The Model Evaluation Tools (MET): Recent Additions and Enhancements

The Model Evaluation Tools (MET) is an internationally recognized software package used for the verification of Numerical Weather Prediction (NWP). The Developmental Testbed Center (DTC) provides community support for MET which forms the core of the DTC testing and evaluation of innovations for Research-to-Operations transition. MET implements, extends, and standardizes the verification methodologies developed at multiple institutions, including the National Oceanic and Atmospheric Administration (NOAA) operational centers (e.g. EMC, NHC, WPC, SPC, OPC) and labs (e.g. ESRL, NSSL), along with the United States Air Force, and the Met Office. The goal is to make verification reproducible across institutions. MET has been engineered to be highly configurable and spatially and temporally agnostic. It has been applied to NWP model output from decadal climate simulations down to convection allowing mesoscale runs. Since the first public release, many capabilities have been added to MET beyond traditional statistics including object-based, neighborhood, and tropical cyclone verification methods.

The National Oceanic and Atmospheric Administration (NOAA) has chosen to unify their verification efforts into a single system. This system must handle a variety of spatial scales and evaluate individual earth system component models (including atmosphere, atmospheric composition, land, ocean, ice and waves) and the entire earth system model, including coupling among system components and linkages with assimilation of observations. MET was chosen as the core of this unified verification system and python wrappers have been developed to automate it. This larger system, METplus, includes the core MET tools, a results database (METdb), aggregation tools (METcalcpy), visualization tools (METplotpy, METviewer, METexpress), and the automation wrapper scripts.

In this talk, I will highlight enhancements to MET from the last two major releases, review current development priorities, and discuss how users can more closely engage with our development team.