filter-stations 0.3.12

Making it easier to navigate and clean station data

Documentation

https://filterstations.netlify.app/

Water Level Pipeline

• A series of functions to be added to the filter-stations module in pypi to evalute which TAHMO stations to use that corroborates with the water level
• All begins with the coordinates of the gauging station(location of the monitoring sensor)

import os
from pathlib import Path
import haversine as hs
import pandas as pd
import numpy as np
import datetime
import statsmodels.api as sm
from matplotlib.dates import DateFormatter
import matplotlib.pyplot as plt

# config_path
config_path = os.path.join(Path(os.getcwd()).parent.absolute(), 'config.json')

from filter_stations import retreive_data, Interactive_maps, Filter
import json
# Authentication
with open(config_path) as f:
    conf = json.load(f)

apiKey = conf['apiKey']
apiSecret = conf['apiSecret']
fs = retreive_data(apiKey, apiSecret)

# given the radius and the longitude and latitude of the gauging station, return the stations within
def stations_within_radius(radius, latitude, longitude, df=False):
    stations  = fs.get_stations_info()
    stations['distance'] = stations.apply(lambda row: hs.haversine((latitude, longitude), (row['location.latitude'], row['location.longitude'])), axis=1)
    infostations = stations[['code', 'location.latitude','location.longitude', 'distance']].sort_values('distance')
    if df:
        return infostations[infostations['distance'] <= radius]
    else:
        return infostations[infostations['distance'] <= radius].code.values

ewaso = stations_within_radius(100, -0.406689, 36.96301)
ewaso

API request: services/assets/v2/stations





array(['TA00283', 'TA00378', 'TA00754', 'TA00074', 'TA00196', 'TA00073',
       'TA00056', 'TA00029', 'TA00416', 'TA00719', 'TA00258', 'TA00622',
       'TA00028', 'TA00414', 'TA00190', 'TA00078', 'TA00024', 'TA00080',
       'TA00166', 'TA00108', 'TA00026', 'TA00189', 'TA00250', 'TA00182',
       'TA00715', 'TA00377', 'TA00027', 'TA00057', 'TA00134', 'TA00448',
       'TA00774', 'TA00773', 'TA00772', 'TA00775', 'TA00771', 'TA00679',
       'TA00770'], dtype=object)

The assumption here is one can have credential but not the data

• From the list of stations get the precipitation data with a certain data completeness check provided
• Additionally the start and end date if the data is not provided
• The default start date is the day the sensors were set up at DSAIL
• Chck the documentation on the types of variables available

def stations_data_check(stations_list, percentage=1, start_date=None, end_date=None, data=None, variables=['pr'], csv_file=None):
    if data is None:
        data = fs.multiple_measurements(stations_list, startDate=start_date, endDate=end_date, variables=variables, csv_file=csv_file)

    # Check the percentage of missing data and return the stations with less than the percentage of missing data
    data.index = data.index.astype('datetime64[ns]')
    data = data.dropna(axis=1, thresh=int(len(data) * percentage))
    data.to_csv(f'{csv_file}.csv')
    return data

stations_df = stations_data_check(list(ewaso), start_date='2022-12-01', end_date='2022-12-31', variables=['pr'], csv_file='ewaso2.csv')

Apart from the completeness another method of validation by eliminating unusable sensors is checking for a positive correlation and lag

• The default lag is 3 days between a particular station and the gauging station
• The required format is a timeseries data
• Provide the column names for evaluation format = [Date, data]
• with the change in parameters one can choose above or below threshold

def stations_lag(weather_stations_df, gauging_stations_df, gauging_station_columns, date=None, lag=3, above=False, below=False):
    
