Home / Study with us / MPAS Tutorials
Large dark blue and grey storm cloud above a ship on the sea

Study the MPAS model

Dates: October 21-25, 2024
Location: Edinburgh, UK

The Model for Prediction Across Scales – Atmosphere (MPAS-A) is next-generation numerical atmospheric model that employs state-of-the-art numerics and software engineering, while incorporating physical parameterization schemes from the WRF-ARW model.

MPAS-A solves the compressible non-hydrostatic equations, supporting applications from global scales to the explicit simulation of clouds, and it uses horizontally unstructured meshes that permit continuous mesh refinement in both global and regional configurations.

Leading scientists from the US National Center for Atmospheric Research (NSF-NCAR) will introduce you to MPAS-A and MPAS-JEDI.

Attendees have the opportunity, as part of enrolling on the course, to participate in a one-day workshop for WRF and MPAS users and developers.

How to apply

Applications opening 2024

Subscribe to our Reach for the Skies mailing list to be kept in the loop about upcoming courses.

Learning outcomes

Specific topics to be covered include:


  • Installation and set-up of MPAS-A modelling system components
  • Global and regional simulation configuration
  • Uniform- and variable-resolution configuration
  • Real-data and idealised simulation initialization
  • The MPAS-A dynamical solver and numerical methods
  • Physical parameterization schemes in MPAS-A
  • The MPAS-A software framework and computational infrastructure
  • Post-processing and visualisation tools

  • MPAS-JEDI data assimilation software build and test
  • Deterministic and ensemble data assimilation algorithms and configurations
  • Preprocessing and assimilation settings of observations
  • Global and regional applications
  • Prerequisites

    The course is aimed at new or beginner users of MPAS-A and MPAS-JEDI.

    Basic knowledge of atmospheric science and numerical modelling, as well as experience working in a Unix computing environment are required.

    Although prior experience with the WRF-ARW model is not necessary, this course will include material to aid modellers in transitioning work from WRF- ARW to MPAS-A.

    Teaching staff

    The course is led by leading scientists from the US National Center for Atmospheric Research:

  • Bill Scamarock
  • Jo Klemp
  • Michael Duda
  • Ming Chen
  • Laura Fowler
  • Jake Liu
  • Course schedule

    MPAS-A Tutorial: Monday morning – Wednesday lunchtime (2.5 days)

    MPAS-JEDI Tutorial: Wednesday afternoon – Thursday afternoon (1.5 days)

    The course will combine lectures by staff from the US National Center for Atmospheric
    Research with hands-on practical sessions, where participants will gain experience building,
    configuring and running MPAS-A and MPAS-JEDI.

    MPAS/WRF Workshop: Friday (1 day)

    The half-day workshop provides a forum for both users and developers to present their work with either the WRF or MPAS modelling system, and to discuss issues related to the model development and applications.

    Course Meal: Wednesday evening

    All attendees will be invited to a dinner.

    Keynote Talks

    Atmospheric model configurations for NWP and climate: Vertical resolution and model filters

    Bill Skamarock, NSF-NCAR/MMM

    The atmospheric component of the global Model for Prediction Across Scales (MPAS-A) was developed for high-resolution climate and NWP applications, and its dynamical core employs a lower-order finite-volume approach similar to many research and operational models. In a 2019 published study using MPAS-A, we found that at current NWP resolutions (i.e. dx ~ 15 km) vertical mesh spacing of 200 meters or less is needed for convergence of the kinetic energy spectrum and to resolve critical flow features in the free troposphere and lower stratosphere.  Subsequent to that study, we discovered that we can configure MPAS-A with less horizontal dissipation when higher vertical resolution is employed, thus increasing the effective vertical and horizontal resolution.  With the increase of the effective resolution, the higher vertical resolution configurations are more efficient than their lower vertical resolution counterparts, and the results argue for moving to higher vertical resolution in many of our applications even if horizontal resolution must be lowered to fit within existing computing limits.  The physical phenomena driving these results are atmospheric fronts and their realization in the simulations, and these results have some implications for dynamics underlying the mesoscale kinetic energy spectrum and model physics. In this talk we will outline our vertical resolution results and supporting simulation evidence, and discuss implications for model configurations including mesh spacing and dissipation mechanisms.

    Modifying the Height-Based Vertical Coordinate in MPAS to Permit a Constant Pressure Upper Boundary for Geospace Applications

    Joseph Klemp and William Skamarock, NSF-NCAR/MMM

    For applications extending into the thermosphere, realistic simulations may require an upper boundary that permits vertical expansion/contraction of the atmosphere in response to strong internal heating/cooling rates. In adapting MPAS for geospace applications, a modification of the height-based vertical coordinate is presented that relaxes the rigid-lid constraint and permits the coordinate surfaces at upper levels to transition toward a constant pressure surface at the model’s upper boundary. This modification is conceptually similar to a terrain-following coordinate at low levels, but now modifies the coordinate surfaces at upper levels to conform to a constant pressure surface at the model top. Since this surface is evolving in time, the height of the upper boundary is adaptively adjusted to follow a designated constant pressure upper surface. This alteration in the original height-based vertical coordinate employed in MPAS requires only minor modifications to the numerics and little additional computational expense. The viability of this modified vertical coordinate formulation is verified in a 2-D prototype of MPAS for an idealized case of upper-level diurnal heating.

