Analysis of the local sea state derived through in-situ, remote sensing, and numerical model

Abstract

For this study, the physical ocean data measured remotely from High Frequency Radars (HFR) and the data calculated from the SHYFEM numerical model were analysed against in-situ observations, which was carried out on two geographical scales located on the West side of the Maltese Islands, referred to hereafter as the micro and macro Areas of Study during 2020, 2021 and 2023. Within the micro AoS, the trajectories observed by Lagrangian-type SouthTEK drifters were visually compared to model-simulated trajectories. The comparision carried out is primarily a visual comparison. In terms of magnitude, for the majority of the plot model replicated quite well the magnitude of drifter’s velocity, while for a few, the model either predicted a faster or a slower moving current. In terms of direction, the model replicated quite well the drifters’ trajectory, while a few of the model-simulated trajectories did not replicate the drifters’ trajectory at all. Additionally, the 2-D temperature and salinity profiles for two transects, which were gathered using CTD sonde casts, were compared to the 2-D temperature and salinity profiles generated by the model. The temperature profiles between the two datasets indicated that the model replicated well the insitu measurements throughout each profile, with some exceptions during the seasonal transition period. With regards to the salinity profiles, the observed salinity ranged from 34.4 to 36.6 pss, with the salinity generally decreasing with increasing depth. Meanwhile, the modelled salinity ranged from 37.6 to 37.95 pss, with the salinity generally decreasing with increasing depth, although with some exceptions. Furthermore, CODE-type drifters were deployed for several days within the macro AoS. Their trajectories were then simultaneously compared in terms of surface current vector fields and magnitude. Overall, the radar successfully replicated 8 out of 10 trajectories, while the model replicated 6 out of 13 trajectories. Thus, it can be concluded that the radar replicated the drifters’ trajectories better than the model. This conclusion is based on the assumption that the drifters’ trajectories are a representation of the ground truth data while assuming a radar-simulated trajectory without any data gaps. Within both AoS, the temperature sensor on both the SouthTEK and CODE drifters also provided SST data, which were also compared to the modelled SST data. Overall, the model replicated fairly well the observed SST data however, given the scale difference between the two areas of studies, the comparison within the micro AoS visualised certain outliers, such as overestimation by the model from winter to spring, and underestimations in June and August potential due to seasonal variability and marine heatwave.