Hourly 4D‐Var in the Met Office UKV operational forecast model. (30th January 2020)

• Record Type: Journal Article
• Authors: Milan, Marco; Macpherson, Bruce; Tubbs, Robert; Dow, Gareth; Inverarity, Gordon; Mittermaier, Marion; Halloran, Gemma; Kelly, Graeme; Li, Dingmin; Maycock, Adam; Payne, Tim; Piccolo, Chiara; Stewart, Laura; Wlasak, Marek
• Is Part Of: Quarterly journal of the Royal Meteorological Society. Volume 146:Number 728(2020)
• Page Start: 1281
• Page End: 1301
• Publication Date: 2020-01-30
• Subjects: data assimilation -- Met Office -- UKV -- 3D‐Var -- 4D‐Var; Meteorology -- Periodicals; 551.5
• DOI: 10.1002/qj.3737
• Languages: English
• ISSNs: 0035-9009

Abstract

Abstract: Hourly cycling four‐dimensional variational data assimilation (4D‐Var) was implemented operationally in the Met Office's convective‐scale UKV forecast model in July 2017, replacing the previous three‐hourly cycling three‐dimensional 3D‐Var scheme. The new system was based on a previous Nowcasting Demonstration Project (NDP), developed and run in real time over a southern UK domain for the London 2012 Olympic and Paralympic Games and focusing on precipitation forecasts. The new operational system extends this capability to the full UK (and surrounding) area, delivering outputs suitable for blending into nowcasting products and for general forecasting of a full range of meteorological variables. We describe the general formulation of the Met Office 4D‐Var system and some particular ingredients. Differences between the new system and its NDP and UK 3D‐Var antecedents are discussed for both the assimilation algorithm and the observational inputs. As an illustration of the impact on forecast performance, we compare the skill of 3D‐Var and 4D‐Var using an hourly cycle for both configurations. For precipitation skill, we also compare three‐hourly 3D‐Var and hourly 4D‐Var with a reference nowcasting system, in order to highlight the improvement from the new method at very short forecast ranges. Future avenues for developing the system are outlined. Abstract : In this paper we describe the hourly cycling four‐dimensional variational data assimilation (4D‐Var) implemented operationally in the Met Office's convective‐scale UKV forecast model. The new operational system extends this capability to the full UK and surrounding area (shown in the figure), delivering outputs suitable for blending into nowcasting products and for general forecasting of a full range of meteorological variables.