MET Stat YAML Configuration

This page describes the structure and fields of the YAML file used to configure the MET Stat module in VCasT. This file controls how .stat files are read, filtered, and processed to extract and aggregate forecast verification statistics.

Example YAML File

Sample MET Stat Configuration YAML
input_stat_folder: /path/to/MET/stats
line_type: cts

date_column: fcst_valid_beg
start_date: "2022-05-01_00:00:00"
end_date: "2022-05-08_00:00:00"

string_filters:  
  model: ["RRFS_GDAS_GF.SPP.SPPT_mem01", "RRFS_GDAS_GF.SPP.SPPT_mem02"]
  fcst_var: ["APCP_03"]
  fcst_thresh: [">0.0"]
  fcst_lead: ['030000', '060000', '090000']

reformat_file: true
output_reformat_file: "/path/to/filtered_output.data"

output_file: true
stat_vars: ["fbias", "gss"]
output_plot_file: "/path/to/output.data"

aggregate: true
group_by: ["fcst_lead"]
output_agg_file: "/path/to/output_agg.data"
ci: true

Configuration Sections

The YAML file is divided into several sections:

Input and Line Type Settings

  • input_stat_folder: Path to the directory containing MET .stat files. All files must be located within a single folder to allow the tool to scan and process them collectively. Recursive search or multiple paths are not supported.

  • line_type: Specifies the MET line type to filter (e.g., cnt, cts, ecnt, etc.). This setting determines which records are extracted and must be compatible with the metrics listed in stat_vars. Only variables valid for the chosen line type will be available.

    For a full list of supported line types and their associated statistics, see the MET User’s Guide - Point-Stat Output Line Types.

Date Settings

  • date_column: Column used to apply the date filter (e.g., fcst_valid_beg).

  • start_date / end_date: Time range to include data for processing (YYYY-MM-DD_HH:MM:SS format).

Filtering Options

  • string_filters: Filter rows by allowed string values in specific columns. Examples include:

    • model: List of models and/or ensemble members, e.g., [ens0, ens1].

    • fcst_var: List of forecast variables, e.g., [REFC, TMP].

    • fcst_lead: List of forecast lead times in HH0000 format, e.g., [010000, 020000].

Output Options

  • reformat_file: [true, false] writes the fully filtered DataFrame to file (with all columns associated to the chosen line type option).

  • output_reformat_file: Path to save the filtered dataset.

  • stat_vars: List of variables to extract (e.g., fbias, gss).

  • output_file: [true, false] exports selected statistical variables.

  • output_plot_file: Path to save the extracted metric values.

Aggregation Options

  • aggregate: [true, false] whether to compute group statistics (mean, CI).

  • group_by: List of column(s) to group by (e.g., [fcst_lead, model]).

  • output_agg_file: Path to save aggregated output.

  • ci: [true, false] whether to compute confidence intervals for each metric.

—

For a working example or template, see the following use cases.