proc_dendro_L2: Clean Time-aligned Dendrometer Data
In treenet/treenetproc: Clean, Process and Visualize Dendrometer Data

Description Usage Arguments Details Value References Examples

proc_dendro_L2 cleans time-aligned (L1) dendrometer data by removing outliers and correcting for erroneous jumps or shifts in the data.

proc_dendro_L2(
  dendro_L1,
  temp_L1 = NULL,
  tol_out = 10,
  tol_jump = 50,
  lowtemp = 5,
  frost_thr = 5,
  interpol = NULL,
  frag_len = NULL,
  plot = TRUE,
  plot_period = "full",
  plot_show = "all",
  plot_export = TRUE,
  plot_name = "proc_L2_plot",
  iter_clean = 1,
  tz = "UTC"
)

`dendro_L1`	`data.frame` with time-aligned dendrometer data. Output of function `proc_L1`.
`temp_L1`	`data.frame` with time-aligned temperature data. Output of function `proc_L1` (see Details for further information).
`tol_out`	numeric, defines the rigidity of the threshold above or below which a value is classified as an outlier. Lower values increase the rigidity (see Details for further information).
`tol_jump`	numeric, defines the rigidity of the threshold above or below which a value is flagged for jump correction. Lower values increase the rigidity (see Details for further information).
`lowtemp`	numeric, specifies the temperature in °C below which frost shrinkage or expansion is expected. Default value is set to `5°C` due to hysteresis shortly before or after frost events.
`frost_thr`	numeric, increases the thresholds for outlier and jump detection in periods of probable frost (i.e. temperature < `lowtemp`). The thresholds are multiplied by the value provided.
`interpol`	numeric, length of gaps (in minutes) for which values are linearly interpolated after data cleaning. Set `interpol = 0` to disable gapfilling. If `interpol = NULL` the default value is set to `interpol = 2.1 * reso` (i.e. two timestamps).
`frag_len`	numeric, specifies the length of data fragments occurring in-between missing data that are automatically deleted during data cleaning. This can be helpful to remove short fragments of erroneous data within a period of missing data, i.e. after jumps. If `frag_len = NULL` the default value is set to `frag_len = 2.1`.
`plot`	logical, specify whether the changes that occurred during data cleaning should be plotted.
`plot_period`	character, specify whether plots should be displayed for the whole period (`plot_period = "full"`), separately for each year (`plot_period = "yearly"`) or for each month (`plot_period = "monthly"`).
`plot_show`	character, specify whether all periods should be plotted (`plot_show = "all"`) or only those periods in which `L1` dendrometer data changed during data cleaning (`plot_show = "diff"`).
`plot_export`	logical, specifies whether the plots are exported as a `PDF` file to the working directory or are plotted in the console.
`plot_name`	character, specify name of the PDF to which the plots are saved.
`iter_clean`	numeric, specifies the number of times the cleaning process is repeated. Can be used to check whether running the cleaning process multiple times has an effect on the results. In most cases, a single iteration is sufficient.
`tz`	specify the desired time zone. Default is `"UTC"`.

Time-aligned temperature data temp_L1 is used to define periods in which frost shrinkage is probable, e.g. when the temperature is below lowtemp. Without temperature data, frost shrinkages may be classified as outliers. For more details and an example see the following vignette: vignette("Introduction-to-treenetproc", package = "treenetproc").

Temperature data can also be provided along with dendrometer data. In this case, the name of the temperature series has to contain the string "temp". In case no temperature dataset is specified, a sample temperature dataset will be used with a warning. The sample temperature dataset assigns permanent frost to the three months December, January and February.

Outliers and jumps are identified when exceeding a lower or upper threshold. Thresholds are obtained on the basis of density distributions of differences between neighbouring data points. The rigidity of the thresholds can be controlled with the arguments tol_jump and tol_out. For more information on the calculation of the thresholds the user is referred to Knüsel et al. (2020, in prep).

The function returns a data.frame with processed dendrometer data containing the following columns:

`series`	name of the dendrometer series.
`ts`	timestamp with format `%Y-%m-%d %H:%M:%S`.
`value`	dendrometer value (`µm`).
`max`	highest measured value up to this timestamp (`µm`).
`twd`	tree water deficit (`µm`), i.e. the amount of stem shrinkage expressed as the difference between `max` and `value`.
`gro_yr`	growth since the beginning of the year (`µm`). Also calculated for years with missing data.
`frost`	indicates frost periods (i.e. periods in which the temperature is below `lowtemp`).
`flags`	character vector specifying the changes that occurred during the processing. For more details see the following vignette: `vignette("Introduction-to-treenetproc", package = "treenetproc")`
`version`	package version number.