distsamp is aerial by default. A shipboard survey differs in three ways
that matter, and this is where all three are set.
Usage
platform_defaults(platform = c("aerial", "vessel", "ship"))Value
A list with platform, kind (the PLATFORM_KIND level to keep),
effort_args (for narwcr::flag_effort()), and distance_sources (for
sighting_distances()).
What changes with the platform
- Effort criteria
The handbook's altitude ceiling (8.A.1) is an aircraft criterion. It cannot be widened to suit a ship: a missing value fails a criterion, so a record with no
ALTfails at any ceiling, andmax_alt_m = Infdoes not help. The vessel defaults drop the criterion instead (narwcr::flag_effort(max_alt_m = NULL)). Sea state and visibility apply to both platforms and are unchanged.The effect depends on the file, so measure it. Some programmes leave
ALTempty on vessel records, and every such record is excluded. Others write the observer's eye height there, which passes an aircraft's ceiling and hides the problem until the empty records go missing. On one real extract of 1,037,180 vessel records, 338,562 had noALT— 32% of the vessel effort, and 277 of 741 eligible sightings.- Right-angle distance sources
"angle"is \(h/\tan\theta\) from an aircraft altitude, and the"strip"code books are the Skymaster's and the AT-11's (8.A.31). Neither exists on a vessel, so the vessel default is"exact"alone: a sighting position projected onto the track, which does not depend on the platform. Seesighting_distances().- Which records are kept
narwcr::classify_platform()labels each stretch from its median speed. The platform names the label to keep.
What does not change
LEGTYPE 5 and 6 mark shipboard records (8.A.21), but they mark a platform
of opportunity (POP): a ship that recorded what it saw while going
somewhere. That is not a designed line-transect. Counting it as effort would
put unsearched track into the denominator of a density estimate. The vessel
defaults therefore keep legtype_on_effort = 2, a designated census line,
as the aerial ones do. A dedicated shipboard line-transect survey codes its
lines LEGTYPE = 2 and needs nothing here. A POP file is not a distance
sampling dataset.
Segmentation itself does not depend on the platform. Great-circle distance between consecutive fixes, splitting at breaks in effort, and cutting a track into lengths work the same for a ship as for a plane.
Naming
"vessel" is the documented spelling, because it is the PLATFORM_KIND
level the filter compares against. "ship" is accepted everywhere "vessel" is, and resolves to
it.
Examples
platform_defaults("aerial")
#> $platform
#> [1] "aerial"
#>
#> $kind
#> [1] "aerial"
#>
#> $effort_args
#> list()
#>
#> $distance_sources
#> [1] "angle" "exact" "strip"
#>
# A vessel drops the altitude ceiling and loses the aircraft distance
# sources
platform_defaults("vessel")
#> $platform
#> [1] "vessel"
#>
#> $kind
#> [1] "vessel"
#>
#> $effort_args
#> $effort_args$max_alt_m
#> NULL
#>
#>
#> $distance_sources
#> [1] "exact"
#>
# "ship" is the same thing
identical(platform_defaults("ship"), platform_defaults("vessel"))
#> [1] TRUE