, New Zealand

Abstract: Hamilton City is currently the fourth most populous territorial authority in New Zealand. The city boundary was extended in 1989 in order to provide sufficient land for urban growth for at least 25 years. Despite being neither unplanned nor unchecked, urban growth within this boundary has been branded by the media as urban sprawl. Urban sprawl is a complex phenomenon with a wide range of definitions incorporating aesthetic judgements, unwanted externalities, policy consequences, land development patterns, and urban growth rates. The negative economic, environmental, social and public health effects of urban sprawl are widely considered to outweigh any positive effects, leading to the term having a negative connotation. Just as there are many ways to define urban sprawl, there are also many ways to measure the phenomenon including urban area growth rates, density measurements, and spatial geometry, as well as differences in access to public services, employment opportunities and commercial areas. The primary research question addressed in this study is whether the post-1989 urban growth in Hamilton should be categorised as urban sprawl? Remote sensing and GIS techniques have been used to measure urban growth, urban form and accessibility in order to address three sub-questions which reflect different ways of defining and measuring urban sprawl: • Has growth in the Hamilton urban area occurred at a greater rate than the city’s population growth? • Are new neighbourhoods more homogeneous, having a higher proportion of single family dwellings on larger land parcels, and do they lack street connectivity? • Do residents of old and new neighbourhoods have different access to essential services, commercial areas, employment areas, and public transport? Post-classification comparison of satellite imagery was used to measure urban growth. A supervised classification was performed on three Landsat images acquired in 1990, 2001 and 2014. Pixels in each image were classified into a common land cover schema comprising eight classes: urban-residential, urban-commercial/industrial, cleared land, grassland, natural vegetation, agriculture-crops, agriculture-fallow and water. Pixels were then reclassified to three classes (urban, non-urban and water) and the Hamilton urban area was quantified for each classification map by multiplying the number of urban pixels by the pixel size. An accuracy assessment showed the classification maps to have overall accuracy of 94-97% and Kappa estimation of 90-96%. Between 1990 and 2014 there was 45% growth in the Hamilton urban area, while census data showed 41% growth in the population residing within Hamilton City between 1991 and 2013. Hence growth in the Hamilton urban area has not occurred at a greater rate than the city’s population growth. Therefore, based on this method of measuring urban sprawl, the post-1989 urban growth in Hamilton should not be categorised as urban sprawl. Cadastral, land use and road data were used to calculate six metrics of urban form for each of the 37 residential neighbourhoods in Hamilton: land use mix, dwelling density, single-dwelling proportion, single-dwelling parcel size, internal street connectivity and external street connectivity. There are statistically significant differences in these metrics between old (developed prior to 1990) and new (developed from 1990 to 2014) neighbourhoods. New neighbourhoods are currently more homogeneous, they have a higher proportion of single family dwellings and lack street connectivity. However these single family dwellings do not occupy larger land parcels, which may be a function of several factors including subdivision policies, market demand and land prices. Therefore, based on five out of the six metrics of urban form, the post-1989 urban growth in Hamilton should be categorised as urban sprawl. Cadastral, facility and road data were used to calculate eleven metrics of accessibility for each residential neighbourhood. For residential land parcels the median distance to the nearest commercial land use, bus stop, primary school, employment area and medial clinic, and from the nearest police station, fire station and ambulance station was calculated for each neighbourhood. Pedestrian access (walkability) was calculated as the percentage of residential parcels in the neighbourhood within walking distance (800m) of a commercial land use, bus stop and primary school. There are statistically significant differences in these metrics between old and new neighbourhoods, and residents currently have different access to commercial areas, employment areas, public transport and some essential services. Residents of new neighbourhoods currently experience increased travelling distances and reduced accessibility and walkability. However there is currently no difference between old and new neighbourhoods in terms of access to police and ambulance services, suggesting that the locations of these facilities are well-balanced across the city. Therefore, based on nine out of the eleven metrics of accessibility, the post-1989 urban growth in Hamilton should be categorised as urban sprawl. The results of this study demonstrate the complexity of the urban sprawl phenomenon and that whether the post-1989 urban growth in Hamilton should be categorised as urban sprawl depends upon the particular definition of urban sprawl that is adopted and the measurement method used.