Advanced models and weather prediction in the Arctic: Enhanced capacity from observations and polar process representations (ALERTNESS)

Improved weather prediction in the Arctic Climate change leads to increased activity in the Arctic, but the high-impact weather (HIW) common in these areas is hazardous to marine operations and industrial development. The ambition of Alertness is to develop world-leading capacity for the delivery of reliable and accurate Arctic weather forecasts and warnings for the benefit of maritime operations, business and society. The unique Arctic weather conditions challenge the observation systems and numerical weather prediction (NWP) models, and thereby our ability to predict the weather. Alertness has established and employs a database of 21 Arctic HIW events, such as polar lows, maritime icing, fog and cold-air outbreaks and relevant field campaigns. The methodological basis of Alertness is formed by AROME Arctic. Alertness has made significant contributions to establish AROME Arctic as an operational NWP model system for the Arctic. Alertness puts science for services and society into practice: AROME Arctic is used by Yr.no, weather forecasters, student on field campaigns at UNIS, etc. AROME Arctic is the backbone of the recently released Copernicus Arctic Regional ReAnalysis (CARRA). Alertness develops and applies verification metrics and diagnostics for NWP in the Arctic. When compared to other weather forecast models, AROME Arctic performs very well. A set of common weaknesses across forecast systems are revealed, such as (i) forecasting temperature during cloud free, calm weather, (ii) an underestimation of temperature in windy conditions and (iii) an underestimation of solid precipitation. Satellite data (ASCAT and SAR) are used to investigate how grid-spacing and convection representation in NWP models affects the wind speed forecasts of polar lows, illustrate how observation, interpolation and representativeness errors can be taken into account, and demonstrate how temperature diagnostics can highlight the problematic NWP behaviour in connection with the stable boundary layer. In Alertness, we implemented individual tendency output as a new diagnostic in AROME Arctic, and demonstrated its utility during a marine cold air outbreak. We could identify controlling factors for the activity of certain physical parameterisation schemes, study the interplay between them, and investigate the model?s subgrid-scale as well as grid-scale responses to changes in its physics package. Alertness enhances the use of existing Arctic observation systems in data assimilation to improve the weather forecast initialization and accuracy, as well as advancing future observing systems. We have developed a methodology (supermodding) to improve the uptake of wind scatterometer (satellite) data by accounting for its effective resolution. Unique regional observing system experiments (OSEs) coordinated with global OSEs of the EU-APPLICATE project, show the relative impact of different observations on forecast accuracy, disentangling the benefits of observations on local forecast accuracy. The model representation and initialization of sea ice as it is very important in the Arctic area. We have advanced its initialization through an adaptive bias-aware Extended Kalman Filter assimilation of satellite derived sea ice temperature. Impact-based forecasting combines a forecast of a weather or climate hazard and an assessment of possible impacts, including when, where and how likely the impacts are. Thus, for a weather forecast to be complete the uncertainty must be quantified. This is typically done through the use of ensembles, whereby many individual forecasts are made for the same forecast period. Alertness has constructed an ensemble from AROME Arctic that takes uncertainty in both the model initial conditions and surface physiography into account: experiments covering the Year of Polar Prediction (YOPP) special observing periods for winter and summer show more accurate forecasts than the leading global ensemble, as well as better quantification of the uncertainty in the forecasts at smaller spatial scales. We are seeing promising results from a new method to estimate the uncertainty in sea surface temperature (SST) based on targeting perturbations to the SST towards areas where the uncertainty is greatest. Work is ongoing into better estimating the uncertainty in model physics (SPP) and model initial conditions (EDA). Alertness cooperates with a multitude of international projects. Of special importance is our contributions to YOPPsiteMIP, a coordinated process-based model evaluation project based on high-frequency multivariate observations at selected Arctic supersites. www.alertness.no gives news and information about the project, as well as open access to data. In May 2020, close to hundred viewers attended a live streaming event on Youtube organised by Alertness on how a weather forecast is made and how researchers work with improving weather forecasts in the Arctic.

Se Resultatrapport

High-impact weather conditions, rapid climate change and limited predictability make the Arctic a challenging operating environment leading to substantial business, societal and environmental risks. ALERTNESS will meet this growing need for reliable and accurate weather predictions by addressing forecast challenges unique to the Arctic: availability and quality of observations, exploitation of satellite observations over snow and ice, model uncertainty due to physics parameterisations, and specific high-impact weather situations. A new set of verification measures appropriate for Arctic will be developed and employed throughout the project to efficiently monitor progress and user needs. ALERTNESS will take advantage of several unique opportunities arising during the Year of Polar Prediction (YOPP) to tackle long-standing issues in atmospheric models in polar environments: advance atmospheric and sea ice analysis in the Arctic using more and new observation types, optimize observation usage and implement state-of-the-art analysis techniques; enhance the representation of Arctic processes including a revised approach to heat flux parameterisation; and improve understanding of error compensation between different processes. Furthermore, ALERTNESS will explore new ways to diagnose uncertainties evolving from representations of small-scale processes, and is set to generate substantial advances in probabilistic forecasting for the Arctic. ALERTNESS builds directly on existing user and stakeholder mechanisms and will enable stakeholders to, for instance, use the advances of probabilistic forecasting for the benefit of safer and more efficient operations in the Arctic. Our tight collaboration between academia and the operational environment at MET Norway will efficiently transfer the results from research to operations, creating a lasting legacy in Arctic weather prediction capacity.