Model Evaluation Tools (MET) is a highly configurable, state-of-the-art suite of verification tools designed for the verification and evaluation of numerical weather forecasts. Although MET was developed at the Developmental Testbed Center (DTC) to be used with the Weather Research and Forecasting (WRF) modeling system, the data formats and standards used allow MET to be run on the output of other modeling systems as well.
The current MET capabilities include traditional verification approaches for standard surface and upper air variables; confidence intervals for most verification measures; spatial forecast verification methods; ensemble verification methods; and tropical cyclone track verification methods. The MET package is designed to be modular and flexible. Each tool can be run individually without running the entire set of tools. New tools can easily be added to the MET package due to this modular design.
Future MET development will take into account the needs of the NWP community - including operational centers and the research and development community.
The following are required to build and run MET:
- C, C++, and Fortran compilers all from the same family (recommend GNU, PGI, or Intel compilers)
- The UNIX make (GNU make) utility
- The HDF5 library is required to support NetCDF4.
- Unidata's NetCDF4 library
- The GNU Scientific Library (GSL)
- NCEP's BUFRLIB for decoding point observations in PrepBufr format
- NCEP's GRIB2 C-Library with JASPER, PNG, and Z libraries for reading GRIB2 files (optional)
- The HDF4 and HDF-EOS2 libraries for using the MODIS-Regrid tool (optional)
- The Cairo and FreeType libraries for using the MODE-Graphics tool (optional)
- Python 2.7 for Python embedding to prepare 2D gridded data fields for reading by the MET tools (optional)
Additional Recommended Software
Although none of the following utilities are required to run MET, they may be helpful in preparing data for use in MET:
- The Unified Post-Processor for users verifying WRF model data
- The copygb utility for regridding GRIB data (also packaged with the Unified Post-Processor)
- The wgrib and wgrib2 utilities to inventory GRIB1 and GRIB2 files.
MET consists of a set of verification tools run from the command line. The current MET tools perform the following verification techniques:
- Prepare point observations for the Point-Stat and Ensemble-Stat tools using the PB2NC, ASCII2NC, and/or MADIS2NC tools.
- Prepare rainfall into desired accumulations for verification using the PCP-Combine tool.
- Compare a gridded forecast to point observations the Point-Stat tool.
- Compare a gridded forecast to a gridded analysis field using the Grid-Stat tool.
- Perform a grid-to-grid comparison over a series of fields, most commonly a time-series, using the Series-Analysis tool.
- Compare an ensemble of gridded forecasts to point and/or gridded observations using the Ensemble-Stat tool.
- Apply an object-based verification approach to a gridded forecast and gridded analysis field using the MODE tool.
- Apply an intensity-scale decomposition verification approach to a gridded forecast and gridded analysis field using the Wavelet-Stattool.
- Aggregate results across many cases using the Stat-Analysis and MODE-Analysis tools.
- Perform verification on ATCF (Automated Tropical Cyclone Forecasting) files using the MET-Tropical Cyclone (MET-TC) tools.
- Dump the contents of binary Gridpoint Statistical Interpolation (GSI) diagnostic files to ASCII format using the MET GSI tools and run the Stat-Analysis tool to derive statistics.
MET Users Documentation consists of a detailed user's guide, slides from previous tutorials, and related white papers. Users are strongly encouraged to download and read the MET Users Guide before they start using MET.