Contenido

turbulence.Graph
turbulence.Graph.time_series
turbulence.Graph.windrose
turbulence.Graph.anomalies_meaning_average
turbulence.Graph.time_series_web
turbulence.Graph.plotly_RidgelinePlot_web
turbulence.Graph.fig_html_web
turbulence.Graph.backup_figure

turbulence.Graph

class turbulence.Graph()

Class using as constructor of the interactive graphics employing bokeh and plotly library; and statics graphics employing matplotlib.pyplot. These graphs are mainly oriented to be used with data from class Sesor but it's possible to use it with other data if this contained in a dataframe object with characteristics given in each method

turbulence.Graph.time_series

turbulence.Graph.time_series(data = None, variables = None, style = 'Bokeh', kind = 'line', ylabel = None, name_graph = None, time_units = None, plot_height = 350, plot_width = 600, line_widths = 2, line_alphas = 1, legend_location = 'top_left', color = None, show_figure = False, save_figure = False)

Generate an interactive graph using the bokeh or static library using matplolib.pyplot. You can use any dataframe, including the turbulent variable information contained in the Sensor object

Parameters:

data: DataFrame containing the time series by columns. The name of index should be 'TIMESTAMP'
variables: Variable as str or List of variables to graph. If more than one variable is passed, they are graphed in the same plane
style: Library to graphic, its options are: 'Bokeh' to interactive graphic with bokeh, or 'plt' to use matplotlib.pyplot
kind: String type. Choose graph type, default is Line, its options are: 'line', 'step'
ylabel: String type. Title of y axis, default is None.
name_graph: String type. Title that will have the graph.
time_units: String type. Time units of the graph
plot_height: Choose the height of the graph, default 200.
plot_width: Choose the width of the graph, default 600.
line_widths: Int or a list with int to select the width of the lines.
line_alphas: Int or a list with int to select the transparency of the lines.
*legend_location: * List of colors. Default is a list of SIATA's colors. The list star with black, and next have SIATA's colors
color: String type. Name used to identify the sensor in the database.
show_figure: Default is False. Option to show or not the figure in the notebook
save_figure: Default is False. Save the figure in the same level of the notebook or script in PDF format. Only to 'plt' style

Returns:

Returns a HTML graph of the given variables using bokeh
Returns the ax object created with matplotlib.pyplot

Examples

Creating random data to graphic


xxxxxxxxxx
import turbulence as tbl
import numpy as np
import pandas as pd
#Dataframe with random data
index = pd.date_range(start='2021-01-01', end='2021-02-15')
columns = list('ABCDE')
df = pd.DataFrame(np.random.rand(len(index),len(columns)), columns=columns, index = index) 
df.index.name = 'TIMESTAMP'
df.head()
                   A         B         C         D         E
TIMESTAMP                                                   
2021-01-01  0.758538  0.324083  0.153662  0.342210  0.498356
2021-01-02  0.044648  0.348006  0.230927  0.135567  0.052285
2021-01-03  0.958152  0.370423  0.969999  0.147159  0.347812
2021-01-04  0.606610  0.151699  0.654082  0.505528  0.183892
2021-01-05  0.327732  0.242570  0.011946  0.083442  0.041856

Interactive graph using bokeh


xxxxxxxxxx
graph = tbl.Graph() #Generating the Graph object
time_series = graph.time_series(data = df, show_figure = True)

series_gif

Static graph. All variables are graphics in the same space


xxxxxxxxxx
time_series = graph.time_series(data = df, style = 'plt', name_graph = 'Sample series', show_figure = True)

series_jpg

turbulence.Graph.windrose

turbulence.Graph.windrose(data = None, nsector = 16, style = 'plotly', lst_color = None, title = None, show_figure = False, save_figure = False)

Create an interactive windrose using plotly library or a static plot using matplotlib.pyplot

