An assessment of the LICOM Forecast System under the IVTT class4 framework

This paper evaluates LFS (LICOM Forecast System) forecasts and compares them with other marine forecast systems under the IVTT (Intercomparison and Validation Task Team) Class 4 framework. LFS delivers real-time daily forecasts driven by the GFS (Global Forecast System) atmospheric analyses and surface forecasts. The nudging method in LFS provides the initial state for forecasting, with only the temperature and salinity restored towards the Mercator PSY4 daily analyses. Assessments show that LFS demonstrates a reasonably good capability in short-term marine environment forecast. For the leading 1-6 days forecasts, the root mean square error (RMSE) ranges between 0.53-0.63 degrees C, 0.57-0.66 degrees C and 0.12-0.13 psu for the sea surface temperature, temperature, and salinity profiles, respectively. The overall performance is comparable to other major marine forecast systems, with a slight advantage in forecasting the temperature and salinity profiles. Different nudging time scales are applied to the upper ocean and deep ocean to preserve the effects of mesoscale processes and correct the large-scale biases in temperature and salinity. However, the absence of other observational constraints, such as the sea level height, significantly affects the regional forecast features. Further analyses are required to improve the performance, and the integration of an assimilation system into LFS is urgently needed.

Automated summary

This paper evaluates the performance of LFS (LICOM Forecast System) forecasts in comparison with other marine forecast systems. LFS provides real-time daily forecasts driven by atmospheric analyses and surface forecasts. The results show that LFS demonstrates reasonably good capabilities in short-term marine environment forecast, with comparable performance to other major marine forecast systems. However, further improvements are needed, such as the integration of an assimilation system and consideration of additional observational constraints.