The impact of urbanization on water and energy fluxes varies according to the characteristics of the urban patch type. Nevertheless, urban flux observations are limited, particularly in arid climates, given the wide variety of land cover present in cities. To help address this need, a mobile eddy covariance tower was deployed at three locations in Phoenix, Arizona, to sample the surface energy balance at a parking lot, a xeric landscaping (irrigated trees with gravel) and a mesic landscaping (irrigated turf grass). These deployments were compared to a stationary eddy covariance tower in a suburban neighborhood. A comparison of the observations revealed key differences between the mobile and reference sites tied to the urban land cover within the measurement footprints. For instance, the net radiation varied substantially among the sites in manners consistent with albedo and shallow soil temperature differences. The partitioning of available energy between sensible and latent heat fluxes was modulated strongly by the presence of outdoor water use, with the irrigated turf grass exhibiting the highest evaporative fraction. At this site, we identified a lack of sensitivity of turbulent flux partitioning to precipitation events, which suggests that frequent outdoor water use removes water limitations in an arid climate, thus leading to mesic conditions. Other urban land covers with less irrigation, however, exhibited sensitivity to the occurrence of precipitation, as expected for an arid climate. As a result, quantifying the frequency and magnitude of outdoor water use is critical for understanding evapotranspiration losses in arid urban areas.