The Irish Atlantic CoCliME case study configuration, validation and application of a downscaled ROMS ocean climate model off SW Ireland.

This ocean numerical modelling study is focused on the shelf waters off southwest Ireland. Outputs from the model are used to provide scientists and policy makers with climate change information related to oceanic conditions that influence harmful algal blooms in the region. Four simulations were developed to include a 20-year hindcast simulation (1997-2016), a 31-year historical simulation (1975-2005), and two 30-year future climate model projections (2006-2035) for the representative carbon pathways (RCP) 4.5 and 8.5 scenarios. We evaluated the hindcast model skill by comparing the simulation outputs with measured observational data and calculated statistics such as the bias, and the root mean square error (RMSE). The observations consisted of satellite sea surface temperature and, CTD temperature and salinity profile data collected in Irish waters. The sea surface temperature RMSEs between the 20-year hindcast simulation and the satellite data were ca. 0.50°C. A minimum RMSE of ca. 0.20°C was recorded in December 2015. The basin-averaged bias (difference) and RMSE for near bottom temperature between the RCP 4.5 and the historical simulation was 0.14°C and 0.27°C respectively. The sea surface temperature anomaly pattern for the RCP 8.5 shows warming across the whole model domain. The sea surface salinity and near bottom salinity climate simulation anomaly maps indicate a general freshening on the southwest Irish shelf. A change of ca. 0.2 sea surface salinity and near bottom salinity was observed. The RCP 8.5 simulation shows the highest current velocities when compared to other simulations. The Irish coastal current pattern under the RCP 8.5 scenario appears very intense and well defined with a velocity > 20 [cm/s].

Harmful algal bloom forecast system for SW Ireland. Part I: Description and validation of an operational forecasting model.

A 3D primitive equation coastal ocean model for southwest Ireland, called the Bantry Bay model, was developed and implemented operationally. Validated model outputs have multiple uses. One of the incentives to develop the model was to explore the possible transport pathways that carry harmful algae blooms (HAB) into Bantry Bay. The model is nested offline in a regional North East Atlantic operational model. Surface forcing is taken from the half-degree Global Forecasting System, available at three-hourly intervals. Heat fluxes are calculated from the bulk formulae. Surface freshwater fluxes are obtained from the prescribed rainfall rates and the evaporation rates calculated by the model. Freshwater discharges from five rivers are included in the model. Model validation and the model skill in representing the water level, currents, temperature and salinity in the bay are reported. A scoring system based on the average adjusted relative mean absolute error for the predicted currents was used. An upgrade to a higher score was achieved through the incorporation of local winds into the surface forcing and by varying the bottom roughness coefficient. The model, designed to work in forecast mode, can replicate the main oceanographic features in the region. The model forecast is used in a decision support system for HAB alerts. An operational HAB alert system did not exist in Ireland prior to the use of this model.

Harmful algal bloom forecast system for SW Ireland. Part II: Are operational oceanographic models useful in a HAB warning system.

This study investigated the application of a three-dimensional physical hydrodynamic model in a harmful algal bloom forecast system for Bantry Bay, southwest Ireland. Modelled oceanographic conditions were studied and used to help understand observed changes in the chemical and biological patterns from the national biotoxins and phytoplankton monitoring program. The study focused on two toxic events in 2013. An upwelling event was predicted by the model prior to the appearance and population increase of potentially toxic diatoms, Pseudo-nitzschia, and associated domoic acid in shellfish. A downwelling episode was provided as a forecast in the model prior to the arrival of a Dinophysis bloom and detection of its associated biotoxins in Bay shellfish. The modelled forecast products developed included expected surface, mid-depth and bottom current pathways at the mouth of the Bay and on the adjacent shelf. The rate and direction of water volume flow at the mouth and mid-bay sections were produced by the model to examine predicted upwelling and downwelling pulses. The model also calculated the evolution of water properties (temperature, salinity and density) with depth along the Bay axis and on the adjacent continental shelf. Direct measurements of water properties at a fixed point, mid-bay, were comparable to model calculations. The operational model for southwest Ireland produces a reliable 3-day physical hydrodynamic forecast of the dominant regional physical processes that result in water exchange events between Bantry Bay and its adjacent shelf. While simulated physical hydrodynamics were provided as a 3-day forecast, the upwelling and downwelling signals from the model, closely linked to toxic HAB episodes, were evident up to 10 days prior to the contamination of shellfish in the Bay.