RavenR (version 2.2.0)

rvn_annual_peak_timing_error: Annual Peak Timing Errors

Description

rvn_annual_peak_timing_error creates a plot of the annual observed and simulated peak timing errors, based on the water year.

Usage

rvn_annual_peak_timing_error(
  sim,
  obs,
  mm = 9,
  dd = 30,
  add_line = TRUE,
  add_labels = TRUE
)

Value

returns a list with peak timing errors in a data frame, and a ggplot object

df_peak_timing_error

data frame of the calculated peak timing errors

p1

ggplot object with plotted annual peak errors

Arguments

sim

time series object of simulated flows

obs

time series object of observed flows

mm

month of water year (default 9)

dd

day of water year (default 30)

add_line

optionally adds a 1:1 line to the plot for reference (default TRUE)

add_labels

optionally adds labels for early peak/late peaks on right side axis (default TRUE)

Details

Creates a plot of the peak timing errors in simulated peaks for each water year. The difference in days between the simulated peak and observed peak are plotted (and/or returned in the data frame) for the water year. This diagnostic is useful in determining how accurate the timing of peak predictions is. Note that a large error in the number of days between simulated and observed peaks indicates that the model predicted a larger event at a different time of year, i.e. overestimated a different event or underestimated the actual peak event, relative to the observed flow series.

The sim and obs should be of time series (xts) format and are assumed to be of the same length and time period. The flow series are assumed to be daily flows with units of m3/s. Note that a plot title is purposely omitted in order to allow the automatic generation of plot titles.

The add_labels will add the labels of 'early peak' and 'late peak' to the right hand side axis if set to TRUE. This is useful in interpreting the plots. Note that values in this metric of less than zero indicate an early prediction of the peak, and positive values mean a late prediction of the peak (since the values are calculated as day index of simulated peak - day index of observed peak).

See Also

rvn_annual_peak_event to consider the timing of peak events rvn_annual_peak_event_error to calculate errors in peak events

Examples

# load sample hydrograph data, two years worth of sim/obs
data(rvn_hydrograph_data)
sim <- rvn_hydrograph_data$hyd$Sub36
obs <- rvn_hydrograph_data$hyd$Sub36_obs

# create a plot of peak timing errors with defaults
peak1 <- rvn_annual_peak_timing_error(sim, obs, add_line=TRUE)
peak1$df_peak_timing_error
peak1$p1

# plot directly and without labels
rvn_annual_peak_timing_error(sim, obs, add_line=TRUE, add_labels=FALSE)