    
    # set the date as axis
    # weather_station_df = weather_stations_df.set_index('Date')
    # weather_stations_df.Date= weather_stations_df.Date.apply(pd.to_datetime,dayfirst = True)
    # weather_stations_df = weather_stations_df.set_index('Date')
    # get the starting date of the gauging station the first value
    if date is None:
        date = gauging_stations_df.loc[0, gauging_station_columns[0]]
    start_date = datetime.datetime.strptime(date, "%d/%m/%Y")
    end_date = start_date + datetime.timedelta(len(gauging_stations_df)-1)
    # get the ddataframe from start date to end date
    df_fit = weather_stations_df[start_date:end_date]
    # get the water data list
    water_list = list(gauging_stations_df[f'{gauging_station_columns[1]}'])
    above_thresh_lag = dict()
    below_thresh_lag = dict()
    # get the lag for every column against the water data 
    for cols in df_fit.columns:
        select_list = list(df_fit[cols])
        coefficient_list = list(sm.tsa.stattools.ccf(select_list,water_list, adjusted=False))
        a = np.argmax(coefficient_list)
        b = coefficient_list[a]
        if a > lag:
            above_thresh_lag[cols] = {
                'lag': a,
                'coefficient': b,
                'coefficient_list': coefficient_list,
                'select_list': select_list,
                'water_list' : water_list
            }
        else:
            below_thresh_lag[cols] = {
                'lag': a,
                'coefficient': b,
                'coefficient_list': coefficient_list,
                'select_list': select_list,
                'water_list' : water_list
            }
    if above and below:
        return above_thresh_lag, below_thresh_lag
    elif above:
        return above_thresh_lag
    elif below:
        return below_thresh_lag

water_six = pd.read_csv('./water-level-data-ewaso/1E2020.csv')
water_six

<style scoped> .dataframe tbody tr th:only-of-type { vertical-align: middle; }

.dataframe tbody tr th {
    vertical-align: top;
}

.dataframe thead th {
    text-align: right;
}

</style>

	time	water_level(m)
0	12/05/2020	2.618646
1	13/05/2020	2.551392
2	14/05/2020	2.507711
3	15/05/2020	2.491130
4	16/05/2020	2.434761
...	...	...
259	26/01/2021	0.947099
260	27/01/2021	0.929186
261	28/01/2021	0.911274
262	29/01/2021	0.910711
263	30/01/2021	0.939971

264 rows × 2 columns

lag_ = stations_lag(stations_df, water_six, ['time', 'water_level(m)'], lag=3,below=True)
lag_

Plotting

Provides visuals of the data

• An option to save the
• An option of choosing the dpi
• provide the startDate based on the water collection starting date

lag_[list(lag_.keys())[0]]['water_list']

import warnings
warnings. filterwarnings('ignore')

def plot_figs(weather_stations, water_list, threshold_list, save=False, dpi=500, date='11-02-2021'):
    start_date = datetime.datetime.strptime(date, "%d-%m-%Y")
    end_date = start_date + datetime.timedelta(len(water_list)-1)
    # weather_stations = weather_stations.set_index('Date')
    df_plot = weather_stations[start_date:end_date]
    df_plot = df_plot[threshold_list].reset_index()
    df_plot.rename(columns={'index':'Date'}, inplace=True)
    
    
    plt.rcParams['figure.figsize'] = (15, 9)
    print('Begin plotting!')
    
    for cols in df_plot.columns[1:]:
        fig, ax1 = plt.subplots()
        color = 'tab:blue'
        ax1.set_xlabel(f'Time', fontsize=24, weight='bold')
        ax1.set_ylabel(f'Rainfall {cols} (mm)', color=color, fontsize=24, weight='bold')
        ax1.bar(pd.to_datetime(df_plot['Date'], format="%d/%m/%Y"), df_plot[f'{cols}'], color=color, width=4, alpha=1.0)
        ax1.tick_params(axis='y', labelcolor=color, labelsize=24)
        ax1.tick_params(axis='x')
        ax1.set_xticklabels(df_plot['Date'], fontsize=21, weight='bold')
        ax1.grid(color='gray', linestyle='--', linewidth=0.8)
        ax1.set(facecolor="white")
        ax2 = ax1.twinx()  # instantiate a second axes that shares the same x-axis

