As of November 2023 the Copernicus Open Access Hub is no longer operational. Querying products needs to be updated to work with the new Copernicus Data Space Ecosystem API.
About: To streamline the process of monitoring emissions above Thailand and the surrounding areas, an automatic pipeline to download, process, and visualize Sentinel-5P data is created.
Data: The atmospheric data used in this project is captured by the TROPOMI instrument onboard the Sentinel-5P satellite, part of the European Space Agency's Copernicus program.
The following precursor and greenhouse gases are monitored:
- Nitrogen Dioxide (NO2)
- Formaldehyde (HCHO)
- Carbon Monoxide (CO)
- Sulfur Dioxide (SO2)
- Methane (CH4)
Workflow: The pipeline consists of three main parts:
- Querying and downloading Level 2 Sentinel-5P data
- Processing the downloaded Level 2 data into Level 3 data
- Temporally-aggregating and visualizing the L3 processed data
Live Implementation:
https://gicait.github.io/sentinel5P-automated
To execute the full processing pipeline, clone the repository, install the necessary dependencies and run execute.py. This will execute query.py, process.py, and multitemporal.py consecutively. The scripts can also be run manually one at a time. Make sure the scripts are all in the same folder if downloading manually.
No modification of the scripts should be necessary for the showcased results but can be done to change the output (e.g., study area, pollutants, visualization style)
Description:
This script is responsible for the high level automation of the entire workflow. It executes the three scripts query.py, process.py, and multitemporal.py consecutively. If any execution errors occur, the exception output is printed.
Note:
- The script assumes that the
query.py,process.py, andmultitemporal.pyscripts are located in the same directory.
Description:
This script is responsible for querying and downloading the Level 2 (L2) Sentinel-5P products from the Copernicus Open Access Hub. It queries Near Real-Time (NRT) and Offline (OFFL) products in a user-defined AOI (defined by a GeoJSON file, parsed as WKT) and time-frame (NRT default: current date - one week ago, OFFL default: one week ago - two weeks ago). For NRT available products such as sulfur dioxide, nitrogen dioxide, carbon monoxide, and formaldehyde, the data may be a mix of NRT and OFFL products. These products are generally available 3 hours after sensing. For OFFL only available products such as methane, the data is available about 5 days after sensing. Due to this a query time-frame more recent than one week ago is not recommended for OFFL products. See the Sentinel-5P Data Products description for more product details.
All products are downloaded as NetCDF files. If it does not already exist, the script creates the directory Products_Raw/ to store the downloaded data in. If any of the target products are already contained in this directory, they will not be downloaded again. Any products older than the specified time-frame are deleted.
Third party dependencies:
Note:
- By default, this script requires the GeoJSON file
thailand_boundary_simple.geojsonto be in theSupport_Files/directory to define the AOI. To use a different GeoJSON file, specify the file path in line 10.
Description:
This script processes the downloaded Level 2 (L2) data products into Level 3 (L3) products using the HARP library and should be executed after query.py. The following processing steps are performed:
- Validity filtering: The
/PRODUCT/qa_value(HARP field name:[product]_validity) is used to filter the product by quality. The default value is set to 75. - Spatial filtering: The data is filtered to the desired spatial extent of the analysis/visualization. Default: Area around Thailand (Lat Lon: 5, 95 – 21, 110).
- Spatial regridding: The data is resampled to a new raster.
- Derivations: The coverage stop time and central coordinates of each cell are derived.
- Attribute filtering: The product attributes to be kept are defined. All other attributes are filtered out.
After processing, the L3 products are saved as NetCDF files. If it does not already exist, the script creates the directory Products_Processed/ to store the files in. If any of the L3 target products are already contained in this directory, they will not be generated again. Any products older than the specified time-frame (same time-frames as in query.py by default) are deleted.
Third party dependencies:
Note:
- It is assumed that L2 products are downloaded and stored in the
Products_Raw/directory. - The exception "Error: product contains no variables, or variables without data." occurs when no cells of an entire product are greater than the minimum validity threshold. In this case, processing is skipped and no L3 product is generated. This has been observed as a common occurrence for methane (CH4) products.
Description:
This script takes care of averaging and visualizing the L3 processed data and should be executed after process.py. The L3 product attributes, their description, value range and units are defined for the visualization. The value range may be adjusted if inadequate. The script iterates through each product, opening all product files and calculating the mean values for each cell of the attribute to be visualized. The results are plotted and saved as PNG images in the Output/[Y_m_d]/ directory. A single JPG image containing all outputs is also created in the Output/ directory. The visualizations are created using the cartopy library and can be modified based on use case/preference. By default, any Output/[Y_m_d]/ directories older than eight weeks are deleted to save space. The script also creates a GIF animation of previous outputs for each product in the Output/ directory. If already present, the GIFs will be overwritten each time the script is run. By default, the eight most recent outputs are included in the GIF.
Although not part of the main workflow, multitemporal_tiff.py can be used to output the average concentrations as GeoTIFF files for further analysis instead of PNG/JPG images. To do this, run this script after process.py instead of multitemporal.py.
Third party dependencies:
Note:
- It is assumed that L3 products are already processed and stored in the
Products_Processed/directory. - It is assumed that L2 products are downloaded and stored in the
Products_Raw/directory.