Addressing the causes of large‐scale circulation error in the Met Office Unified Model. (16th May 2020)

• Record Type: Journal Article
• Authors: Williams, K. D.; van Niekerk, A.; Best, M. J.; Lock, A. P.; Brooke, J. K.; Carvalho, M. J.; Derbyshire, S. H.; Dunstan, T. D.; Rumbold, H. S.; Sandu, I.; Sexton, D. M. H.
• Is Part Of: Quarterly journal of the Royal Meteorological Society. Volume 146:Number 731(2020)
• Page Start: 2597
• Page End: 2613
• Publication Date: 2020-05-16
• Subjects: circulation error -- drag partitioning -- model evaluation; Meteorology -- Periodicals; 551.5
• DOI: 10.1002/qj.3807
• Languages: English
• ISSNs: 0035-9009

Abstract

Abstract: The Met Office Unified Model develops a systematic error consisting of a large high‐pressure bias over the North Pole. The error begins to develop early in numerical weather prediction forecasts and is the leading large‐scale circulation error in the model's climatological mean, with implications for principle circulation regimes and blocking. The cause of the error is investigated using a variety of diagnostic techniques, including analysis of a perturbed parameter ensemble for the model, evaluation of changes in both weather and climate simulations to minimise the risk of introducing compensating errors, and detailed comparisons with another model. A reduction in this systematic error is achieved through an overall reduction in the near‐surface drag; however, to accomplish this without affecting forecast predictability adversely, a redistribution of the drag is required between the roughness from vegetation in the land‐surface scheme, the near‐surface drag from the orographic gravity‐wave drag scheme, the turbulent orographic form drag, and the resolved drag (through changing the orographic filtering). A modest reduction in the climatological bias is achieved, along with a reduction in bias and improved predictability of near‐surface winds. Abstract : The Met Office Unified model develops a large high surface pressure bias and associated geopotential height error over the Arctic (figure shows the climatological bias against ERA‐Interim for a) PMSL, b) geopotential height (m)). The cause of the error is investigated and a redistribution of near‐surface drag between parametrization schemes is found to be required