Skip to contents

Set the model output temporal frequency

Source: R/set_output_time_step.R
set_output_time_step.Rd

A thin, generic wrapper around set_time() for the common case of only wanting to change how often every model in the ensemble writes output. The frequency applies to all models: build_glm(), build_gotm() and build_simstrat() each translate output_time_step into their native output-cadence setting (nsave, output.yaml time_step, and Output/Times respectively), and the internal aeme_time_axis() reconstructs the matching output time axis when the results are read back.

Usage

set_output_time_step(aeme, frequency, daily_mean = FALSE)

Arguments

aeme

An Aeme object.

frequency

The desired output frequency. Either a single positive number of seconds, or a character string:

  • a named frequency: "subdaily"/"hourly" (3600 s), "daily" (86400 s), "weekly" (604800 s);

  • a "<n> <unit>" string, e.g. "6 hours", "30 min", "1 day", "900 sec" (units: sec/min/hour/day/week, singular or plural).

daily_mean

logical; if TRUE, also produce a daily-mean output stream (see Details). Default FALSE.

Value

The aeme object with time(aeme)$output_time_step (and, when supplied, time(aeme)$output_daily_mean) updated. output_time_step must be greater than or equal to time(aeme)$time_step; set_time() raises aeme_error_output_time_step otherwise.

Details

AEME does not temporally disaggregate forcing, so requesting output more frequent than daily is only meaningful when the meteorological (and, where relevant, inflow/outflow) forcing is supplied at least as often. See vignette("hourly-vs-daily-met").

Daily-mean output

daily_mean = TRUE makes every model additionally produce a daily-mean output stream alongside the raw frequency output, so a run can be done at a sub-daily cadence (for accuracy, or to average out a diurnal cycle) while the results compared against daily observations are true daily means rather than instantaneous snapshots. It is modelled on GOTM's native output_daily stream (time_method: mean, one record per day):

  • GOTM-WET writes the daily means itself (its output_daily.nc).

  • GLM-AED and Simstrat have no native time-averaging, so AEME averages their sub-daily output.nc by calendar day after the run and writes a companion output_daily.nc. The raw sub-daily output is kept.

The daily stream stores only the targeted variables — the model_controls simulate set plus the fixed set AEME's readers require — not every model variable.

See also

set_time()

Examples

aeme <- readRDS(system.file("extdata/aeme.rds", package = "AEME"))
aeme <- set_output_time_step(aeme, "hourly")
time(aeme)$output_time_step
#> [1] 3600
aeme <- set_output_time_step(aeme, "6 hours")
time(aeme)$output_time_step
#> [1] 21600
aeme <- set_output_time_step(aeme, "hourly", daily_mean = TRUE)
time(aeme)$output_daily_mean
#> [1] TRUE