ctdcal.flagging.outliers¶

ctdcal.flagging.outliers(data, old_flags=None, flag_good=2, flag_outlier=4, n_sigma1=2, n_sigma2=20, ignore_nan=True)[source]¶

Flag extreme outliers using standard deviations from the mean as a threshold.

Outliers are identified over two passes. For the first pass, mean and standard deviation of data are calculated for all data. Values more than n_sigma1 standard deviations from mean are (temporarily) flagged questionable. For the second pass, mean and standard deviation are re-calculated with questionable data excluded. Data more than n_sigma2 standard deviations from mean are flagged as outliers.

Parameters

data (array-like) – Variable to be flagged
old_flags (array-like, optional) – Original data flags to be merged in (if provided)
flag_good (int, optional) – Flag value for good data
flag_outlier (int, optional) – Flag value for outliers
n_sigma1 (int, optional) – Number of standard deviations away from mean needed to be excluded from statistics
n_sigma2 (int, optional) – Number of standard deviations away from mean needed to be outlier
ignore_nan (bool, optional) – Ignore nan values in data

Returns

flags – Flag for each data point in input

Return type

array-like

Notes

Functionality is similar to Sea-Bird’s “Wild Edit” in Seasoft V2.