Spatial Analysis of Water Use in Oregon, USA, 1985-2005

Water use patterns are not distributed evenly over space and time. Determining the amount of water used within a region, as well as the various ways in which water is used is important for making adequate and sustainable water management policies and determining future water availability. We examined differences in spatial trends in Oregon freshwater use (total, municipal, and agricultural water withdrawals), by county, between the years 1985 and 2005. We also explored biophysical and socioeconomic factors that explain spatial patterns using Moran's I, local index of spatial autocorrelation (LISA), and spatial regression models. There was a moderate positive spatial autocorrelation among counties that had similar total and irrigation withdrawals. LISA analysis identified hot spots between certain arid agricultural counties in the southeastern Oregon and cold spots between certain humid northwestern counties, including within the Portland metro area. Annual precipitation and income are negatively associated with total water withdrawals. Summer temperature and farm size is positively associated with irrigation water withdrawals, while net cash return and income are negatively associated with irrigation water withdrawals. When compared to ordinary least square regression models, spatial error models that take into account spatial dependence provide a more comprehensive explanation of the variations of water use, suggesting that water resource planning and management should incorporate spatial and neighborhood effects to effective manage limited natural resources.