Calculate model performance metrics

Calculates model performace metrics from the netcdf file generated by running `run_ensemble`

Usage

calc_fit(
  ncdf,
  list = NULL,
  model,
  var = "temp",
  dim = "model",
  dim_index = 1,
  qualfun = qual_fun,
  avfun = "mean"
)

Arguments

ncdf

Path to the netcdf file generated by running `run_ensemble`

list

Alternatively to `ncdf` a list of siimulated variables, as returned by `run_ensemble()` when argument `return_list = TRUE`

model

Vector of models for which to calculate the performance measures

var

Variable for which to calculate the performance measures. Defaults to "temp".

dim

character; NetCDF dimensions to extract. Must be either "member" or "model". Defaults to "model". Only used if using the netCDF file. Currently only works with "model".

dim_index

numeric; Index of dimension chosen to extract from. Defaults to 1. Only used if using the netCDF file.

qualfun

Function to calculate the performance measures. Per default calculates root mean suqared error (rmse), Nash-Shutcliff efficiency (nse), Pearson correlation (r), bias (bias), mean absolute error (mae), normalized mean absolute error (nmae), and bias. Can be any function that takes observed data as first, and simulated data at the same time and depth as the second argument.

avfun

Name of the function to calculate the ensemble average, defaults to "mean"

Author

Johannes Feldbauer

Examples

if (FALSE) { # \dontrun{
# using standard quality measures
calc_fit(ncdf = "output/ensemble_output.nc",
                   model = c("FLake", "GLM",  "GOTM", "Simstrat", "MyLake"),
                   var = "temp")
# using own performance measure
 calc_fit(ncdf = "output/ensemble_output.nc",
                  model = c("FLake", "GLM",  "GOTM", "Simstrat", "MyLake"),
                  var = "temp", qualfun = function(O, S) mean(O - S, na.rm = TRUE))
} # }