Code related to calculating time series trends

This repo will potentially contain both Python and R versions of the algorithms but both implementations should run the same series of tests against the test data.

Input and output data (gridded and station) is NetCDF4 (do we want to support NetCDF3 or NetCDF4 classic?) and CF 1.6 (http://cfconventions.org/cf-conventions/v1.6.0/cf-conventions.html) compliant. Station data in NetCDF is supported under CF1.6 descrete sampling geometries (http://cfconventions.org/cf-conventions/v1.6.0/cf-conventions.html#discrete-sampling-geometries) and orthogonal multi-dimensional array representation is used.

In the meantime, the repo contains a set of (crudely cobbled together) functions that can be used for trend analysis and derivation of trend significance level and confidence interval of individual time series, as described below:

v1.0 created by P.Wolski September 2016

functions.py contains functions: test_autocorr() trend_CI() get_TheilSen()

trend.py contains functions: get_linear() get_TheilSen() get_quantreg()

in functions.py -lowess in trend_CI() crashes if time series starts with NAs -there is no implementation of significance level for Durbin-Watson test for autocorrelation -test_autocorr() and get_TheilSen() missing structured docstrings -example datasets and wrapper functions would be nice to have

in trend.py

• all functions give only analytical pval
• no correction for autocorrelation