In this branch, pygeodesy uses the pyre package (https://github.com/pyre/pyre) to handle configuration and documentation of parameters. Similar to the previous version of pygeodesy, all tasks flow through a single application binary called pygeodesy.
> pygeodesy --help
pygeodesy:
pygeodesy 0.0.1
copyright (c) 2016-2019 all rights reserved
Author:
Bryan V. Riel <bryanvriel@gmail.com>
commands:
makedb: Make a time series database
subnet: Subset a time series database file
clean: Clean a time series by removing outliers and removing bad values
filter: Apply low-pass/smoothing filter to time series database
cme: Perform common mode estimation on data
modelfit: Fit temporal model to time series database
detrend: Remove time series model from time series data
elasticnet: Fit temporal model to time series database using Elastic Net regression
plot: Plot component time series for specified stations
netmap: Makes interactive map of station network
velmap: Plot vector map of displacements for network
Each task handled by pygeodesy also comes with its own help explaining all its parameters:
> pygeodesy modelfit --help
pygeodesy:
pygeodesy 0.0.1
copyright (c) 2016-2019 all rights reserved
Author:
Bryan V. Riel <bryanvriel@gmail.com>
pygeodesy modelfit
Fit temporal model to time series database.
usage:
pygeodesy modelfit [command]
where [command] is one of
help: show this help screen
options:
--input: Input time series database [str]
--output: Output time series model database [str]
--user: Python file defining temporal dictionary [str]
--nstd: Number of deviations for outlier threshold (default: 5) [int]
--t0: Starting decimal year of model (default: None) [float]
--tf: Ending decimal year of model (default: None) [float]
--penalty: Regularization parameter for model estimator (default: 1.0) [float]
--output_amp: Filename for optionally saving seasonal amplitudes for each station [str]
--output_phase: Filename for optionally saving seasonal phase for each station [str]
--num_iter: Number of iterations for iterative least squares (default: 1) [int]
--rw_iter: Number of re-weighting iterations for LassoRegression (default: 1) [int]
--solver: Name of solver from [RidgeRegression, LassoRegression] [str]
--scale: Scale observations by factor (default: 1.0) [float]
--special_stats: List of stations to allow for 10x higher std threshold [str]
--std_thresh: Absolute deviation threshold for bad stations (default: 1.0e10) [float]
--min_timespan: Minimum timespan (days) of valid data to keep station (default: 365.0) [float]
--min_valid: Minimum number of valid observations to keep station (default: 100) [int]
--dry: show what would get done without actually doing anything [bool]
Also similar to the previous version of pygeodesy, all parameters for all tasks can be stored in a single configuration file. Here, we use the pfg file format which uses whitespace to demarcate different sections of the pfg file for different tasks. For example, here is a pfg file that specifies parameters for the tasks makedb and subnet:
;
; Configuration file for GPS data
;
pygeodesy:
; Global configuration
global:
; The type of time series data
data_type = gps
; Supported data format?
data_format = sopac
; Make database from time series files
makedb:
; Input time series directory
directory = /Users/briel/data/WNAM
; The column format of files
column_fmt = year: 1, doy: 2, north: 3, east: 4, up: 5, sigma_north: 6, sigma_east: 7, sigma_up: 8
; Output database name
dbname = raw.db
; File containing coordinates of stations
metafile = /Users/briel/data/SopacStack/SopacCoordinates.txt
; The relevant columns in metafile
metafmt = id: 0, lat: 8, lon: 9, elev: 10
; The file extension for time series files
extension = .neu
; Subset database for a given spatial region and temporal window
subnet:
; Input raw database
input = sqlite:///raw.db
; Output subset database
output = sqlite:///data.db
; Polygon coordinates file
poly = poly_slab.txt
; Start time
tstart = 2010-01-01
; End time
tend = 2019-01-01
;station_list = cascadia_stations.txt
If you name your parameter file pygeodesy.pfg, then the pygeodesy application will find the pfg file automatically. If you name your pfg file something else, you just need to pass the name of the file via the --config flag, e.g.:
> pygeodesy makedb --config=myfile.pfg
Parameters can also be specified on the command line to override the parameters in the .pfg file, e.g.:
> pygeodesy modelfit --std_thresh=100.0