CAMS global atmospheric composition forecasts

CAMS global atmospheric composition forecasts#

Hint

Execute the notebook on the training platform >>

This notebooks provides an introduction to the CAMS global atmospheric composition forecasts and shows you how the variable Dust Aerosol Optical Depth can be used to monitor and predict dust events.

CAMS produces global forecasts for atmospheric composition for the next five days twice a day. The forecasts consist of more than 50 chemical species (e.g. ozone, nitrogen dioxide, carbon dioxide) and seven different types of aerosol (desert dust, sea salt, organic matter, black carbon, sulphate, nitrate and ammonium aerosol).

The initial conditions of each forecast are obtained by combining a previous forecast with current satellite observations through a process called data assimilation. This best estimate of the state of the atmosphere at the initial forecast time step, called the analysis, provides a globally complete and consistent dataset allowing for estimates at locations where observation data coverage is low or for atmospheric pollutants for which no direct observations are available.

The notebook examines the Saharan Dust event which occured over Europe in April 2024.

Basic facts

Spatial resolution: 0.4° x 0.4°
Spatial coverage: Global
Temporal resolution: 1-hourly up to leadtime hour 120
Temporal coverage: since 2015 to present
Data format: GRIB or zipped NetCDF

How to access the data

CAMS global atmospheric composition forecasts are available for download via the Copernicus Atmosphere Data Store (ADS). You will need to create an ADS account here.

Data from the ADS can be downloaded in two ways:

manually via the ADS web interface
programmatically with a Python package called cdsapi (more information)

Load required libraries

import xarray as xr

import matplotlib.pyplot as plt
import matplotlib.colors
from matplotlib.cm import get_cmap
from matplotlib import animation
from matplotlib.axes import Axes

import cartopy.crs as ccrs
from cartopy.mpl.gridliner import LONGITUDE_FORMATTER, LATITUDE_FORMATTER
import cartopy.feature as cfeature
from cartopy.mpl.geoaxes import GeoAxes
GeoAxes._pcolormesh_patched = Axes.pcolormesh

from IPython.display import HTML

import warnings
warnings.simplefilter(action = "ignore", category = RuntimeWarning)

Load helper functions

%run ../functions.ipynb

Optional: Retrieve CAMS global atmospheric composition forecasts programmatically#

The CDS Application Program Interface (CDS API) is a Python library which allows you to access data from the ADS programmatically. In order to use the CDS API, follow the steps below:

Self-register at the ADS registration page (if you do not have an account yet)
Login to the ADS portal and go to the api-how-to-page.
Copy the CDS API key displayed in the black terminal window and replace the ###### of the KEY variable below with your individual CDS API key

Note: You find your CDS API key displayed in the black terminal box under the section Install the CDS API key. If you do not see a URL or key appear in the black terminal box, please refresh your browser tab.

URL='https://ads.atmosphere.copernicus.eu/api/v2'
KEY='#########################'

The next step is then to request the data with a so called API request. Via the ADS web interface, you can specificy the characteristics of the CAMS global atmospheric composition forecasts and at the end of the web interface, you can open the ADS request via Show API request. Copy the request displayed there in the cell below. Once you execute the cell, the download of the data starts automatically.

In the web interface we specified the following characteristics:

Variable: Dust aerosol optical depth at 550 nm
Date: 2024-04-07 to 2024-04-07
time: 12:00 (Start of the model run)
Lead time hour: every 3 hours up to forecast hour 90
Format: netcdf_zip

import cdsapi

c = cdsapi.Client(url=URL, key=KEY)

c.retrieve(
    'cams-global-atmospheric-composition-forecasts',
    {
        'variable': 'dust_aerosol_optical_depth_550nm',
        'date': '2024-04-07/2024-04-07',
        'time': '12:00',
        'leadtime_hour': [
            '0', '12', '15',
            '18', '21', '24',
            '27', '3', '30',
            '33', '36', '39',
            '42', '45', '48',
            '51', '54', '57',
            '6', '60', '63',
            '66', '69', '72',
            '75', '78', '81',
            '84', '87', '9',
            '90',
        ],
        'type': 'forecast',
        'format': 'netcdf_zip',
    },
    '../20240407_duaod.netcdf_zip')

CAMS global atmospheric composition forecasts can be retrieved either in GRIB or in a zipped NetCDF. Above, we requested the data in a zipped NetCDF and for this reason, we have to unzip the file before we can open it. You can unzip zip archives in Python with the Python package zipfile and the function extractall().

import zipfile
with zipfile.ZipFile('../20240407_duaod.netcdf_zip', 'r') as zip_ref:
    zip_ref.extractall('../')

Visualize Dust Aerosol Optical Depth at 550nm#

The next step is to visualize the dataset. You can use the function visualize_pcolormesh, which makes use of matploblib’s function pcolormesh and the Cartopy library.

