Provides a rough estimate on the light loss and corresponding range of scale values to consider during optimization. A percentile parameter can be provided to remove outliers. It is recommended to first use `st_screen_twl()` to remove poor quality days.
stk_calibrate(
df,
percentile = 100,
floor = 1.5,
distribution = FALSE,
verbose = TRUE
)Arguments
- df
a skytrackr dataframe
- percentile
percentile of the diurnal data (abvoe the floor value) used in calculating daily statistics (default = 100)
- floor
threshold to remove low (nighttime) values
- distribution
provide the full distribution across the track of scale factors (default = FALSE, providing only a range as a global constraint)
- verbose
Give detailed feedback (TRUE or FALSE, default = TRUE)
Value
An estimated range of scale values to be used in optimization. Values are not log transformed.
Examples
# Estimate the upper scale value for fitting routine
upper_scale_value <- cc876 |> stk_calibrate()
#>
#> ── Estimating suggested scale values ───────────────────────────────────────────
#>
#> ! Upper bound is lower than the light model's normal maximum range (10).
#> ℹ Suggesting 10 instead of the calculated 7.
#> → The suggested scale range is c(0.9,10)!
#> ℹ Note, these estimates are approximations! Always inspect your data (e.g. by
#> using stk_filter()) and consider using the stk_screen_twl() to remove days
#> with poor twilight and other characteristics. When using the 'individual'
#> light model you might have to relax the upper scaling value.