Household Water Demand and Land Use Context: A Multilevel Approach.

摘要

Urban water use arises from a mix of scale-dependent biophysical and socioeconomic factors. In Portland, Oregon, single-family residential water use exhibits a tightly coupled relationship with summertime weather, although this relationship varies with land use patterns across households and neighborhoods. This thesis developed a multilevel regression model to evaluate the relative importance of weather variability, parcel land use characteristics, and neighborhood geographic context in explaining single-family residential water demand patterns in the Portland metropolitan area. The model drew on a high-resolution panel dataset of weekly mean summer water use over five years (2001--2005) for a sample of 460 single-family households spanning an urban-to-suburban gradient. Water use was found to be most elastic with respect to parcel-scale building size. Building age was negatively related to water use at both the parcel and neighborhood scale. Half the variation in water use can be attributed to between-household factors. Between-neighborhood variation exerted a modest but statistically significant effect. The analysis decomposed household temperature sensitivity into four components: a fixed effect common to all households, a household-specific deviation from the fixed effect, a separate extreme heat effect, and a land use effect, where lot size exaggerated the effect of temperature on water use. Results suggested that land use planning may be an effective non-price mechanism for long-range management of peak demand, as land use decisions have water use implications. The combined effects of population growth, urbanization, and climate change expose water providers to risk of water stress. Modeling fine-grain relationships among heat, land use, and water use across scales plays a role in long-range climate change planning and adaptation.