Look at a NARWC extract on a map

Opens an interactive map of a survey extract: where the platform went, what it saw, and when. Sightings, effort and acoustic stations are drawn together over a chosen time period, and the record behind any point is one click away.

Usage

run_narwc_app(dat = NULL, pam = NULL, ...)

Arguments

dat: A data frame of NARWC survey data, ideally from read_narwc(), or a path — or anything else narwc_fetch() resolves — to read one from. NULL (the default) opens the shipped example, and a file can be loaded from inside the app.
pam: Acoustic detections: one row per station and day, carrying a station, a position, a date, a species, whether it was detected, and how long the recorder was listening. A data frame or a path to a CSV; column names are matched loosely and what matched is shown on the app's Source tab. NULL (the default) draws no acoustic layer.
...: Passed to shiny::runApp(), which is where launch.browser, port and host go.

Value

The value of shiny::runApp(), invisibly.

What it draws

Sightings — every record naming a species, coloured by species or by data type, sized by group size, with a popup giving the record and the handbook's meaning for each of its codes rather than the bare number.
Effort — the track, split into the part that met flag_effort()'s criteria and the part that did not. The criteria are controls: moving the Beaufort cutoff and watching the on-effort track shrink is the quickest way to see what a threshold costs.
Depth contours — isobaths at chosen depths, cut from a bathymetry grid. Off until asked for, because the first draw fetches one.
Acoustic stations — a fixed recorder with a detection rate, from a pam table. Recording effort is the denominator, because a station that heard whales on four of five days it listened is not the station that heard them on four of two hundred.

Depth contours

Isobaths are cut from an ETOPO grid rather than read off a basemap tile, so a line on the map is a depth that can be named — this is the 100 m contour, that is the 200 m — where a tinted tile is only a picture of one. Which depths are drawn is a control, because the isobath a distribution is read against is the reader's question and not this app's.

The grid is fetched by marmap::getNOAA.bathy() the first time contours are asked for, which is why they start switched off: an app that reaches a server the moment it opens is not what the rest of this package does. It is cached on disk afterwards, under tools::R_user_dir("narwcr", "cache") by default and following marmap's own filename convention — so a grid already downloaded by anything else in this stack is read rather than fetched, and one downloaded here serves them back. Point the cache wherever the rest of the stack keeps its own:

options(narwcr.cache = tools::R_user_dir("datamatch", which = "cache"))

Once a grid is loaded, a sighting's popup also reports the depth under it, from the nearest grid cell. Nothing is fetched for a popup's sake: the depth appears when the contours do, and is absent rather than guessed at until then.

Aerial, vessel, opportunistic, PAM

The first three arrive in one extract and are told apart by LEGTYPE (handbook 8.A.21): codes 0–4 are line-transect aerial, 5–6 shipboard, and 7 and 9 aerial platforms of opportunity. Where LEGTYPE says nothing the platform is inferred from how fast it was moving, via classify_platform(), and the app reports which records were read and which were inferred — those are not the same claim.

The handbook has no code for a dedicated shipboard survey, so every shipboard record in the archive is a platform of opportunity. Where a programme reads its own extract differently, the mapping is one option:

options(narwcr.legtype_types = c("5" = "opportunistic", "6" = "opportunistic"))

options(narwcr.species_labels = c(RIDO = "Risso's dolphin")) extends the species names the popups show, the same way. Neither ever replaces a code: what was recorded is always on screen beside what it means.

Preparation runs once, over everything

make_leg_id() and fill_legstage() read a record's neighbours, so running them on a table already filtered to one year would split every line crossing the boundary and strip the state the first record of January inherited from December. The app runs the whole pipeline on the whole table when it loads, and the time controls filter what that produced.

Examples

if (FALSE) { # \dontrun{
# The shipped example
run_narwc_app()

# A real extract, read first so the reading report is on the console
dat <- read_narwc("extract.csv")
run_narwc_app(dat)

# With acoustic detections beside it
run_narwc_app(dat, pam = "detections.csv")
} # }