Skip to contents

Mixed-layer depth of one CTD cast (threshold criterion)

Source: R/ctd_derived.R
ctd_mld.Rd

The first depth below ref_depth where the profile departs from its value at ref_depth by threshold, linearly interpolated between the two bracketing samples (CalCOFI/workflows#101). The default is the provider's definition, the CalCOFI legacy one (Rasmus Swalethorp, 2026-09-23, questions.csv Q01, adopted 2026-10-01): "MLD is the depth at which sigma-theta is 0.02 kg/m3 greater than at a reference depth of 10 m". The criterion stays an argument: the alternatives published beside it are the classic 0.125 kg m^-3 (Levitus 1982; Monterey & Levitus 1997) and the temperature criterion (|Delta T| >= 0.2 deg C, de Boyer Montegut et al. 2004), which needs no salinity, so it is the only one a preliminary_without_bottle cast can have. The 0.03 kg m^-3 default of calcofi4db 4.16.0-4.17.x (de Boyer Montegut et al. 2004) is superseded and only reachable as threshold = 0.03.

Usage

ctd_mld(
  depth,
  value,
  qual = NA,
  criterion = c("sigma_theta", "temperature"),
  threshold = NULL,
  ref_depth = 10
)

Arguments

depth

depth (m), positive down; one cast.

value

sigma-theta (kg m^-3) or temperature (deg C), per criterion.

qual

quality codes of value (NA = good).

criterion

"sigma_theta" (value must increase by threshold) or "temperature" (absolute departure of threshold).

threshold

the departure that ends the mixed layer; default 0.02 for sigma-theta (the provider's definition), 0.2 for temperature.

ref_depth

reference depth (m), default 10.

Value

one-row tibble: mld_m, ref_value, depth_max_m (deepest good sample), status (ok, mixed_to_bottom, no_reference, no_data), criterion, threshold, ref_depth.

Details

The provider's answer is silent on the edge cases, so these are unchanged from 4.16.0: the value at ref_depth is interpolated from the samples either side of it (it is NA, status no_reference, when the cast does not bracket it, e.g. a cast starting below 10 m); a cast that never crosses the threshold has mld_m = NA and status mixed_to_bottom (its deepest sample is reported, so a consumer can say "deeper than"). Flagged (8/9), missing and non-finite samples are dropped first, so a NaN bin is a gap the interpolation spans.

Examples

z <- 0:100
ctd_mld(z, ifelse(z < 40, 25, 25.5))           # step at 40 m, the provider's 0.02
#> # A tibble: 1 × 7
#>   mld_m ref_value depth_max_m status criterion   threshold ref_depth
#>   <dbl>     <dbl>       <dbl> <chr>  <chr>           <dbl>     <dbl>
#> 1  39.0        25         100 ok     sigma_theta      0.02        10
ctd_mld(z, ifelse(z < 40, 25, 25.5), threshold = 0.125)
#> # A tibble: 1 × 7
#>   mld_m ref_value depth_max_m status criterion   threshold ref_depth
#>   <dbl>     <dbl>       <dbl> <chr>  <chr>           <dbl>     <dbl>
#> 1  39.2        25         100 ok     sigma_theta     0.125        10