In ianjonsen/duckConfit: State-space filtering code for RAATD data paper

This short vignette documents the data preparation and SSM-filtering steps used for the RAATD data. The workflow is illustrated using the royal penguin (**) data as an example.

Data preparation

Load example royal penguin data and apply prefilter() to prepare data for state-space filtering. The function prints information about the number of tracks discarded and location flagged to be ignored by the state-space filter

require(dplyr)
require(duckConfit)
data(rope)

pfd <- prefilter(
  dat = rope,
  metadata = meta,
  sp = "ROPE",
  min_obs = 30,
  min_days = 5,
  vmax = 10
  )

State-space filtering

Fit state-space model to pre-filtered data pfd and use dplyr::do() to apply sfilter() to each individual track. Use redo_sfilter() to attempt to recover cases where the SSM failed to converge. redo_sfilter() will try tries = n times before giving up. Output is a tibble grouped by individual id.

ssm_by_id <- pfd %>%
  do(ssm = sfilter(., span = 0.4, nu = 5))

ssm_by_id <- redo_sfilter(ssm_by_id, pfd, tries = 10)

Generating Quality Control Plots

A .pdf file of plots to aid quality control decisions are generated using qc_plot() with 1 individual track per page. The file is saved to the working directory by default, this can be changed by specifying the file path via the fpath argument. Three plots are printed per page: (1) a map of the estimated track (blue points and line) overlaid on the data passed to the state-space filter (red points); (2) a plot of estimated longitudes (blue) overlaid on the observed longitudes (red); (3) a plot of estimated latitudes (blue) overlaid on the observation latitudes (red). In all cases, the plots are scaled to the estimated track for ease of visualisation - some observed data may occur outside the plotted range.