Importance of high-resolution regional atmospheric modelling for the polar regions: Insights from the Polar CORDEX project

Reliable weather forecasts and future climate predictions for the polar regions are required by many users. For example, they are necessary for forecasters to predict severe and hazardous weather, such as extreme winds, to ensure the safety of flight operations and the polar tourism industry. Reliable climate predictions are also necessary for the polar regions as they are expected to see some of the largest temperature changes in response to global warming, which will result in important changes to other parts of the Earth system such as the ocean and ice.

Operational weather forecasts are often produced using atmosphere-only global numerical weather prediction (NWP) models at grid-spacings of around 50 km, while climate projections are produced using global climate models (GCMs) at grid-spacings of around 100 km.

However, the users in these regions are increasingly asking for more granulated, higher-resolution weather forecasts and climate predictions at grid-spacings of 1-10 km, which presents a major challenge for the current and next generation of atmospheric models. The World Weather Research Programme (WWRP) aims to address this challenge through the Polar Coupled Analysis Prediction for Services (PCAPS) project. Dr Andrew Orr - who is a climate scientist at the British Antarctic Survey, a member of the PCAPS steering group, one of the coordinators of the Polar Coordinated Regional Climate Downscaling Experiment (Polar CORDEX) project, and an investigator for the Polar Regions in the Earth System (PolarRES) project - reflects on how high-resolution regional atmospheric modelling is being used to address these challenges and help improve our understanding and decisions related to the polar regions. 

One way of providing accurate and detailed weather forecasts and climate predictions for the polar regions at spatial scales of 1-10 km is to use high-resolution regional atmospheric models. This approach is known as downscaling, with the high-resolution model used to dynamically downscale output from the relatively coarse-resolution global NWP and GCM simulations, as well as global atmospheric analysis / reanalysis products at grid-spacings of around 30 km, i.e., taking information known at large-scales to make predictions at local scales.

Crucially, atmospheric models at such fine spatial scales are able to represent the key complex processes affecting the local meteorology (and extremes) of the polar regions, including polar mesocyclones, marginal sea ice zone processes, clouds, land-sea contrasts, atmospheric boundary layer, and pronounced local topographic effects (including katabatic winds, warming induced by foehn winds, and enhanced precipitation).

An example of the benefits of high-resolution regional atmospheric modelling is shown in the figure below for the South Orkney Islands, which is a group of remote islands near Antarctica that support rich terrestrial ecosystems and areas of biodiversity. They consist of four main islands that are largely glaciated, with the largest (Coronation Island) being around 46 km long, 5-15 km wide, and 1,265 m high.

These islands are subject to highly spatially variable patterns (including extremes) of temperature, precipitation, and winds due to a combination of their pronounced local orography and their location within a region of enhanced storm activity over the Southern Ocean. In this example, the UK Met Office Unified Model (MetUM) at a grid-spacing of 1 km is used to downscale ECMWF Reanalysis v5 (ERA5) at a grid spacing of 30 km for the period 1 to 17 January 1991.

Panel a shows that the large variability in air temperature at 2 m that is observed over the islands during this period (including a distinct foehn-induced peak on 13 January that reaches 17°C) are captured reasonably well by the MetUM, but completely missed by ERA5. Furthermore, panels b and c show that the MetUM simulates the expected highly complex wind patterns over the islands, which are also completely missed by ERA5. The reason for these differences is that the MetUM is able to resolve the local orography of the islands due to its high spatial resolution, while for ERA5 the island orography is essentially unresolved and actually treated as ocean and not land, due to its relatively coarse resolution (panels d and e). 

Comparison of the observed temperature time-series at 2 m (°C) at Signy station (red dot) on South Orkney Islands from 1 to 17 January 1991 against output from MetUM model and ERA5 dataset (panel a). Comparison of 10 m wind over South Orkney Islands on 13 January 1991 from the MetUM model (panel b) and ERA5 dataset (panel c). Comparison of the height of orography from the MetUM model (panel d) and ERA5 dataset (panel e). 

The grid-spacing used by the MetUM model for the simulations is 1 km, and for ERA5 it is 30 km. 

Figure credits: Adapted from Lu et al. 2023.

A long-standing and successful example of high-resolution regional atmospheric modelling is the Polar CORDEX project. Polar CORDEX is a World Climate Research Programme (WCRP) initiative aimed at developing regional climate downscaling of the polar regions to provide skillful description of global to regional scale climate projections for input into impact and adaptation studies. Polar CORDEX strengthens cooperation and knowledge exchange on climate model simulations between polar climate modelling groups globally. It has also been hugely bolstered in the last few years by the EU-funded project PolarRES, which focuses on regional climate projections.

A coordinated set of simulations enables regional model inter-comparison studies, which are required to characterise uncertainties. Outputs from this project also contribute to studies assessing the impact of climate change on the polar regions and the development of adaptation and mitigation strategies, which are required by a diverse range of stakeholders. 

Both PCAPS and Polar CORDEX therefore recognise the importance of high-resolution regional modelling for the polar regions and the development of a coordinated way to achieve this, as well as being an excellent example of WWRP and WCRP initiatives working together.

The next step for this work will be the completion by the end of 2024 of a set of climate projections for the polar regions (for Polar CORDEX, via the PolarRES project) at an unprecedented grid spacing of 11 km, which will pave the way for more reliable assessment of the potential social and environmental aspects of climate change.