With ?visualize_pcolormesh you can open the function’s docstring to see what keyword arguments are needed to prepare your plot.

?visualize_pcolormesh

You can make use of the variables we defined above:

units
long_name
latitude
longitude

Additionally, you can specify the color scale and minimum (vmin) and maxium (vmax) data values.

time_index = 10
visualize_pcolormesh(data_array=du_aod[time_index,:,:],
                     longitude=longitude,
                     latitude=latitude,
                     projection=ccrs.PlateCarree(),
                     color_scale='hot_r',
                     unit=units,
                     long_name=long_name + ' ' + str(du_aod[time_index,:,:].time.data)[0:10],
                     vmin=0, 
                     vmax=1.5)

(<Figure size 2000x1000 with 2 Axes>,
 <GeoAxes: title={'center': 'Dust Aerosol Optical Depth at 550nm 2024-04-08'}>)

../_images/02e0ebfc0bd057d82f2e5c7d9e3beb1ae3c9fca49c1531cd3e94e3b72893a302.png

Animate changes of Dust AOD at 550nm over time#

In the last step, you can animate the Dust Aerosol Optical Depth at 550nm in order to see how the trace gas develops over a period of 5 days, from 7th to 11th April 2024.

You can do animations with matplotlib’s function animation. Jupyter’s function HTML can then be used to display HTML and video content.

The animation function consists of 4 parts:

Setting the initial state:
Here, you define the general plot your animation shall use to initialise the animation. You can also define the number of frames (time steps) your animation shall have.
Functions to animate:
An animation consists of three functions: draw(), init() and animate(). draw() is the function where individual frames are passed on and the figure is returned as image. In this example, the function redraws the plot for each time step. init() returns the figure you defined for the initial state. animate() returns the draw() function and animates the function over the given number of frames (time steps).
Create a animate.FuncAnimation object:
The functions defined before are now combined to build an animate.FuncAnimation object.
Play the animation as video:
As a final step, you can integrate the animation into the notebook with the HTML class. You take the generate animation object and convert it to a HTML5 video with the to_html5_video function

# Setting the initial state:
# 1. Define figure for initial plot
fig, ax = visualize_pcolormesh(data_array=du_aod_subset[0,:,:],
                               longitude=du_aod_subset.longitude, 
                               latitude=du_aod_subset.latitude,
                               projection=ccrs.PlateCarree(), 
                               color_scale='hot_r', 
                               unit='-',
                               long_name=long_name + ' - '+ str(du_aod_subset.time[0].data)[0:10],
                               vmin=0,
                               vmax=1, 
                               lonmin=lonmin, 
                               lonmax=lonmax, 
                               latmin=latmin, 
                               latmax=latmax,
                               set_global=False)

frames = 30

def draw(i):
    img = plt.pcolormesh(du_aod_subset.longitude, 
                         du_aod_subset.latitude, 
                         du_aod_subset[i,:,:], 
                         cmap='hot_r', 
                         transform=ccrs.PlateCarree(),
                         vmin=0,
                         vmax=1,
                         shading='auto')
    
    ax.set_title(long_name + ' - '+ str(du_aod_subset.time[i].data)[0:10], fontsize=20, pad=20.0)
    return img

def init():
    return fig

def animate(i):
    return draw(i)

ani = animation.FuncAnimation(fig, animate, frames, interval=800, blit=False,
                              init_func=init, repeat=True)

HTML(ani.to_html5_video())
plt.close(fig)

Play the animation video as HTML5 video

HTML(ani.to_html5_video())

CAMS global atmospheric composition forecasts

Contents

CAMS global atmospheric composition forecasts#

Optional: Retrieve CAMS global atmospheric composition forecasts programmatically#

Load and browse CAMS global atmospheric composition forecasts#

Retrieve Dust Aerosol Optical Depth at 550nm as data array#

Visualize Dust Aerosol Optical Depth at 550nm#

Create a geographical subset for Europe#

Animate changes of Dust AOD at 550nm over time#