Parameters:

data: Dataframe containing two columns with the direction and speed of the wind, an index in datetime format and the names of the columns are "direction" and "speed"
nsector: Number of sectors into which you want to divide the windrose. Default is 16
style: Library that you want to use to make the windrose. Options are 'plotly' for an interactive graph, or 'plt' to use the WindRose library
kind: String type. Choose graph type, default is Line, its options are: 'line', 'step'
lst_color: List of colors in hexadecimal format. SIATA colors are used by default
show_figure: Default is False. Option to show or not the figure in the notebook
save_figure: Default is False. Save the figure in the same level of the notebook or script. Only to 'plt' style

Returns:

If style is 'plt', returns the ax object created with WindRose library
If style is 'plotly', returns the interactive figure created with plotly

Examples

Creating random data to graphic


xxxxxxxxxx
import turbulence as tbl
import numpy as np
import pandas as pd
#Dataframe with random data
index = pd.date_range(start='2021-01-01', end='2021-02-15')
columns = ['direction', 'speed']
df = pd.DataFrame(np.random.rand(len(index),len(columns)), columns=columns, index = index)
df.direction = df.direction * 100
df.speed = df.speed * 10
df.head()
            direction     speed
2021-01-01  21.872307  0.301827
2021-01-02  90.461161  7.083188
2021-01-03  64.340076  7.408397
2021-01-04  26.599929  6.254186
2021-01-05  56.810256  6.578352

Interactive graph using plotly


xxxxxxxxxx
graph = tbl.Graph() #Generating the Graph object
windrose = graph.windrose(data = df, show_figure = True)

windrose_gif

Static graph using WindRose


xxxxxxxxxx
windrose = graph.windrose(data = df, style = 'plt', show_figure = True)

windrose

turbulence.Graph.anomalies_meaning_average

turbulence.Graph.anomalies_meaning_average(data = None, variables = ['LE', 'H', 'TAU'], period = '30T', units = ['W/m²', 'W/m²', 'N/m²'], window = 12, min_periods = 11, plot_width = 800, plot_height = 350, line_width = 4, bar_alpha = 1, style = 'Bokeh', show_figure = False, save_figure = False, path = None)

The database is resampled according to the 'period' parameter. The moving average is then calculated by taking a width window 'window' with a minimum of 'min_periods' data. Anomalies are calculated as: anomaly = (resampled data) - (moving average) The initial configuration is for processed data from EddyPro Parameters:

data: By default it is the data of the object. It must be a DataFrame with the type DatetimeIndex, and columns with variable names, which must have the index name 'TIMESTAMP'
variables: List with variables of interest for anomalies. Default are turbulent fluxes
period: Used for resampling the database to smooth the series. Default is '30T'
units: List of units for the variables of interest
window: Integer with the data number for the mobile window. Default is 12 (Corresponding to 6 hours if period is '30T')
min_periods: Minimum number of data to make the moving average. Default is 11
plot_width: Choose the width of the graph, default 800
plot_height: Choose the height of the graph, default 350
line_widths: Int or a list with int to select the width of the lines
bar_alpha: Int or a list with ints to select the transparency of the bars
style: Library that you want to use to make the figure. Options are 'Bokeh' for an interactive graph, or 'plt' to use matplotlib.pyplot
show_figure: Default is False. Option to show or not the figure in the notebook
save_figure: Default is False. Only to 'plt' style
path: Path to save the figure in PDF format

Returns:

If style is 'Bokeh', then an interactive figure is returned with the anomalies as bars starting from the value of the moving average. Anomalies can be turned off by clicking on the legends. Also, a DataFrame with the data used for the figure is returned
If style is 'plt', then a list of tuples is returned, each tuple has the figure and the ax object of the figure. Also, a DataFrame with the data used for the figure is returned

