Run ensemble analysis on finished model runs

Usage

run.ensemble.analysis(
  settings,
  plot.timeseries = NA,
  ensemble.id = NULL,
  variable = NULL,
  start.year = NULL,
  end.year = NULL,
  ...
)

Arguments

settings: PEcAn settings object
plot.timeseries: if TRUE plots a modeled timeseries of target variable(s) with CIs
ensemble.id: database ID, taken from settings if not specified
variable: variable name to process, taken from settings if not specified
start.year, end.year: taken from settings if not specified
...: additional arguments passed to ensemble.ts

Value

Nothing (called for side effects). Creates ensemble plots as PDF files and optionally saves time series analysis results.

Details

Upstream contract (reads from settings$outdir):

ensemble.output.<var>.<years>.<id>.Rdata: Produced by get.results. Contains ensemble.output: a numeric vector of model output values, one per ensemble member.

File-based side effects (saved to settings$outdir):

ensemble.analysis.<var>.<years>.<id>.pdf: Histogram and boxplot of ensemble output distribution.
ensemble.ts.<var>.<years>.<id>.pdf: (Optional) Time series plot with confidence intervals. Generated when plot.timeseries is not NA.
ensemble.ts.analysis.<var>.<years>.<id>.Rdata: (Optional) Contains ensemble.ts.analysis: time series summary statistics. Saved when plot.timeseries is not NA.

Note: This is a terminal step — nothing downstream loads these files programmatically. The results are consumed by visualization or user inspection.

Author

David LeBauer, Shawn Serbin, Ryan Kelly