A daily precipitation-runoff model, referred to as the Los Angeles Basin watershed model (LABWM), was used to estimate recharge and runoff for a 5,047 square kilometer study area that included the greater Los Angeles area and all surface-water drainages potentially contributing recharge to a 1,450 square kilometer groundwater-study area underlying the greater Los Angeles area, referred to as the Los Angeles groundwater-study area. The recharge estimates for the Los Angeles groundwater-study area included spatially distributed recharge in response to the infiltration of precipitation, runoff, and urban irrigation, as well as mountain-front recharge from surface-water drainages bordering the groundwater-study area. The recharge and runoff estimates incorporated a new method for estimating urban irrigation, consisting of residential and commercial landscape watering, based on land use and the percentage of pervious land area.

The LABWM used a 201.17-meter gridded discretization of the study area to represent spatially distributed climate and watershed characteristics affecting the surface and shallow sub-surface hydrology for the Los Angeles groundwater study area. Climate data from a local network of 201 monitoring sites and published maps of 30-year-average monthly precipitation and maximum and minimum air temperature were used to develop the climate inputs for the LABWM. Published maps of land use, land cover, soils, vegetation, and surficial geology were used to represent the physical characteristics of the LABWM area. The LABWM was calibrated to available streamflow records at six streamflow-gaging stations.

Model results for a 100-year target-simulation period, from water years 1915 through 2014, were used to quantify and evaluate the spatial and temporal variability of water-budget components, including evapotranspiration (ET), recharge, and runoff. The largest outflow of water from the LABWM was ET; the 100-year average ET rate of 362 millimeters per year (mm/yr) accounted for 66 percent of the combined water inflow of 551 mm/yr, including 488 mm/yr from precipitation and 63 mm/yr from urban irrigation. The simulated ET rate varied from a minimum of 0 mm/yr for impervious areas to high values of more than 1,000 mm/yr for many areas, including the south-facing slopes of the San Gabriel Mountains, stream channels underlain by permeable soils and thick root zones, and pervious locations receiving inflows both from urban irrigation and surface water. Runoff was the next largest outflow, averaging 145 mm/yr for the 100-year period, or 26 percent of the combined precipitation and urban-irrigation inflow. Recharge averaged 45 mm/yr, or about 8 percent of the combined inflow from precipitation and urban irrigation.

Simulation results indicated that recharge in response to urban irrigation was an important component of spatially distributed recharge, contributing an average of 56 percent of the total recharge to the eight LABWM subdomains containing the Los Angeles groundwater study area. The 100‑year average recharge rate for the eight subdomains was 41 mm/yr, or 8,473 hectare-meters per year (ha-m/yr), with urban irrigation included in the simulation compared to a recharge rate of 18 mm/yr, or 3,741 ha-m/yr, with urban irrigation excluded. In contrast to recharge, the effect of urban irrigation on runoff was slight; runoff was 72,667 ha-m/yr with urban irrigation included compared to 72,618 ha-m/yr with urban irrigation excluded, an increase of only 48 ha-m/yr (about 0.1 percent).

Simulation results also indicated that potential recharge from hilly drainages outside of, but bordering and tributary to, the lower-lying area of the Los Angeles groundwater study area, in this study referred to as mountain-front recharge, could provide an important contribution to the total recharge for the groundwater basins. The time-averaged recharge rate was similar to the combined direct and mountain-front recharge components estimated in a previous study and used as input for a calibrated groundwater model. The annual (water year) recharge estimates simulated in this study, however, indicated much greater year-to-year variability, which was dependent on year-to-year variability in the magnitude and distribution of daily precipitation, compared to the previous estimates.