Smoothing data from one TDL valve

Tool for applying a smoothing function to the time series corresponding to measurements from a single valve in a tunable diode laser (TDL) data set.

Usage

smooth_tdl_data(
    tdl_exdf,
    column_to_be_smoothed,
    valve_column_name,
    valve_number,
    smoothing_function
  )

Arguments

tdl_exdf

An exdf object representing data from a TDL data logger.

column_to_be_smoothed

The name of the column in tdl_exdf that contains the data to be smoothed; typically, this is 'Conc12C_Avg' or 'Conc12C_Avg'.

valve_column_name

The name of the column in tdl_exdf that contains the valve number; typically, this is 'valve_number'.

valve_number

The value of the valve_column_name column that indicates the valve to be smoothed.

smoothing_function

A function that accepts two vectors Y and X (in that order) and returns a smoothed version of Y(X); typically, smoothing_function is based on smooth.spline or a filter from the signal package.

Details

The output from a TDL is highly sensitive to electronic and atmospheric noise, and it is often helpful to smooth the data from one or more valves before attempting to apply calibration corrections or determine the content of an unknown gas mixture. smooth_tdl_data is a convenience function that extracts a time series corresponding to data from one valve, applies a smoothing operation, and replaces the original data in tdl_exdf with the smoothed version. The smoothing function is user-supplied to allow more flexbility.

In addition to the column_to_be_smoothed and valve_column_name columns, the tdl_exdf must also contain an 'elapsed_time' column, which is typically created by a call to identify_tdl_cycles.

Value

An exdf object based on tdl_exdf, where the time series of column_to_be_smoothed vs. 'elapsed_time' has been replaced by a smoothed version obtained by applying the smoothing_function.

Examples

# Example 1: Smoothing the 12C signal from one TDL valve using a spline fit
tdl_file <- read_gasex_file(
  PhotoGEA_example_file_path('tdl_sampling_1.dat'),
  'TIMESTAMP'
)

tdl_file <- identify_tdl_cycles(
  tdl_file,
  valve_column_name = 'valve_number',
  cycle_start_valve = 20,
  expected_cycle_length_minutes = 2.7,
  expected_cycle_num_valves = 9,
  timestamp_colname = 'TIMESTAMP'
)

spline_smoothing_function <- function(Y, X) {
    ss <- smooth.spline(X, Y)
    return(ss$y)
}

spline_smoothed_tdl_file <- smooth_tdl_data(
  tdl_file, 'Conc12C_Avg', 'valve_number', 20, spline_smoothing_function
)

# Example 2: Smoothing the same data with a Butterworth filter (as an
# alternative to the spline_smoothing_function above)

if (FALSE) { # \dontrun{

butterworth_smoothing_function <- function(Y, X) {
  # Create a low-pass Butterworth filter
  lpf <- signal::butter(1, 0.25, type = "low")

  # Apply it to the Y data
  signal::filter(lpf, Y)
}

butterworth_smoothed_tdl_file <- smooth_tdl_data(
  tdl_file, 'Conc12C_Avg', 'valve_number', 20, butterworth_smoothing_function
)
} # }