A set of Python tools to make it easier to extract weather station data (e.g., temperature, precipitation) from the Global Historical Climatology Network - Daily (GHCND)

More information on the data can be found here

The code can be downloaded from the get_station_data github repository

from get_station_data import ghcnd
from get_station_data.util import nearest_stn

%matplotlib inline

Read station metadata

stn_md = ghcnd.get_stn_metadata()

Choose a location (lon/lat) and number of nearest neighbours

london_lon_lat = -0.1278, 51.5074
my_stns = nearest_stn(stn_md, 
                        london_lon_lat[0], london_lon_lat[1], 
                        n_neighbours=5 )
my_stns
station lat lon elev name
52113 UKE00105915 51.5608 0.1789 137.0 HAMPSTEAD
52165 UKM00003772 51.4780 -0.4610 25.3 HEATHROW
52098 UKE00105900 51.8067 0.3581 128.0 ROTHAMSTED
52191 UKW00035054 51.2833 0.4000 91.1 WEST MALLING
52131 UKE00107650 51.4789 0.4489 25.0 HEATHROW

Download and extract data into a pandas DataFrame

df = ghcnd.get_data(my_stns)

df.head()
station year month day element value mflag qflag sflag date lon lat elev name
0 UKE00105915 1959 12 1 TMAX NaN 1959-12-01 0.1789 51.5608 137.0 HAMPSTEAD
1 UKE00105915 1959 12 2 TMAX NaN 1959-12-02 0.1789 51.5608 137.0 HAMPSTEAD
2 UKE00105915 1959 12 3 TMAX NaN 1959-12-03 0.1789 51.5608 137.0 HAMPSTEAD
3 UKE00105915 1959 12 4 TMAX NaN 1959-12-04 0.1789 51.5608 137.0 HAMPSTEAD
4 UKE00105915 1959 12 5 TMAX NaN 1959-12-05 0.1789 51.5608 137.0 HAMPSTEAD

Filter data for, e.g., a single variable

var = 'PRCP'   # precipitation
df = df[ df['element'] == var ]

### Tidy up columns
df = df.rename(index=str, columns={"value": var})
df = df.drop(['element'], axis=1)

df.head()
station year month day PRCP mflag qflag sflag date lon lat elev name
93 UKE00105915 1960 1 1 2.5 E 1960-01-01 0.1789 51.5608 137.0 HAMPSTEAD
94 UKE00105915 1960 1 2 1.5 E 1960-01-02 0.1789 51.5608 137.0 HAMPSTEAD
95 UKE00105915 1960 1 3 1.0 E 1960-01-03 0.1789 51.5608 137.0 HAMPSTEAD
96 UKE00105915 1960 1 4 0.8 E 1960-01-04 0.1789 51.5608 137.0 HAMPSTEAD
97 UKE00105915 1960 1 5 0.0 E 1960-01-05 0.1789 51.5608 137.0 HAMPSTEAD 
df.drop(columns=['mflag','qflag','sflag']).tail(n=10)
station year month day PRCP date lon lat elev name
83938 UKE00107650 2016 12 22 0.0 2016-12-22 0.4489 51.4789 25.0 HEATHROW
83939 UKE00107650 2016 12 23 1.4 2016-12-23 0.4489 51.4789 25.0 HEATHROW
83940 UKE00107650 2016 12 24 0.0 2016-12-24 0.4489 51.4789 25.0 HEATHROW
83941 UKE00107650 2016 12 25 1.0 2016-12-25 0.4489 51.4789 25.0 HEATHROW
83942 UKE00107650 2016 12 26 0.0 2016-12-26 0.4489 51.4789 25.0 HEATHROW
83943 UKE00107650 2016 12 27 0.0 2016-12-27 0.4489 51.4789 25.0 HEATHROW
83944 UKE00107650 2016 12 28 0.2 2016-12-28 0.4489 51.4789 25.0 HEATHROW
83945 UKE00107650 2016 12 29 0.4 2016-12-29 0.4489 51.4789 25.0 HEATHROW
83946 UKE00107650 2016 12 30 0.0 2016-12-30 0.4489 51.4789 25.0 HEATHROW
83947 UKE00107650 2016 12 31 0.4 2016-12-31 0.4489 51.4789 25.0 HEATHROW

Save to file

df.to_csv('London_5stns_GHCN-D.csv', index=False)

Plot histogram of all data

df['PRCP'].plot.hist(bins=40)

<matplotlib.axes._subplots.AxesSubplot at 0x11ae36898>

png

Plot time series for one station

heathrow = df[ df['name'] == 'HEATHROW' ]
heathrow['PRCP'].plot()

<matplotlib.axes._subplots.AxesSubplot at 0x81f0d7240>

png