Skip to main content

Package to study BSRN data and their quality control

Project description

PyBsrnQC

A library for BSRN Quality Control (QC)

Repository github with the source code : https://github.com/LE2P/PyBsrnQC

Installation

Using pip

pip install pybsrnqc

Quick use

Put your solar data in a directory (dataset for example). Data is registered in CSV files, there is a value each minute.

The data has to have at least the following columns in the header :

timestamp global2_avg direct_avg diffuse_avg downward_avg temperature
2019-08-01 00:00:00 145.1 665.300 69.340 365.65 25

Example :

Data under the form YYYYMM_brut.csv with a csv file per month gathered in a directory dataset.

- dataset/
  - 201906_brut.csv
  - 201907_brut.csv
  - 201908_brut.csv
  - ...

You can then study this data :

  • Calculate the BSRN coefficient of a certain Quality Control chosen
  • Create the flagged data file associated with your datas and the coefficients selected
  • Visualize the datas

BSRN coefficient calculation

Use the coef_calculator module.

from pybsrnqc import coef_calculator as cc 

name_coef, coef = cc.compute('./dataset')

You can then load your coefficient to your configuration file. Thus, your automatic control will take into account the new specific coefficient calculated.

from pybsrnqc import config

config.load(name_coef, coef)

If your QC have a maximum and a minimum limit (QC5 for instance) you can use the following code :

from pybsrnqc import coef_calculator as cc 
from pybsrnqc import config

name_coef, coef, name_coef_min, coef_min = cc.compute('./dataset')
config.load(name_coef, coef, name_coef_min, coef_min)

Quality Control tool

Use the automaticQC module in order to generate the flagged data :

from pybsrnqc.automaticQC import generateQCFiles

generateQCFiles('./dataset/201908_brut.csv')

Be careful : when you generate your files, it is in the same directory than you brut data. Don't forget to move your generated files if you want to continue to study your brut data.

If you want to visualize the QC of your data and your data use :

  • plotQCFiles

    from pybsrnqc.automaticQC import plotQCFiles
    
    plotQCFiles('./dataset/201908_brut.csv')
    
  • plotBSRN from visualPlot

     from pybsrnqc.visualPlot import plotBSRN
    
     plotBSRN('./dataset/201908_brut.csv')
    

Visualization tools

You can use other functions of the pybsrnqc packages to plot and visualize the data.

  • Access to data

    Inquire the path of your directory (you can select the time period over which you want to observe the data). If the period isn't specified all the files are opened.

    from pybsrnqc import open_data as od
    
    df = od.open_all('./dataset')
    

    If you want to specify the period, put your data under the form YYYYMM_brut.csv and inquire the period as[YYYYMM, YYYYMM].

     # Example to select data from June to December 2019
    
     df = od.open_all('./dataset',period=['201906','201912'])
    
  • You can plot the data over the time period chosen with the current limits.

    import json 
    import importlib.resources 
    from pybsrnqc import plot_limits as pl
    from pybsrnqc import qc_functions as qcf
    
    # Get data conf from JSON file
    with importlib.resources.path("pybsrnqc", "qcrad_conf.json") as data_path:
    with open(data_path, 'r') as f:
        coefs = json.load(f)
    
    # Plot the limits for the QC chosen (here QC1)
    pl.limit_plot(df, qcf.QC1(), coefs)
    
  • Plotting limits with differents coefficient values

    # Plot the limits for the QC chosen (here QC1)
    pl.multiplot_coef(df, qcf.QC1(), coefs)
    
  • Plotting the 3D histogram of the data

    # Plot the histogram for the QC chosen (here QC1)
    pl.hist_data(df, qcf.QC1())
    

Density analysis

  • Computation of the data density Compute the density of each elements over the dataset given

     # Calculation of the KDE for a dataset
     log_kernel = pl.kde_computing(df, qcf.QC1())
    
  • Time series plotting Zoom on a certain period of time : plot on sza or time with density values.

    # Time series plotting and density
    pl.plot_series_kde(df, log_kernel, QC, begin, end)
    
  • Coefficient indicators Plot the evolution of certain values according to he coefficient Return a dataframe containing the indicators values

    from pybsrnqc import coef_study as cs
     
    # Indicators plotting
    df_var = cs.coef_variation(df, log_kernel, qcf.QC1())
    
  • Getting the coefficient This function give the best coefficient giving a density threshold defined as the outlier limit. Return the score dataframe and the best score with the linked coefficient.

