Comparing Winter 2013 PHX Temperatures to the 40 year average using Python Pandas and GSOD (https://www.ncdc.noaa.gov/cgi-bin/res40.pl) data.

example2.py

#!/usr/bin/env python

import pandas as pd
import matplotlib.pyplot as plt

# Read in the CSV Temperature data retrieved from Global Surface Summary of Day
# https://www.ncdc.noaa.gov/cgi-bin/res40.pl
temps = pd.read_csv('phx-temps.csv', index_col=0,
                    names=['highs', 'lows'],
                    parse_dates=True)

# Add an Average Column
temps['avg'] = (temps.highs + temps.lows) / 2

# Add a Day of Year Column and then group by that column
# This will allow us to align years by DOY so we can take the mean of each DOY
temps['doy'] = temps.index.dayofyear
temps_group = temps.groupby(temps['doy'])

# Get an array of Days of Year that matches the ROI
roi_doy = temps['20121101':'20130401'].index.dayofyear
# Get the matching ROI DateTime Index
roi_index = temps['20121101':'20130401'].index

# Dataframe containing the 2012-2013 values in the ROI as well as the 40 year
# mean, mapped into the ROI by day of year
df = pd.DataFrame({
    'highs': pd.Series(temps['20121101':'20130401'].highs),
    'mean_highs': pd.Series(temps_group.highs.mean()[roi_doy].values,
                            index=roi_index),
    'avg': pd.Series(temps['20121101':'20130401'].avg),
    'mean_avg': pd.Series(temps_group.avg.mean()[roi_doy].values,
                          index=roi_index),
    'lows': pd.Series(temps['20121101':'20130401'].lows),
    'mean_lows': pd.Series(temps_group.lows.mean()[roi_doy].values,
                           index=roi_index)
})

# Compute Rolling Means
df['highs_10day_ra'] = pd.rolling_mean(df['highs'], 10)
df['avg_10day_ra'] = pd.rolling_mean(df['avg'], 10)
df['lows_10day_ra'] = pd.rolling_mean(df['lows'], 10)

# Begin Plotting
y_max = 90
y_min = 28

df[['avg', 'mean_avg']].plot(ylim=[y_min, y_max], label=['A', 'B'])
plt.title('Phoenix Winter 2013 Average Daily Temperatures')
plt.ylabel('Temperature (F)')
plt.savefig('phxtemps-winter2012-ADT.png')

df[['avg_10day_ra', 'mean_avg']].plot(ylim=[y_min, y_max])
plt.title('Phoenix Winter 2013 Average Daily Temperatures')
plt.ylabel('Temperature (F)')
plt.savefig('phxtemps-winter2012-ADT-ra.png')

df[['lows', 'mean_lows']].plot(ylim=[y_min, y_max])
plt.title('Phoenix Winter 2013 Daily Low Temperatures')
plt.ylabel('Temperature (F)')
plt.savefig('phxtemps-winter2012-lows.png')

df[['lows_10day_ra', 'mean_lows']].plot(ylim=[y_min, y_max])
plt.title('Phoenix Winter 2013 Daily Low Temperatures')
plt.ylabel('Temperature (F)')
plt.savefig('phxtemps-winter2012-lows-ra.png')

df[['highs', 'mean_highs']].plot(ylim=[y_min, y_max])
plt.title('Phoenix Winter 2013 Daily High Temperatures')
plt.ylabel('Temperature (F)')
plt.savefig('phxtemps-winter2012-highs.png')

df[['highs_10day_ra', 'mean_highs']].plot(ylim=[y_min, y_max])
plt.title('Phoenix Winter 2013 Daily High Temperatures')
plt.ylabel('Temperature (F)')
plt.savefig('phxtemps-winter2012-highs-ra.png')

get_phx.py

#!/usr/bin/env python

# uses gsod module: https://github.com/imankulov/gsod

from gsod import StationLoader
from gsod import WeatherLoader

# figure out what station identifiers are
#loader = StationLoader()
#stations = loader.get_stations()
#for station in stations:
#    if station['station_name'] == 'PHOENIX/SKY HARBOR':
#        phx_usaf = station['usaf']
#        phx_wban = station['wban']

# PHX Sky Harbor IDs
u = 722780
w = 23183

wloader = WeatherLoader()

for year in range(1950,2014):
    weather = wloader.get_weather(u, w, year)
    for obs in weather:
        print str(obs['date']) + ',' + str(obs['max_temp']) + ',' + str(obs['min_temp'])

requirements.txt

Bottleneck==0.6.0
argparse==1.2.1
distribute==0.6.24
ipython==0.13.2
matplotlib==1.2.1
numexpr==2.1
numpy==1.7.1
openpyxl==1.6.2
pandas==0.11.0
python-dateutil==2.1
pytz==2013b
pyzmq==13.1.0
scipy==0.12.0
six==1.3.0
statsmodels==0.5.0
tornado==3.1
wsgiref==0.1.2
xlrd==0.9.2
xlwt==0.7.5

Note that the requirements file is a bit excessive for just this example. These requirements give a working IPython notebook environment as well as include pandas features not used here.