Time-integrated covariate — integrate

Accumulates a covariate over preceding time steps at each location. A survey does not sample the food available that instant so much as the food that has built up since the season began, which is what the original pipeline's int_chl captured for chlorophyll.

Usage

integrate_covariate(env_dat, vars = NULL, window = "year", suffix = "_int")

Arguments

env_dat: an sf POINT object with one row per location and time step, as datamatch's access functions return
vars: covariate columns to integrate; NULL does all of them
window: "year" to accumulate from the start of each calendar year, "all" to accumulate over the whole record, or a positive integer for a rolling window of that many time steps
suffix: suffix for the new columns

Value

env_dat with an integrated column per covariate

Details

The default window = "year" reproduces int_chl of Ross et al. (2023), which integrated chlorophyll from January: a running sum from January of each year, reset at the year boundary. A numeric window instead sums over that many trailing time steps, giving a rolling total that does not reset.

References

Ross C, Runge J, Roberts J, Brady D, Tupper B, Record N (2023). Estimating North Atlantic right whale prey based on Calanus finmarchicus thresholds. Marine Ecology Progress Series 703, 1-16. doi:10.3354/meps14204

Examples

if (FALSE) { # \dontrun{
env <- integrate_covariate(env, "CHL")               # int_chl, Ross et al. 2023
env <- integrate_covariate(env, "CHL", window = 3)   # trailing 3-step total
} # }