Spatiotemporal Drivers of Municipal Water Consumption
MetadataShow full metadata
This study analyzed the individual and joint influences of social, urban, and physical landscape characteristics on patterns of municipal water consumption at the county scale for the state of Texas using a cross-sectional research design on three distinct temporal slices (1990, 2000, and 2010). Global multiple linear regression models and measures of global and local spatial association were combined to determine which landscape characteristics significantly influenced per capita municipal water consumption at the county scale, whether or not the statistically significant landscape characteristics varied over time, and to assess the degree to which the patterns and landscape characteristics of municipal water consumption reflected spatially stationarity. Overall results suggested that the social, urbanized, and physical environments contributed significantly to the patterns of per capita municipal water consumption to varying degrees in each year. The social and urbanized environments consistently exerted the strongest influences on per capita municipal water consumption, while the physical environment was generally less important. Additionally, the social environment had the greatest cumulative influence in all three years, and the urbanized environment singly accounted for the majority of the variation in per capita municipal water consumption when the joint influences of the other statistically significant landscape characteristics were considered. Differences in the composites of significant independent variables and the magnitudes of recurring significant variables between years, suggested that time influenced the landscape characteristics of per capita municipal water consumption. The spatial analysis of municipal water consumption patterns and its landscape characteristics suggested that they both exhibited weak to moderate degrees of spatial non-stationarity in each year. Furthermore, the measures of global and local spatial autocorrelation in the residuals of the multiple linear regression models suggested that spatial processes confounded the ability of a global model to adequately explain the significant driving landscape characteristics of per capita municipal water consumption.