shrink_by_hour: Shrink a dataframe with predictions at integer hours to one...

In edwardlavender/fvcom.tbx: An R Toolbox for the Unstructured Grid Finite Volume Community Ocean Model (FVCOM)

Shrink a dataframe with predictions at integer hours to one with interpolated predictions at non-integer hours

Description

FVCOM predictions are resolved hourly. For predictions at non-integer hours, expand_by_hour expands a dataframe with non-integer hours to include both surrounding integer hours, the predictions of which can be obtained with extract. This function is designed to 'shrink' the expanded dataframe back down to its original size by interpolating predictions between hours to generate a single prediction for each original timestamp.

Usage

shrink_by_hour(dat, verbose = TRUE)

Arguments

dat

A dataframe that contains FVCOM predictions (see expand_by_hour and extract). At a minimum, this should contain the following columns: 'hour_dbl', a numeric vector that defines the time of day in hours; 'hour', an integer vector that defines the hour for which predictions have been extracted; 'index_hour', an integer vector that defines the index in the FVCOM arrays for each hour; order', an integer vector that defines each unique, original row (prior to dataframe expansion by expand_by_hour); and 'wc', an numeric vector that defines the prediction extracted from FVCOM array(s).

verbose

A logical input that defines whether or not to print messages to the console to monitor function progress.

Value

The function returns a dataframe, as inputted, but in which any predictions for non-integer hours have been interpolated from the predictions derived at the two surrounding hours. The 'hour' and 'index_hour' columns are dropped to avoid confusion.

Author(s)

Edward Lavender & Janneke Ransijn

See Also

interp_btw_hours linearly interpolates predictions between hours.

Examples

#### Step (1): Define a dataframe that defines the FVCOM predictions required
# Times, including non integer hours
timestamp <- as.POSIXct(c("2016-03-01 00:00:00",
                          "2016-03-02 00:02:00",
                          "2016-03-01 00:00:00"))
dat <- data.frame(timestamp = timestamp)
# Define columns required to extract FVCOM predictions (see fvcom.tbx::extract())
dat$hour_dbl <- lubridate::hour(dat$timestamp) +
                lubridate::minute(dat$timestamp)/60 +
                lubridate::second(dat$timestamp)/3600
dat$date_name <- date_name(dat$timestamp)
dat$mesh_ID <- as.numeric(as.character(dat_mesh_around_nodes$ID))[1]
# Examine dat
dat

#### Step (2): Implement expand_by_hour()
# ... to define integer hours at which to extract predictions
dat_exp <- expand_by_hour(dat)
# Examine dat_exp
dat_exp

#### Step (3): Extract predictions
# Define matches (see ?fvcom.tbx::extract)
match_hour <- data.frame(hour = 0:1, index = 1:2)
match_mesh <- data.frame(mesh = dat_nodexy$node_id, index = 1:length(dat_nodexy$node_id))
# Define path
path <- system.file("WeStCOMS_files/tidal_elevation",
                     package = "fvcom.tbx", mustWork = TRUE)
                     path <- paste0(path, "/")
# Extract predictions
dat_with_wc <- extract(dat = dat_exp,
                       match_hour = match_hour,
                       match_mesh = match_mesh,
                       dir2load = path)
# Examine dat_with_wc
dat_with_wc

#### Step (4): Implement shrink_by_hour() to interpolate predictions
dat_with_wc_interp <- shrink_by_hour(dat = dat_with_wc)
# Examine interpolated dataframe
dat_with_wc_interp
# Check interpolated value with interp_btw_hours()
interp_btw_hours(x = dat_with_wc$hour_dbl[2],
                 h1 = dat_with_wc$hour[2],
                 h2 = dat_with_wc$hour[3],
                 p1 = dat_with_wc$wc[2],
                 p2 = dat_with_wc$wc[3]
                 )

edwardlavender/fvcom.tbx documentation built on Nov. 26, 2022, 10:28 p.m.