Uncertainty in climate projections for the 21st century northwest European shelf seas

Abstract

There are a number of sources of uncertainty that impact climate projections for regional seas. We have assessed the impact that uncertain large-scale climate forcings have on the projections for the north-west European shelf seas. An ensemble of global Atmosphere-Ocean climate model (GCM) projections made by perturbed (atmospheric) parameter model variants which were designed to span uncertainty in climate sensitivity, was dynamically downscaled with the shelf seas model POLCOMS. The simulations were run as transient experiments (from 1952 to 2098) under a medium emissions scenario (SRES A1B). This study has focused on centennial changes over the period 2069–2098 relative to 1960–1989, but also refers to the full transient simulation to assess the significance of projected changes given interannual and low-frequency variability. The ensemble mean of the POLCOMS projections showed a shelf and annual mean Sea Surface Temperature (SST) rise of 2.90°C (±2σ=0.82°C), and a Sea Surface Salinity (SSS) freshening of −0.41psu (±2σ=0.47psu) between these periods. We described the spread in a field for a particular period using the variances associated with both the time mean ensemble dispersion (ensemble variance) and with the interannual variability. For SST in the present-day period, the magnitudes of both ensemble and interannual variance were fairly spatially homogenous. While the future interannual variance is of similar magnitude to that of the present day, the ensemble variance increased considerably into the future period. For SSS, both sources of variance were more spatially heterogeneous, and both increased into the future period. We investigated relationships between the projected shelf seas changes across the ensemble and changes in the large-scale climate forcing. We found that the near surface-air temperature from the driving GCM (averaged over the domain) and the GCM surface salinity to the west of the POLCOMS domain are good proxies for the changes within the shelf seas. We then compared these GCM indicators of shelf changes in our ensemble (under A1B) to the same measures across a number of CMIP5 models, under the RCP6.0 and RCP8.5 scenarios. The spread of these indicators, for our ensemble, fall within the range of the CMIP5 models (particularly under RCP8.5), suggesting our shelf projections would be consistent with an ensemble of projections driven by CMIP5 models.