earth_surface_temperature.py

import pandas as pd
from datetime import date
import streamlit as st
import plotly.express as px
import zipfile
import kaggle
from scipy import signal

st.set_page_config(layout='wide', page_title='Land Surface Temperature')

st.title('Global Land Surface Temperature')
st.header('Monthly average since 1890')

load_kaggle = st.checkbox('Load from Kaggle')

@st.cache_data
def load_data(load_kaggle=load_kaggle):
    if load_kaggle:
        kaggle.api.authenticate()
        kaggle.api.dataset_download_files('berkeleyearth/climate-change-earth-surface-temperature-data', path='.',
                                          unzip=False)
        with zipfile.ZipFile('climate-change-earth-surface-temperature-data.zip', 'r') as zip_ref:
            zip_ref.extractall('earth_surface_temperature')
    df = pd.read_csv('earth_surface_temperature/GlobalLandTemperaturesByCity.csv').dropna(axis=0)
    df['dt'] = df['dt'].apply(date.fromisoformat)
    df['Year'] = df['dt'].apply(lambda x: x.year)
    df = df[['Year', 'dt', 'AverageTemperature', 'City', 'Latitude', 'Longitude']]
    df['AdjustedTemperature'] = (df['AverageTemperature'] * 9 / 5) + 32 + 45
    df['Latitude'] = df['Latitude'].apply(lambda x: float(x[:-1]) if x[-1] == 'N' else -float(x[:-1]))
    df['Longitude'] = df['Longitude'].apply(lambda x: float(x[:-1]) if x[-1] == 'E' else -float(x[:-1]))
    return df[df['Year']>=1890]

@st.cache_data
def min_temp(thres_l, thres_u):
    return load_data().drop('dt', axis=1)[load_data()['Latitude'].apply(lambda x: thres_l<=abs(x)<=thres_u)]\
                    .groupby(['City', 'Year']).min().reset_index().drop('City', axis=1).groupby('Year').mean().reset_index()

@st.cache_data
def max_temp(thres_l, thres_u):
    return load_data().drop('dt', axis=1)[load_data()['Latitude'].apply(lambda x: thres_l<=abs(x)<=thres_u)]\
                    .groupby(['City', 'Year']).max().reset_index().drop('City', axis=1).groupby('Year').mean().reset_index()

@st.cache_data
def smooth(data):
    return signal.savgol_filter(data, 21, 1)


with st.expander('Min-Max temperature'):
    threslc, thresuc = st.columns(2)
    with threslc:
        thres_l = st.slider('min latitude', 0, 50, value=0)
    with thresuc:
        thres_u = st.slider('max latitude', thres_l + 10, 90, value=max(thres_l + 10, st.session_state.get('thres_u', 90)))
    st.session_state['thres_u'] = thres_u
    minc, maxc = st.columns(2)
    with minc:
        st.plotly_chart(px.scatter(min_temp(thres_l, thres_u),
                              x='Year', y='AverageTemperature',
                                   title='Minimum yearly temperature, averaged over cities in latitude range',
                                   labels={'AverageTemperature': 'Temperature in °C'}#, color='City'
                                   ).update_traces(
            marker={'size': 4}).add_traces(
            px.line(
                min_temp(thres_l, thres_u), x='Year',
                y=smooth(min_temp(thres_l, thres_u)['AverageTemperature']), #color='City',
            ).data
        ).add_hrect(y0=min(smooth(min_temp(thres_l, thres_u)['AverageTemperature'])),
                    y1=max(smooth(min_temp(thres_l, thres_u)['AverageTemperature'])),
                    opacity=0.2, fillcolor='purple', line_width=0),
                        use_container_width=True)
    with maxc:
        st.plotly_chart(px.scatter(max_temp(thres_l, thres_u),
                                  x='Year', y='AverageTemperature',
                                   title='Maximum yearly temperature, averaged over cities in latitude range',
                                   labels={'AverageTemperature': 'Temperature in °C'}#, color='City'
                                   ).update_traces(
            marker={'size': 4}).add_traces(
            px.line(
                max_temp(thres_l, thres_u), x='Year',
                y=smooth(max_temp(thres_l, thres_u)['AverageTemperature']), #color='City',
            ).data
        ).add_hrect(y0=min(smooth(max_temp(thres_l, thres_u)['AverageTemperature'])),
                    y1=max(smooth(max_temp(thres_l, thres_u)['AverageTemperature'])),
                    opacity=0.2, fillcolor='purple', line_width=0),
            use_container_width=True)

with st.expander('On the map over time'):
    st.plotly_chart(px.scatter_geo(load_data()[load_data()['Year'].apply(lambda x: x%10==0)], lat='Latitude', lon='Longitude',
                         size="AdjustedTemperature",
                       size_max=6,
                       animation_frame='dt', labels={'AverageTemperature': 'Monthly Average Temperature'},
                   range_color=[-45, 45],
                   color_continuous_scale='Cividis',
                         color="AverageTemperature", height=600
                                   ),
                    use_container_width=True)