Graham et al. (2026) The Met Office Unified Model Global Atmosphere 8.0 and JULES Global Land 9.0 configurations

Identification

• Journal: Geoscientific model development
• Year: 2026
• Date: 2026-02-20
• Authors: Tim Graham, Melissa Brooks, Andrew Bushell, Paul Earnshaw, Samantha Smith, Lorenzo Tomassini, Martin Best, Ian Boutle, Jennifer Brooke, John M. Edwards, Andrew D. Elvidge, Wuhu Feng, Catherine Hardacre, Andrew James Hartley, Alan J. Hewitt, Ben Johnson, ADRIAN C. LOCK, Andy Malcolm, Jane P. Mulcahy, Eike Müller, Ian A. Renfrew, Heather Rumbold, G. G. Rooney, Alistair Sellar, Masashi Ujiie, Irina Sandu, Andy Wiltshire, Michael Whitall
• DOI: 10.5194/gmd-19-1473-2026

Research Groups

• Met Office, FitzRoy Road, Exeter, Devon, EX1 3PB, United Kingdom
• School of Environmental Sciences, University of East Anglia, Norwich, NR4 7TJ, United Kingdom
• Department of Mathematical Sciences, University of Bath, Bath, BA2 7AY, United Kingdom
• Japanese Meteorological Agency, Tokyo, Japan
• National Oceanography Center, European Way, Southampton, SO14 3ZH, United Kingdom (now at)
• School of Physical and Chemical Sciences, University of Canterbury, Canterbury, New Zealand (now at)
• ECMWF, Shinfield Park, Reading, RG2 9AX, United Kingdom (now at)

Short Summary

This paper describes the Global Atmosphere 8.0 and Global Land 9.0 (GA8GL9) configurations of the Met Office Unified Model and JULES land surface model, detailing their scientific advancements over previous versions (GA7GL7) and evaluating their improved performance across weather and climate timescales. GA8GL9 demonstrates reduced errors, enhanced spatial structure in numerical weather prediction, and improved mean climate, particularly in top-of-atmosphere outgoing shortwave radiation.

Objective

• To provide a standalone scientific description of the Met Office Unified Model Global Atmosphere 8.0 and JULES Global Land 9.0 (GA8GL9) configurations.
• To detail the differences and improvements of GA8GL9 compared to its predecessor, GA7GL7, across both numerical weather prediction (NWP) and climate timescales.

Study Configuration

• Spatial Scale:
  • Horizontal: Global configurations with nominal resolutions ranging from 135 km (N96) to 10 km (N1280). Specific tests conducted at N96 (135 km), N216 (60 km), N320 (40 km), N640 (20 km), and N1280 (10 km).
  • Vertical: 85 levels with a model lid at 85 km above sea level (L85) for climate configurations, and 70 levels with a model lid at 80 km (L70) for NWP configurations. Consistent tropospheric vertical resolution.
• Temporal Scale:
  • Climate Simulations: 20-year and 27-year atmosphere/land-only AMIP (Atmospheric Model Intercomparison Project) simulations.
  • NWP Simulations: 3-month long atmosphere/land-only Data Assimilation (DA) trials (September to November 2019) and 5-day N320 case studies.
  • Model Time Steps: Ranging from 20.0 minutes (N96) to 4.0 minutes (N1280). Radiation calculations performed hourly. Convection scheme sub-stepped with two sequential calls per model time step.

