Adding baroclinicity to a global operational model for forecasting total water level: Approach and impact

Abstract

Operational flood forecast systems are typically based on barotropic models in order to limit computational cost. In this study, a computationally efficient way of adding spatial and temporal variations of water density (henceforth “adding baroclinicity”) to a global barotropic forecast system is described and evaluated. The approach is illustrated using a global high-resolution (1/12°) model with an optimized vertical grid of nine ocean levels. Temperature (T) and salinity (S) for each layer are weakly nudged to forecast fields provided by a coarser resolution (1/4°), data-assimilative ocean model. This adds realistic baroclinic variability with periods exceeding about 15 d to the 1/12° model while allowing the higher frequency variability to evolve freely. The computational cost of running the new model is only 10% higher than the cost of running the two-dimensional barotropic version of the model that forms the basis of the present total water level (TWL) forecast system. The value of adding baroclinicity is demonstrated across the sea level frequency spectrum using hourly observations of TWL made between October 20, 2019 and February 28, 2021 by a global distribution of 211 tide gauges. Large improvements in the tidal residuals (TWL-tide) are shown for periods exceeding 15 d. This is consistent with the known importance of baroclinic processes to sea level variability in this frequency range. For periods between 10 h and 15 d, baroclinicity also improves the predictions of extreme residuals by up to 15 cm, with the largest improvements occurring on narrow shelves. In terms of extremes, adding baroclinicity improves predictions of the three largest positive tidal residuals for 2020 at 89% of the tide gauges with the largest improvements reaching 42 cm. Similar results were found for the three largest negative residuals for each tide gauge record. The importance of modeling baroclinically-modified coastal trapped waves is illustrated for the west coast of North and South America.