MODES: advancing wave-space diagnostics for global weather and climate models

Numerical weather prediction (NWP) and climate model systems are commonly evaluated using first- and second-order moment statistics of prognostic and diagnostic variables, such as winds and vertical velocity. While informative, such diagnostics provide only limited insight into the underlying physical processes, particularly those affected by changes in model dynamics or physical parameterizations. A more process-oriented understanding of atmospheric variability in space and time relies on simplified representations of the governing equations and their linear wave solutions.

 

By projecting the global circulation onto wave solutions of the linearized primitive equations on the sphere, one can obtain dynamical insight into the spatial and temporal variability of Rossby, inertia–gravity (IG), Kelvin, and mixed Rossby–gravity (MRG) modes. This projection is implemented in the MODES software.

 

First released as open-access software in 2015, MODES has been used for real-time diagnostics of ECMWF medium-range forecasts, including energy spectra, balanced and unbalanced circulation and equatorial waves in both physics-based and machine learning–based forecasts (https://modes.cen.uni-hamburg.de). Applications of MODES include studies of extreme events in climate models, trends in subseasonal variability in the tropics and midlatitudes, and interactions between large-scale circulation and regional processes such as convection.

 

This poster presents MODES v2, which introduces a decomposition of vertical velocity and vertical momentum fluxes. The new framework enables a detailed characterization of vertical momentum fluxes associated with Kelvin, Rossby, MRG, and eastward- and westward-propagating IG waves, without imposing spatial scale or frequency cutoffs. We illustrate the capabilities of MODES v2 using operational analysis and reanalysis data from ECMWF, highlighting its potential for diagnosing the processes underlying atmospheric variability.