Urban irrigation is an important component of the hydrologic cycle in many areas of the arid and semiarid
western United States. This paper describes a new approach that uses readily available datasets to
estimate the location and rate of urban irrigation. The approach provides a repeatable methodology at
1/3 km2 resolution across a large urbanized area (500 km2). For this study, Landsat Thematic Mapper
satellite imagery, air photos, climatic records, and a land-use map were used to: (1) identify the fraction
of irrigated landscaping in urban areas, and (2) estimate the monthly rate of irrigation being applied to
those areas. The area chosen for this study was the San Fernando Valley in Southern California.
Identifying irrigated areas involved the use of 29 satellite images, air photos, and a land-use map. The
fraction of a pixel that consists of irrigated landscaping (Firr) was estimated using a linear-mixture model
of two land-cover endmembers (selected pixels within a satellite image that represent a targeted landcover).
The two endmembers were impervious and fully-irrigated landscaping. In the San Fernando Valley,
we used airport buildings, runways, and pavement to represent the impervious endmember; golf
courses and parks were used to represent the fully irrigated endmember. The average Firr using all 29
satellite scenes was 44%. Firr calculated from hand-digitizing using air photos for 13 randomly selected
single-family-residential neighborhoods showed similar results (42%).