        color = 'tab:red'
        ax2.set_ylabel('Water level/Stage (m)', color=color, fontsize=24, weight='bold')
        ax2.plot(pd.to_datetime(df_plot['Date'], format="%d/%m/%Y"), water_list, color=color, linewidth=4)
        ax2.tick_params(axis='y', labelcolor=color, labelsize=24)
        ax2.set(facecolor="white")
        plt.title('Stage and Rainfall against Time', fontsize=22, weight='bold')

        date_form = DateFormatter("%m-%y")
        ax1.xaxis.set_major_formatter(date_form)
        fig.tight_layout()

        if save:
            fig.savefig(f'{cols}.png', dpi=dpi)

plot_figs(stations_df, lag_[list(lag_.keys())[0]]['water_list'], list(lag_.keys()), save=True, date='12-05-2020')

Begin plotting!

Format to get the stations maetadata

def filter_metadata(lag_keys):
    captured_list = [i.split('_')[0] for i in list(lag_keys)]
    return fs.get_stations_info(multipleStations=captured_list)

filter_metadata(list(lag_.keys()))

API request: services/assets/v2/stations

<style scoped> .dataframe tbody tr th:only-of-type { vertical-align: middle; }

.dataframe tbody tr th {
    vertical-align: top;
}

.dataframe thead th {
    text-align: right;
}

</style>

	code	status	installationdate	elevationground	sensorinstallations	dataloggerinstallations	creatorid	created	updaterid	updated	...	location.countrycode	location.zipcode	location.latitude	location.longitude	location.elevationmsl	location.note	location.creatorid	location.created	location.updaterid	location.updated
26	TA00028	1	2015-08-31T00:00:00Z	9.0	None	None	2	2018-12-11T08:35:17.888233Z	2	2018-12-11T08:35:17.888233Z	...	KE		0.055219	37.136747	2003.6	{}	2	2018-10-26T13:32:16.15537Z	37	2022-06-30T11:11:50.27135Z
27	TA00029	1	2015-09-02T00:00:00Z	2.0	None	None	2	2018-12-11T08:36:19.30342Z	2	2018-12-11T08:36:19.30342Z	...	KE		-0.500776	36.587511	2545.8	{}	2	2018-10-26T13:33:31.451613Z	37	2022-02-28T12:25:09.578242Z
53	TA00057	1	2015-10-08T00:00:00Z	2.0	None	None	2	2018-12-11T09:21:29.092833Z	2	2018-12-11T09:21:29.092833Z	...	KE		-1.253030	36.856487	1645.3	{}	2	2018-10-29T09:13:33.768613Z	2	2022-07-26T07:34:06.603938Z
68	TA00074	1	2015-11-19T00:00:00Z	2.0	None	None	2	2018-12-11T09:38:25.742397Z	2	2018-12-11T09:38:25.742397Z	...	KE		-0.566080	37.074412	1726.8	{}	2	2018-10-29T10:35:28.49617Z	2	2022-07-26T07:38:42.100985Z
74	TA00080	1	2016-01-28T00:00:00Z	2.0	None	None	2	2018-12-11T09:43:10.523398Z	2	2018-12-11T09:43:10.523398Z	...	KE		-1.087589	36.818402	1777.3	{}	2	2018-10-29T10:53:47.845042Z	37	2022-02-28T13:07:04.709903Z
150	TA00166	1	2017-05-11T00:00:00Z	2.0	None	None	2	2018-12-12T08:29:28.10697Z	2	2018-12-12T08:29:28.10697Z	...	KE		-0.319508	37.659139	1404.0	{}	2	2018-11-10T08:47:37.949135Z	2	2018-11-10T08:47:37.949135Z

6 rows × 28 columns