    # Coefficient calculation
    df_score, score = cs.calc_coef(df, log_kernel, qcf.QC1(), threshold=-15)
    

Small dictionary of functions

  • open_data.open_all(path, period=None, select_day=False, select_Zenith=True)

    • path : path of the directory with the data files
    • period : you can choose a period (see above)
    • select_day : if True, select only the hour between 5 AM and 7 PM
    • select_Zenith : if True, select only the SZA between 0° and 90°
  • plot_limits.limit_plot(df, QC, coefs, save=False, level='all', display=True, values=True, fig=True)

    • df : the dataframe studied
    • QC : the QC studied declared thanks to qc_functions
    • coefs : a set of coefficients, imported from qc_rad.json
    • save : if True, the graph is saved
    • level : 'all' (default) to plot all the limits, 'level_2', 'level_1' or 'level_bsrn' for only one of them
    • display : if True, the graph is displayed
    • values : if True, the data elements are plotted with the limits
    • fig : if True, a new figure is created
  • plot_limits.multiplot_coef(df, QC, coefs, level='level_2', coef_values=[0.0, 0.5, 1.2], level_min=None, coef_values_min=None)

    • df : the dataframe studied
    • QC : the QC studied declared thanks to qc_functions
    • coefs : a set of coefficients, imported from qc_rad.json
    • level : 'level_2' (default) to plot the 2nd level limits, 'level_1' or 'level_bsrn' for the others
    • coef_values : the coefficient values you want to plot. Ex: [value1, value2]
    • level_min : if you want to add a lower limit, 'level_2_min' or 'level_1_min'
    • coef_values_min : the coefficient values for the lower limit
  • plot_limits.hist_data(df, QC, dimension='3D')

    • df : the dataframe studied
    • QC : the QC studied declared thanks to qc_functions
    • dimension : '3D'to plot a 3D histogram, '2D' to plot a 2D one
  • plot_limits.kde_computing(df, QC, display=True, coefs=None, limits=False, level='All', log_form=True, save='KDE_result', bw_sel=None)

    • df : the dataframe studied
    • QC : the QC studied declared thanks to qc_functions
    • display : if True, the graph is displayed
    • coefs : a set of coefficients, imported from qc_rad.json
    • limits : if True, limits are plotted on the KDE graph
    • level : 'all' (default) to plot all the limits, 'level_2', 'level_1' or 'level_bsrn' for only one of them
    • log_form : if True, the computed kernel is returned under the log form
    • save : if True, the graph is saved
    • bw_sel : bandwidth selection method. None correspond to the scott method. 'silverman'or others can be inquired.
  • plot_limits.plot_series_kde(df, log_kernel, QC, begin, end, var='timestamp', line=True)

    • df : the dataframe studied
    • log_kernel : the computed KDE under the log form
    • QC : the QC studied declared thanks to qc_functions
    • begin : start date (see above)
    • end : end date (see above)
    • var : if 'timestamp' (default), time is on the abscissa. If 'SZA' it is the zenith angle
    • line : if True, dots are connected by a line
  • coef_study.coef_variation(df, log_kernel, QC, level='level_2', coef_range=[0.0, 1.2], step=0.01, coef_range_min=None, step_min=None, verbose=True, display=True)

    • df : the dataframe studied
    • log_kernel : the computed KDE under the log form
    • QC : the QC studied declared thanks to qc_functions
    • level : 'level_2' (default) to plot the 2nd level limits, 'level_1' for the 1st level
    • coef_range : panel of the coefficients you want to try
    • step : step in the range
    • coef_range_min : panel of the coefficients you want to try for the lower limit
    • step_min : step in the range for the coefficient of the lower limit
    • verbose : if True, waiting point are displayed
    • display : if True, the graph is displayed
  • coef_study.calc_coef(df, log_kernel, QC, threshold, level='level_2', coef_range=[0.0, 1.2], coef_range_min=[0.0, 1.2], step=0.01, step_min=0.1, verbose=True)

    • df : the dataframe studied
    • log_kernel : the computed KDE under the log form
    • QC : the QC studied declared thanks to qc_functions
    • threshold : density threshold chosen for outliers (see above)
    • level : 'level_2' (default) to plot the 2nd level limits, 'level_1' for the 1st level
    • coef_range : panel of the coefficients you want to try
    • step : step in the range
    • coef_range_min : panel of the coefficients you want to try for the lower limit
    • step_min : step in the range for the coefficient of the lower limit
    • verbose : if True, waiting point are displayed
  • coef_calculator.compute(path=None, level='level_2', bw_sel=None)

    • path : path of the directory with the data files
    • level : 'level_2' (default) to plot the 2nd level limits, 'level_1' for the 1st level
    • bw_sel : bandwidth selection method. None correspond to the scott method. 'silverman'or others can be inquired.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

pybsrnqc-0.1.62.tar.gz (22.6 kB view hashes)

Uploaded Source

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page