High and medium resolution ocean models for the Great Barrier Reef

Abstract This article describes 3D medium (4km) and high (500m) horizontal resolution ocean models developed for the Coral Sea with a particular focus on the Great Barrier Reef lagoon. The medium resolution model is underpinning a 10 year reanalysis and a forecast demonstrator system for maritime monitoring. The ocean models have been developed to assist the management of the Great Barrier Reef given the existence of a number of stressors acting on the reef. The presented models aim to represent physical stressors that include thermal, salinity and flooding rivers. The river discharge is tracked by passive and age tracers. They can provide concentrations to identify the area/volume influenced by flood water. This paper gives an overview of the forward non-linear models with the aim to assess their performance. The models have been evaluated over a two year period spanning January 2009–March 2011 including the 2010/2011 La Nina and the advent of Tropical Cyclone Yasi in early 2011 that contributed to one of the largest river discharge events recorded over northern Queensland. The models’ performance is investigated on a full domain scale via means of satellite sea surface temperature (SST), expendable bathythermograph (XBT) and drifter data. Over the shelf and the Great Barrier Reef lagoon, available tide gauges, Ocean Colour, Chlorophyll-a and Queensland’s Integrated Marine Observing System (QIMOS) mooring data of temperature, salinity and velocity have been used to assess the models. Results suggest that the high resolution model gives better statistics in terms of mean absolute deviation and bias compared to the medium resolution model for subsurface temperature and salinity over the deep ocean. Moreover, sea surface temperature performance is similar for both models. Over the lagoon and shelf both models’ performance is mixed. The high resolution model displays a consistent warm bias on the order of 0.3 °C over the shallow well mixed stations throughout the year. However, the 4km model displays larger variability in this region. It tends to be too warm in summer and too cool in winter when compared to moorings. Modelled velocities are assessed via means of the QIMOS moorings. Overall the high resolution model fares better in terms of current direction with an average veering of 13° vs. 22°. The models’ performance is comparable in terms of current speed over the shallow moorings. However, the high resolution model is better equipped to represent steeper shelves and this results at times in better performance for velocity and temperature, particularly over Palm Passage. A large focus has been placed on including and assessing passive tracer dynamics emanating from river discharge. 25 river sources have been added across the models domain together with a set of one-dimensional channels representing the volume flux of brine. Two case studies are presented to demonstrate the models’ ability to reproduce mean features of observed plumes. A major river discharge event has been examined in further detail. Both models are able to reproduce the low salinity signal present in the observations in the upper level to a satisfactory degree. The 4km model matches the observed strength and duration of the fresh water plume substantially better than the high resolution model in this case.