Current numerical models used for irrigation scheduling rely on single point measurements to predict irrigation needs for entire fields. To incorporate the spatial dimension, remotely-sensed imagery was investigated as a tool for estimating water use by cotton in central Arizona. This report presents the second year of a study situated in a furrow-irrigated field with large variation in soil texture. Aerial imagery was obtained every two to three weeks using a high resolution camera system equipped with narrow band-pass filters and calibrated with ground-based reference tarps. The normalized difference vegetation index (NDVI) was used to estimate the crop coefficient (K cb) for cotton via a previously defined relationship. The K cb plus estimated soil evaporation coefficients were multiplied by reference evapotranspiration (ET o) determined from a nearby weather station and summed during each irrigation interval to provide a seasonal water use map of the field. The map was validated using soil water balance with periodic soil moisture measurements. Areas of the field with less vigorous canopy were poorly predicted in 2003, although ET in areas with more robust growth was well characterized. Although ET predictions in 2003 were not as accurate as in 2002, the study demonstrated that if relationships between imagery and ET can be developed, maps showing the spatial and temporal dynamics of crop water use can be created. Water use maps can offer insight into spatially-variable crop response and help define and manage zones in surface irrigated fields. Copyright ASCE 2005.