Examples

Creating random data to graphic


xxxxxxxxxx
import turbulence as tbl
import numpy as np
import pandas as pd
#Dataframe with random data
index = pd.date_range(start='2021-01-01', end='2021-01-04', freq = '5T')
columns = ['LE', 'H', 'TAU']
df = pd.DataFrame(np.random.rand(len(index),len(columns))*100, columns=columns, index = index)
df.index.name = 'TIMESTAMP'
df.head()
                            LE          H        TAU
TIMESTAMP                                           
2021-01-01 00:00:00  73.787154  78.890860  37.334627
2021-01-01 00:05:00  45.476565  97.376636  40.864121
2021-01-01 00:10:00  79.536425  57.533989  44.323393
2021-01-01 00:15:00  72.835109  98.125444  50.045119
2021-01-01 00:20:00  85.315780  81.654879  42.771131

Interactive graph using bokeh


xxxxxxxxxx
graph = tbl.Graph() #Generating the Graph object
mean, anomalies_fig = graph.anomalies_meaning_average(data = df, show_figure = True)

windrose_gif

Static graph using matplotlib.pyplot to show the sensible heat 'H'


xxxxxxxxxx
mean, anomalies_figs = graph.anomalies_meaning_average(data = df, style = 'plt', show_figure = True)

windrose

Using the same data but modify the period of smooth to 20 minutes and taking a window of 3 hours


xxxxxxxxxx
mean, anomalies_figs = graph.anomalies_meaning_average(data = df, style = 'plt', period = '20T', window = 9, min_periods=8, show_figure = True)

windrose

turbulence.Graph.time_series_web

turbulence.Graph.windrose(data = None, style = 'Bokeh', show_figure = False)

This function uses time_series and graph_AllVariables to generate the graphs used on the COMPLEX campaign website. View them at https://siata.gov.co/COMPLEX/Website/ by selecting the "Go to diagnostics" button on each station.

Parameters:

data: By default it is the data of the object. It must be a DataFrame with the type DatetimeIndex, and columns with variable names, which must have the index name 'TIMESTAMP'
style: Library that you want to use to make the figure. Options are 'Bokeh' for an interactive graph, or 'plt' to use matplotlib.pyplot
show_figure: Default is False. Option to show or not the figure

Returns:

If style is 'Bokeh', returns a set of interactive graphics that are displayed on the website of the COMPLEX campaign
If style is 'plt', Generates graphs in pdf format of the variables of interest with record of the last 6, 12 and 24 hours, these are saved here¹

turbulence.Graph.plotly_RidgelinePlot_web

turbulence.Graph.plotly_RidgelinePlot_web(data = None, variables = None, title = None, legend = False, color = None, fig_size = None, html_save = False, show_figure = False)

The ridgeline plot allows us to study the distribution of the numerical variables of interest. For this case, the following have been selected: 'TAU', 'Ux', 'Uy', 'Uz', 'WS', 'WS_MAX', 'T_SONIC', 'H2O_density', read from EasyFlux, but you can use data that has the format of dataframe described

Parameters:

data: By default it is the data of the object. It must be a DataFrame with the type DatetimeIndex, and columns with variable names, which must have the index name 'TIMESTAMP'
variables: String list with variables for which distribution is required
title: Library that you want to use to make the figure. Options are 'Bokeh' for an interactive graph, or 'plt' to use matplotlib.pyplot
legend: Library that you want to use to make the figure. Options are 'Bokeh' for an interactive graph, or 'plt' to use matplotlib.pyplot
color: Library that you want to use to make the figure. Options are 'Bokeh' for an interactive graph, or 'plt' to use matplotlib.pyplot
fig_size: Default is False. Option to show or not the figure
html_save: Bool. Default is False, if it's changed to True then saved a HTML file
show_figure: Default is False. Option to show or not the figure

Returns:

Creates a standalone HTML that is saved locally and opened inside your web browser. More information here

