Details of the forecast model Met Office, UK GloSea4 is an ensemble seasonal forecasting system which has been running at the Met Office since September 2009 and is based on the coupled ocean-atmosphere model, HadGEM3_AO_r1.1. This model is part of the HadGEM3 family of models, a group of models sharing the same physical and dynamical schemes at different resolutions, which can be used for prediction across a range of time and space scales HadGEM3_AO_r1.1 consists of the following components: UM (Met Office Unified Model atmosphere; NEMO (Nucleus for European Modelling of the Ocean) ocean; CICE (Los Alamos Sea-ice Model) sea-ice; and MOSES (Met Office Surface Exchange Scheme) land surface. The model resolution used in GloSea4 is 1.25 degrees latitude by 1.875 degrees longitude (N96) and 38 levels in the vertical for the atmosphere; 1 degree with 1/3 of a degree refinement between 20S and 20N (ORCA1 grid) and 42 levels in the vertical for the ocean. Uncertainties in initial conditions are represented via a lagged-initialisation approach, with batches of ensembles run from initial conditions which are updated weekly. All simulations initialised on a particular start date have the same initial conditions but evolve differently due to the stochastic physics schemes used to represent model uncertainties. The RP (Random Parameters) scheme aims to represent the structural uncertainty arising from subjective parameters in physical parameterizations. SKEB2 (Stochastic Kinetic Energy Backscatter version2.0) aims to represent the sub-grid scale uncertainty arising from advection and numerical dissipation. The GloSea4 forecast system has 14 members per week (initialised on Mondays) and the operational forecast starting at the beginning of the month combines the 3 weeks centred around that time. Hindcasts are initialised on fixed dates (1st, 9th, 17th and 25th of each month) and have 3 members per start-date. The skill assessments submitted here for nominal monthly start dates (denoted 'month') are based on aggregates of ensembles from three start dates: 25th 'month'-1 and 1st and 9th 'month', for the period 1989-2002. Results have been submitted initially for 2m temperature and precipitation (although other parameters will be verified in due course). Global grid-point diagnostics are available for 3-month rolling seasons corresponding to months 2-4, 3-5 and 4-6; measures aggregated over the Tropics, the Northern extratropics and the Southern extratropics are available for conventional seasons DJF, MAM, JJA and SON. The verification datasets used were the ERA-Interim Reanalysis data set for 2m temperature and the GPCP (NASA) observations dataset for precipitation. Verification was carried out on the atmospheric (N96) grid. References: To be provided.