Pull a per-time-step index back out as a series

Several functions here compute one value per time step and broadcast it onto every row, so the object keeps its shape and can carry on down a pipe: box_anomaly(), section_transport(), eastern_gom_salinity(), northeast_channel_inflow() and the rest of the region-scale indices all behave this way. That is right for modelling, where the covariate has to line up with the observations, and wrong for almost everything else. Plotting a 22-year monthly index from a broadcast column means plotting each value a few thousand times; writing one out means exporting a file mostly made of repetition.

Usage

index_series(env_dat, vars = NULL)

Arguments

env_dat: an sf POINT object with one row per location and time step, as datamatch's access functions return
vars: index columns to extract, or NULL to take every column that is constant within each time step

Value

a data frame with YEAR, MONTH, DAY and one column per index, one row per time step, ordered by date. Not an sf object: an index has no location

Details

This collapses them back: one row per time step, in date order, as a plain data frame.

What it will not do

A column that varies within a time step is a map, not an index, and collapsing it would silently throw away the spatial pattern and keep an arbitrary one of its values. Naming such a column is an error rather than a quiet mean. If a summary of a map per step is what you want, that is a different operation and an explicit one — take the mean yourself, or use box_anomaly(), which is exactly that with a region attached.

With vars = NULL the constant-within-step columns are found for you, so passing an object through several index functions and then calling this returns all of them at once.

Examples

if (FALSE) { # \dontrun{
env <- eastern_gom_salinity(env)
env <- section_transport(env, from = c(-67.5, 44.5), to = c(-66.0, 43.5))

series <- index_series(env)
plot(with(series, as.Date(paste(YEAR, MONTH, DAY, sep = "-"))),
     series$egom_salinity, type = "l")
} # }