Examples

Creating random data to graphic


xxxxxxxxxx
import turbulence as tbl
import numpy as np
import pandas as pd
#Dataframe with random data
index = pd.date_range(start='2021-01-01', end='2021-01-31', freq = '5T')
distributions = {'normal': np.random.normal(size = len(index)), 'uniform': np.random.uniform(size = len(index)),
                'power':np.random.power(3,size = len(index)), 'weibull':np.random.weibull(5, size = len(index)),
                'poisson':np.random.poisson(5, size = len(index))}
df = pd.DataFrame(data = distributions, index = index)
df.index.name = 'TIMESTAMP'
df.head()
                       normal  poisson     power   uniform   weibull
TIMESTAMP                                                           
2021-01-01 00:00:00  0.628836        8  0.920784  0.679841  1.052952
2021-01-01 00:05:00  0.405619        6  0.166144  0.163154  0.720098
2021-01-01 00:10:00  0.153764        4  0.301181  0.613829  1.197749
2021-01-01 00:15:00  0.984777        3  0.779014  0.422416  0.765074
2021-01-01 00:20:00 -1.275737        5  0.898457  0.345466  1.201293

Interactive graph using plotly


xxxxxxxxxx
graph = tbl.Graph() #Generating the Graph object
tag_divs = graph.plotly_RidgelinePlot_web(data = df, variables = df.columns.to_list(), show_figure = True)

ridgelineplot

turbulence.Graph.fig_html_web

turbulence.Graph.fig_html_web(lst_figs_bokeh, lst_caption_bokeh, lst_figs_plotly, lst_caption_plotly, path = None)

Save interactive figures created with bokeh or plotly in .html format²

Parameters:

data: By default it is the data of the object. It must be a DataFrame with the type DatetimeIndex, and columns with variable names, which must have the index name 'TIMESTAMP'
variables: String list with variables for which distribution is required
title: Library that you want to use to make the figure. Options are 'Bokeh' for an interactive graph, or 'plt' to use matplotlib.pyplot
legend: Library that you want to use to make the figure. Options are 'Bokeh' for an interactive graph, or 'plt' to use matplotlib.pyplot
color: Library that you want to use to make the figure. Options are 'Bokeh' for an interactive graph, or 'plt' to use matplotlib.pyplot
fig_size: Default is False. Option to show or not the figure
html_save: Bool. Default is False, if it's changed to True then saved a HTML file
show_figure: Default is False. Option to show or not the figure

Returns:

Creates a standalone HTML that is saved locally and opened inside your web browser. More information here

Examples

Generating the random data for the bokeh and plotly figures


xxxxxxxxxx
import turbulence as tbl
import numpy as np
import pandas as pd
#Dataframe with random data
index = pd.date_range(start='2021-01-01', end='2021-01-31')
columns = list('ABCDE')
distributions = {'normal': np.random.normal(size = len(index)), 'uniform': np.random.uniform(size = len(index)),
                'power':np.random.power(3,size = len(index)), 'weibull':np.random.weibull(5, size = len(index)),
                'poisson':np.random.poisson(5, size = len(index))}
df_plotly = pd.DataFrame(data = distributions, index = index)
df_bokeh = pd.DataFrame(np.random.rand(len(index),len(columns)), columns=columns, index = index) 
df_plotly.index.name, df_bokeh.index.name  = 'TIMESTAMP', 'TIMESTAMP'

Creating the figures, saving in lists and making the captions and saving HTML file


xxxxxxxxxx
graph = tbl.Graph() #Generating the Graph object
time_series = graph.time_series(data = df_bokeh)
tag_divs = graph.plotly_RidgelinePlot_web(data = df_plotly, variables = df_plotly.columns.to_list())
lst_caption_bokeh = ['1. Time series created with bokeh library']
lst_caption_plotly = ['2. Ridgeline plot created with plotly library']
lst_figs_bokeh = [time_series]
lst_figs_plotly = [tag_divs]
graph.fig_html_web(lst_figs_bokeh, lst_caption_bokeh, lst_figs_plotly, lst_caption_plotly)

join_html

turbulence.Graph.backup_figure

turbulence.Graph.backup_figure()

This method is employed just with the Sensor object and generate the figure used in backups process

1 If you are a developer you can also find these graphs in this location on the GOMITA server ↩

2 If HTML is created from Sensor object and the no path is given the file is saved here in GOMITA server ↩