    Intercomparison of Regional-MPAS and WRF

    Ming Chen, May Wong, Bill Skamarock, Wei Wang, NSF-NCAR/MMM

    The Model for Prediction Across Scales (MPAS) is a global nonhydrostatic weather and climate prediction system developed at the National Center for Atmospheric Research (NCAR) in partnership with Los Alamos National Laboratory (COSIM). The MPAS version 7, released in June 2019, has added the capability to run MPAS for regional domains with specified lateral boundary conditions. It also encompasses a subset of WRF physics schemes that have been well tested and widely applied for a variety of applications. To document the viability of this regional implementation of MPAS, we have conducted similarly configured WRF and MPAS model simulations over extended retrospective time periods, and applied standard verification techniques to evaluate these retrospective forecasts. Results for the two models are compared with a focus on important meteorological properties such as wind, temperature, humidity and precipitation. The main goal of this study is to test and evaluate the performance of MPAS for medium-range weather forecasting based on comparison with WRF. With quasi-uniform horizontal meshes in MPAS and WRF, it is to be expected that regional simulations employing the same model physics should produce very similar results. The verification statistics presented here provide this confirmation. Statistically, the spatial patterns, diurnal variations and time series of surface air temperature and precipitation between regional MPAS and WRF simulations agree very well. Despite some slight differences, vertical profiles of biases and root mean square errors of wind speed, temperature and moisture content demonstrate the same pattern between WRF and regional MPAS simulations. The circulation patterns simulated by WRF and regional MPAS are almost identical in the mid- and upper troposphere. This study has also been constructive in that the intercomparisons exposed several inconsistencies in both MPAS and WRF that have now been corrected in the models. The use of variable resolution meshes in MPAS may provide additional benefit for regional simulations. With this capability, high resolution in the interior of the regional domain can be relaxed near the lateral boundaries to grids comparable to the resolution of the boundary forcing data. This provides more flexibility in maintaining consistency between the lateral boundary conditions and the interior, and may remove the need for multiply nested domains. As the next step in evaluating the performance of MPAS in regional applications, we will extend the above verification study to include results from a variable resolution implementation of the regional MPAS.

    Dependence of deep convection schemes on horizontal grid-spacing in MPAS: Difference in formulation and impact on forecasts

    Laura D. Fowler, Mesoscale and Microscale Meteorology Laboratory, National Center for Atmospheric Research, Boulder, Colorado, USA
    Mary C. Barth, Atmospheric Chemistry Observations and Modeling Laboratory National Center for Atmospheric Research, Boulder, Colorado, USA.

    Scale-aware parameterizations of deep convection allow parameterized subgrid-scale deep convection to be gradually replaced by dynamically resolved convection as horizontal grid-spacing decreases. Using uniform and variable-resolution meshes, we compare 30-day simulations run with the convective parameterization developed by Grell and Freitas (2014) and the multi- scale version of the Kain-Fritsch convection parameterization (Alapati et al. 2014). Results focus on the impact of the scale dependence of the cloud base mass flux on the global distribution of convective and grid-scale precipitation.


    There is a charge to attend the MPAS Tutorial and MPAS/WRF Workshop. This charge contributes to the cost of teaching, facilities, refreshments and all activities. You will be asked to pay for your place on the course via the University of Leeds online store once you have been offered a place.

    Full Academic Rate (tutorial & workshop) to be announced

    Educational Discounted Rate (tutorial & workshop) to be announced

    May be available to PhD students or post-doctoral researchers not eligible for the subsidised rate.

    NCAS Subsidised Rate (tutorial & workshop) to be announced

    UK & EU students attending UK universities or students supervised by NCAS staff

    Once you have been offered a place on this course, you will be asked to pay the course fees via the University of Leeds online store.
    The course fee includes all tuition, activities and materials. Refreshments including lunch and a conference dinner will be provided.

    Travel and accommodation

    Accommodation is not provided as part of the course fee and attendees should arrange their own accommodation.

    Attendees are responsible for arranging and paying for their own transport to and from the course.

    Terms and conditions


    Attendees are required to pay in full by credit/debit card or provide a valid purchase order from their home organisation by the confirmation deadline. All payments and refunds are processed by the University of Leeds.

    Cancellations/ refunds

    Cancellations of places on courses are subject to a £50 administration fee where the cancellation is made more than six weeks before the date of the training course.
    The remaining balance of the course fee will be refunded.

    No refund is payable for a cancellation made within six weeks immediately prior to the training course – in this event the full course fee will remain payable.
    In the event that a place can be filled by another student then the full fee (less the £50 administration fee) will be refunded once the replacement student has paid in full.

    All cancellations must be sent in writing to training@ncas.ac.uk.

    Amendments and cancellations by NCAS

    It may be necessary for NCAS to change the content and timing of a course, the staff, the date or the venue. In the unlikely event of the course being cancelled by NCAS, a full refund will be made. For all bookings, the liability of NCAS/University of Leeds shall be limited to the amount of the fee actually paid to NCAS/University of Leeds by the delegate.