Evaluation Declaration Example 3

# An example of an evaluation of ensemble, operational stage forecasts using two types of 
# thresholds: stage thresholds in a CSV file and probability classifier thresholds 
# specified within this configuration. The evaluation is of the 4-day maximums (peak
# stages) between the lead times of 2 days and 6 days.
label: Example 3

# Observations are provided in a PI-timeseries file as 6h stages. 
observed:
  sources:
    - data/CKLN6_STG.xml
  variable:
    name: STG
    label: STAGE

# The forecasts are HEFS provided in a gzipped tarball of PI-timeseries XML files.
predicted:
  sources:
    - data/CKLN6_HEFS_STG_forecasts.tgz
  variable:
    name: SSTG
    label: STAGE
  label: HEFS

# The feature to evaluate, specified by NWS Handbook-5 ID, which differs between 
# the "observed" and "predicted" sources.
features:
  - observed: CKLN6
    predicted: CKLN6SUQ

# The measurement unit of the evaluation.
unit: FT

# The evaluation is restricted to a lead hour that ends at 144 hours, or 4 days. The
# time_scale of the evaluation is the 4 day maximum that ends at 4 days and all those 
# pairs are placed in a pool that is zero hours wide and centered on 144 hours
lead_times:
  minimum: 144
  maximum: 144
  unit: hours
time_scale:
  function: maximum
  period: 96
  unit: hours

# The lead time pool, one big pool that is zero hours wide.
lead_time_pools:
  period: 0
  unit: hours

# The stage thresholds are spedcified in a CSV file at the location specified.
threshold_sources:
  uri: thresholds/scenario801/thresholds.csv
  type: value

# The classifier thresholds are defined here and indicate different
# probabilities for which an ensemble forecast is considered to forecast 
# exceedance of a stage threshold.
classifier_thresholds: [0.1, 0.3, 0.5, 0.7, 0.9]

# The metrics to calculate.
metrics:
  - contingency table

# The decimal format to use when writing numeric outputs.
decimal_format: '#0.000000'

# The output formats to write.
output_formats:
  - csv