Methodology and Data

• Models used:
  • Met Office Unified Model (UM) with ENDGame dynamical core.
  • Joint UK Land Environment Simulator (JULES) land surface model.
  • SOCRATES (Suite Of Community RAdiative Transfer codes based on Edwards and Slingo) radiative transfer scheme.
  • UK Chemistry and Aerosol (UKCA) code for prognostic aerosols and activation.
  • GLOMAP-mode (Global Model of Aerosol Processes) aerosol scheme.
  • CLASSIC (Coupled Large-scale Aerosol Simulator for Studies in Climate) aerosol scheme (for climatological aerosols and mineral dust).
  • TRIP (Total Runoff Integrating Pathways) river routing model.
  • SKEB2 (Stochastic Kinetic Energy Backscatter scheme version 2) and SPT (Stochastic Perturbation of Tendencies scheme) for stochastic physics.
  • COARE (Coupled Ocean–Atmosphere Response Experiment) 4.0 for ocean surface momentum roughness.
  • TOPMODEL (TOPography-based rainfall-runoff MODEL) for sub-grid soil moisture heterogeneity.
• Data sources:
  • Orography: GMTED (Global Multi-resolution Terrain Elevation Data) and RAMP2 (Radarsat Antarctic Mapping Project Digital Elevation Model).
  • Land Use: ESA CCI (European Space Agency’s Land Cover Climate Change Initiative) land-cover dataset, IGBP (International Geosphere-Biosphere Programme).
  • Soil Properties: HWSD (Harmonized World Soil Database), STATSGO, ISRIC-WISE.
  • Vegetation: MODIS (Moderate Resolution Imaging Spectroradiometer) Collection 5 4 km LAI (Leaf Area Index) data, IGBP plant canopy height.
  • Albedo: MODIS (bare soil albedo), GlobAlbedo (snow-free surface albedo).
  • Ocean: GlobColour (sea surface chlorophyll content), Lana et al. (2011) (DMS seawater concentration).
  • Aerosol Emissions: CEDS-CMIP6 (anthropogenic), GFED-CMIP6 (biomass burning), Dentener et al. (2006) (volcanic SO2), UKCA-tropospheric chemistry simulations (aerosol precursor oxidants).
  • Stratospheric Aerosols: CMIP6 forcing of Thomason et al. (2018), Cusack et al. (1998) climatology.
  • River Paths: Oki and Sud (1998).
  • Observational/Reanalysis Data for Evaluation:
    • CERES-EBAF (Clouds and Earth’s Radiant Energy System Energy Balanced and Filled) Edition 4.1 for radiative fluxes.
    • MODIS Edition 2.6 for cloud droplet effective radius.
    • GPCP (Global Precipitation Climatology Project) version 3.2 for precipitation.
    • GPM IMERG (Global Precipitation Measurement Integrated Multi-satellitE Retrievals for GPM) V06B for diurnal cycle of precipitation.
    • ERA-interim reanalyses for various atmospheric fields (temperature, winds, humidity).
    • Radiosondes for tropical temperature forecasts.
    • HadSLP2 (Hadley Centre Sea Level Pressure) for mean sea level pressure.
    • SSMI (Special Sensor Microwave/Imager) for precipitable water.
    • CRUTEM3 (Climatic Research Unit Temperature) for 1.5 m temperature.
    • AERONET (Aerosol Robotic Network) for aerosol optical depth.
    • FLUXNET sites for land surface observations.

Main Results

• NWP Performance: GA8GL9 shows reduced errors (1.3% against observations, 1.5% against own analyses) and improved spatial structure in forecasts compared to GA7GL7, particularly in fine-scale features of precipitation and radiative fluxes.
• Climate Performance: The mean climate in GA8GL9 is improved relative to GA7GL7, with notable enhancements in top-of-atmosphere outgoing shortwave radiation, especially at 60° S due to improved mixed-phase processes.
• Tropical Biases: Significantly reduced tropical temperature biases, with mean biases below the tropopause less than 0.3 K at day 5 forecasts. The tropical tropopause layer (TTL) temperature and humidity biases become largely independent of resolution.
• Convection:
  • Prognostic-based entrainment improves the instantaneous spatial structure of convection and associated fields (precipitation, cloud condensate, outgoing longwave and shortwave radiation).
  • Time-smoothed convective increments and an additional termination condition for convection reduce the frequency of spurious vertically propagating gravity waves in the lower stratosphere and mitigate moist biases.
  • Reduced excessive amplitude in the diurnal cycle of precipitation in regions like the Sahel.
• Cloud Microphysics and Radiation:
  • New riming parametrisation increases supercooled water content, leading to beneficial reductions in shortwave flux biases in the Southern Ocean.
  • Improved TKE diagnostic and Liu cloud droplet spectral dispersion increase cloud droplet effective radius over land and Northern Hemisphere oceans, reducing spatial variability of bias.
  • Black Carbon refractive index update and increased resolution of refractive index imaginary part in RADAER reduce low bias in absorption aerosol optical depth.
• Land Surface: A package of land surface changes (updated orography, soil roughness, vegetation roughness, land use dataset, canopy radiation, distributed form drag, canopy snow for broad-leaved trees) removes the need for an aggregate surface tile in NWP applications and improves near-surface fields.
• Numerical Stability: The new multigrid solver produces smoother solutions at the poles. Improved computational stability in the Gregory-Kershaw convective momentum transport (CMT) prevents numerical noise in wind increments. Increased non-linear solver term for unstable boundary layers reduces failure mechanisms.
• Hydrological Cycle: The overactive hydrological cycle seen in GA7GL7 is reduced in GA8GL9, with a clear reduction in extratropical precipitation.

