As in many of the world's reservoirs, floods provide most of the water to Lake Burragorang (Sydney, Australia). In this study a winter flood through this large (2km3), long (ca. 50 km) and narrow (ca. 0.5-2km) reservoir was monitored at multiple stations over several weeks. The nutrient-laden underflow took 7-8 days to travel from the river confluence to the dam and upwardly displaced the pre-flood anoxic hypolimnion. Particulate phosphorus and nitrogen, and turbidity settled at an exponential rate (0.34-0.42 d-1) in the head of the underflow. Application of a three-dimensional model (ELCOM) with large grid sizes (200×200×1m) for reasonable run times (ca. 1 day) did not resolve the underflow hydrodynamics in this large, narrow and complex geometry. A technique for 'straightening' morphologies that are narrow and curvy was applied to the Lake Burragorang bathymetry to allow large model grid sizes. Simulations with the 'straightened' bathymetry had low run times and compared well with field data. Then, an aquatic ecological model (CAEDYM) was coupled to ELCOM to simulate the biogeochemical fate of the floodwaters. Simulated spatial patterns of dissolved oxygen, and filterable and particulate nutrients compared well with field data. Further, the ELCOM-CAEDYM simulation demonstrated that the dynamic spatial variations in water quality from a Eularian perspective during such flood events are dominated by transport. Settling is the dominant process in the head of the underflow from a Lagrangian frame of reference. To simulate spatial variations during a flood it is adequate to couple an accurate hydrodynamic driver to a simply configured biogeochemical model.