Percent bias in the midsegment slope of a Flow Duration Curve (FDC)

Computes the percent bias for the midsegment slope of the flow duration curves (FDC) between simulated and observed flows, delimited by exceedance probability (EP) thresholds, useful for evaluating model performance in intermediate flow regimes.

Usage

pbias_fdc_midsegment(
  sim,
  obs,
  lowQ_EP = 0.7,
  highQ_EP = 0.2,
  na.rm = TRUE,
  plot = FALSE,
  ...
)

Arguments

sim

numeric. Simulated flow values.

obs

numeric. Observed flow values.

lowQ_EP

numeric. Upper EP bound of the midsegment (e.g., 0.7 for 70%).

highQ_EP

numeric. Lower EP bound of the midsegment (e.g., 0.2 for 20%).

na.rm

logical. Whether to remove NA values before computation. Default: TRUE.

plot

logical. Whether to generate a log-scale plot highlighting the selected segment. Default: FALSE.

...

additional arguments passed to plot().

Value

numeric. Percent bias for the FDC midsegment slope.

Details

The EP thresholds must satisfy \(1 \ge lowQ\_EP > highQ\_EP \ge 0\).

Percent bias for the midsegment slope is calculated as: $$ pbiasFDC_{\text{midsegment}} = \frac{\big[(S_{m1} - S_{m2}) - (O_{m1} - O_{m2})\big]} {(O_{m1} - O_{m2})} \times 100 $$ where \(S_{m1}\), \(S_{m2}\) and \(O_{m1}\), \(O_{m2}\) are the simulated and observed flows at the EP thresholds defining the segment.

See also

Other Performance evaluation: evaluate_swat(), fdc_signature(), gof_metrics(), pbias_fdc_target_segment(), plot_fdc(), plot_mean_annual_cycle(), plot_timeseries()

Examples

# Synthetic daily flow data
set.seed(123)
obs <- abs(rnorm(730, mean = 50, sd = 20))
sim <- obs * runif(730, min = 0.8, max = 1.5)

# Percent bias for midsegment slope (20–70% EP)
pbias_fdc_midsegment(sim, obs, lowQ_EP = 0.7, highQ_EP = 0.2)

# Plot selected segment
pbias_fdc_midsegment(sim, obs, lowQ_EP = 0.7, highQ_EP = 0.2,
                     plot = TRUE)