Contributions

• Consolidation and Integration: GA8GL9 consolidates scientific advancements from both climate-specific (GA7.1GL7.1) and NWP-specific (GA7.2GL8.1) branch configurations into a single, unified configuration.
• Enhanced Convection Scheme: Introduction of prognostic-based entrainment adds "memory" to the convection scheme, allowing for a wider, more realistic range of entrainment rates and improving the spatial structure of convective processes.
• Improved Convection-Dynamics Coupling: Time-smoothed convective increments and an additional termination condition for convection significantly reduce unphysical intermittency and spurious gravity wave generation, leading to more robust and resolution-independent tropical tropopause layer biases.
• Advanced Microphysics: A new riming parametrisation, based on aircraft observations, and prevention of riming for small liquid droplets, physically improves the representation of mixed-phase clouds, leading to reduced shortwave radiation biases in the Southern Ocean.
• Comprehensive Land Surface Updates: A package of land surface changes, including updated orography, spatially varying soil roughness, explicit vegetation roughness lengths, and an updated land use dataset, eliminates the need for an aggregate surface tile in NWP, improving consistency and accuracy.
• Numerical Efficiency and Stability: Implementation of a new multigrid solver for the Helmholtz problem enhances computational efficiency and produces smoother solutions. Improvements to the computational stability of the convective momentum transport scheme prevent numerical noise.
• Reduced Resolution Sensitivity: More realistic treatments of melting snow and freezing rain, along with boundary layer scheme changes, reduce the sensitivity of model processes to vertical resolution, facilitating future model development.
• Operational Readiness: GA8GL9, in its coupled form (GC4), has been adopted as the operational global NWP model at the Met Office, demonstrating its robustness and superior performance.

Funding

• Met Office Hadley Centre Climate Programme (funded by DSIT)
• NERC ACCACIA grant (NE/I028297/1)
• NERC Iceland Greenland Seas Project grant (NE/N009754/1)
• Japan Meteorological Agency (agreement number P014013)

Citation

@article{Graham2026Met,
  author = {Graham, Tim and Brooks, Melissa and Bushell, Andrew and Earnshaw, Paul and Smith, Samantha and Tomassini, Lorenzo and Best, Martin and Boutle, Ian and Brooke, Jennifer and Edwards, John M. and Elvidge, Andrew D. and Feng, Wuhu and Hardacre, Catherine and Hartley, Andrew James and Hewitt, Alan J. and Johnson, Ben and LOCK, ADRIAN C. and Malcolm, Andy and Mulcahy, Jane P. and Müller, Eike and Renfrew, Ian A. and Rumbold, Heather and Rooney, G. G. and Sellar, Alistair and Ujiie, Masashi and Sandu, Irina and Wiltshire, Andy and Whitall, Michael},
  title = {The Met Office Unified Model Global Atmosphere 8.0 and JULES Global Land 9.0 configurations},
  journal = {Geoscientific model development},
  year = {2026},
  doi = {10.5194/gmd-19-1473-2026},
  url = {https://doi.org/10.5194/gmd-19-1473-2026}
}