Downloading ERA5 Atmospheric Reanalysis
Published:
To begin, make sure you have:
- A registered account with the Climate Data Store (CDS)
- cdsapi installed by following the CDSAPI setup
- Python 3 installed
Because we want to download multiple years of daily-resolution wind data, we’ll use the CDS API.
Step 1: Configure Your Data Request
Go to the dataset ERA5 hourly data on single levels from 1940 to present and fill in the following options:
- Product Type: Reanalysis
- Variables:
10m_u_component_of_wind10m_v_component_of_wind
- Years: Start with
1979(full loop through 2025 shown below) - Months and Days: Select all
- Hours
12:00– most representative for the day - Geographical Boundaries:
- North:
-50 - West:
-180 - South:
-90 - East:
180
(Covers the Southern Ocean)
- North:
- Data Format:
NetCDF4 (experimental)
At the bottom of the CDS interface, click “Show API request” and copy the generated Python code.
Step 2: Use the CDS API with a Loop for Multiple Years
Below is an example script using the CDS API to download data from 1979 to 2024.
import cdsapi
from datetime import datetime, timedelta
from concurrent.futures import ThreadPoolExecutor
import os
# Function to download ERA5 data for a specific day
def download_day(date_str):
client = cdsapi.Client()
year = date_str[:4] # breaking up the string
month = date_str[4:6]
day = date_str[6:8]
# Define the folder and filename
output_dir = os.path.join("data", year) # e.g., data/1985
os.makedirs(output_dir, exist_ok=True) # Create folder if it doesn't exist
filename = os.path.join(output_dir, f"era5_wind_{date_str}_12UTC.nc")
print(f"→ Downloading {filename}")
# Submit request to CDS API
client.retrieve(
"reanalysis-era5-single-levels",
{
"product_type": "reanalysis",
"variable": ["10m_u_component_of_wind", "10m_v_component_of_wind"],
"year": year,
"month": month,
"day": day,
"time": ["12:00"],
"data_format": "netcdf",
"area": [-50, -180, -90, 180]
},
filename
)
print(f" ✔ Done: {filename}")
# Generate list of all days from Jan 1, 1979 to Dec 31, 2024
start_date = datetime(1979, 1, 1)
end_date = datetime(2024, 12, 31)
date_list = []
current_date = start_date
while current_date <= end_date:
date_list.append(current_date.strftime("%Y%m%d"))
current_date += timedelta(days=1)
# Use a thread pool to download in parallel (adjust max_workers if needed)
with ThreadPoolExecutor(max_workers=4) as executor:
executor.map(download_day, date_list)
Step 3: Running the Script as a Batch Job or in the Background
Because we are downloading about 394,200 time steps (24hrx365dx45yr) it can take a while. In order to avoid potential interruptions we can run the script in the background. In my case, I will submit it as a job on a cluster.
My python file is saved as UV_Winds_1979-2025.py, and I will run it using nohup:
nohup python UV_Winds_1979-2025.py > UV_download.log 2>&1 &
This will keep it running in the background and you can check the progress/errors on UV_download.log
I recommend running:
tail -f UV_download.log
…to check that it is running without any errors :)