Towards Winter Seasonal Predictability of the North West European Shelf Seas

Abstract

We investigate the winter predictability of the North West European shelf seas (NWS), using the Met Office seasonal forecasting system GloSea5 and the Copernicus NWS reanalysis. We assess GloSea5’s representation of NWS climatological winter and its skill at forecasting winter conditions on the NWS. We quantify NWS winter persistence and compare this to the forecast skill. GloSea5 simulates the winter climatology adequately. We find important errors in the residual circulation (particularly in the Irish Sea) that introduce temperature and salinity biases in the Irish Sea, English Channel, and southern North Sea. The GloSea5 winter skill is significant for SST across most of the NWS but is lower in the southern North Sea. Salinity skill is not significant in the regions affected by the circulation errors. There is considerable NWS winter temperature and salinity persistence. GloSea5 exhibits significant predictive skill above this over ∼20% of the NWS, but for most of the NWS this is not the case. Dynamical downscaling is one method to improve the GloSea5 simulation of the NWS and its circulation, which may reduce biases and increase predictive skill. We investigate this approach with a pair of case studies, comparing the winters of 2010/2011 and 2011/2012 (with contrasting temperature and salinity anomalies, and NAO state). While 2 years are insufficient to assess skill, the differences in the simulations are evaluated, and their implications for the NWS winter predictability are considered. The NWS circulation is improved (where it was poor in the GloSea5), allowing more realistic advective pathways for salinity (and temperature) and enhancing their climatological spatial distributions. However, as the GloSea5 SST anomaly is already well simulated, downscaling does not substantially improve this – in other seasons or for other variables, downscaling may add more value. We show that persistence of early winter values provides some predictive skill for the NWS winter SST, and that the GloSea5 system adds modestly to this skill in certain regions. Such information will allow prospective end-users to consider how seasonal forecasts might be useful for their sector, providing the foundation on which marine environmental seasonal forecasts service and community may be developed for the NWS.

The influence of tides on the North West European shelf winter residual circulation

“…tidally induced mean flow over bathymetric features; Polton et al, 2005). There are many studies of climate impacts (Townhill et al, 2017;King et al, 2021;Nagy et al, 2021), present conditions (Tinker et al, 2020), and seasonal predictability (Tinker et al, 2018;Tinker and Hermanson, 2021) which rely on dynamic downscaling, however this is technically challenging, and computationally expensive. A better option would be to use the GCM data directly, but their simulation of the NWS is rarely realistic enough.…”

“…The Met Office seasonal forecasting modelling system, GloSea5 (MacLachlan et al, 2014), is based on the Met Office Hadley Centre GCM, HadGEM3-GC2 (Williams et al, 2015), with a non-tidal ocean model component based on NEMO ORCA025 (Storkey et al, 2018). Tinker and Hermanson (2021) assessed GloSea5's NWS winter circulation, temperature and salinity. They concluded that there were important circulation errors that would affect some advective pathways, and so temperature and salinity -these errors improved when the GloSea5 was dynamically downscaled with a tidal shelf seas model [NEMO Coastal Ocean version 6 (CO6, O'Dea et al, 2017), see later for details].…”

“…They concluded that there were important circulation errors that would affect some advective pathways, and so temperature and salinity -these errors improved when the GloSea5 was dynamically downscaled with a tidal shelf seas model [NEMO Coastal Ocean version 6 (CO6, O'Dea et al, 2017), see later for details]. These improvements (to circulation, temperature, and salinity) were thought to be caused by the presence of tides (Tinker and Hermanson, 2021): therefore, this study will investigate the impact of tides on the NWS, when CO6 is run with and without tides.…”

“…1.1 GloSea5 simulation of the NWS Tinker and Hermanson (2021) evaluated the NWS winter (Dec-Feb) conditions of the (non-tidal) global ocean model component of GloSea5 by comparing with the Copernicus Marine Environment Monitoring Service NWS reanalysis 1 (Renshaw et al, 2019, termed CMEMS hereafter). CMEMS is based on the NEMO CO6 tidal model run for 23 years , with sea surface temperature, and temperature and salinity profile data assimilation.…”

“…In GloSea5, the fresher cooler English Channel is (incorrectly) of similar temperature and salinity to the southern North Sea, so the plume is not visible. Tinker and Hermanson (2021) used model data from GloSea5 to drive a regional shelf seas model (NEMO CO6) to dynamically downscale two winters as a case studythis improved the spatial resolution over the NWS, and added important shelf seas processes (including tides). The circulation errors (between GloSea5 and CMEMS) reduced in these downscaled simulations, as did the temperature and salinity biases.…”