Convective DA

p1-9 January 21 15:10-16:10

UK regional hybrid variational data assimilation developments

Adam Clayton (Met Office), Gordon Inverarity (Met Office) and Marco Milan (Met Office)


Following the introduction of hybrid 4DVar variational data assimilation to the Met Office's global model in July 2011 and hourly 4DVar to the UK model in July 2017, work is now underway to implement hybrid 4DVar in the UK model. The Met Office has run a global ensemble MOGREPS-G, which currently uses an ensemble transform Kalman filter, operationally since June 2005 and a UK ensemble MOGREPS-UK since November 2016. The latter downscales from the MOGREPS-G members and runs forecasts on the UK domain with a 2.2 km grid in the central part of the domain compared to the deterministic UK forecast's 1.5 km central grid. MOGREPS-UK large-scale perturbations are centred around the high-resolution, deterministic 4DVar analysis. To construct the ensemble perturbations for hybrid 4DVar, we are comparing use of forecasts from both MOGREPS-G and MOGREPS-UK. We are also running experiments to see whether the training data used to construct the static covariance component are better processed using the options normally used to transform and localise the ensemble data. Initial results of this work will be presented.

Contact information

Dr. Gordon Inverarity Met Office