XSeasonsDetect

XSeasonDetect is a Python tool designed for the detection and analysis of meteorological and climatological seasons using Machine Learning, designed to be compatible with xarray.

The tool is built upon a Machine Learning algorithm proposed by A. J Cannon in the article Defining climatological seasons using radially constrained clustering (2005).

Script mode

Create a new project with:

XSeas_newproj --name <project_name>

In the data/raw/ERA5 create a file for each variable you wanto to include into the analysis. Fill each folder wit han .nc file named final.nc.

Data Preprocessing Script for ERA5 NetCDF Files

Then run the script for the preprocessing of ERA5 data:

XSeas_preprocessERA

This script automates the preprocessing of ERA5 climate data stored in NetCDF format. Below is a step-by-step explanation of how the script works:

• Input Data Requirements:

  • Raw NetCDF files should be organized into subfolders within data/raw/ERA5.
  • A target grid file (config/target_grid.txt) must be present to define the spatial grid resolution.
  • A geographic boundary file (data/raw/shapefiles/boundary.gpkg) is required to clip the data spatially.

• Processing Workflow:

  1. Folder Detection: The script scans data/raw/ERA5 for subfolders containing raw NetCDF files.
  2. Directory Setup: For each folder, it creates intermediate (data/temp/ERA5) and output (data/preprocessed/ERA5) directories if they don’t already exist.

• Preprocessing: The following operations are applied to each folder:

  1. Regridding: Matches the spatial resolution defined in the target grid file.
  2. Clipping: Restricts the data to the area defined in the boundary file.
  3. Temporal Filtering: Keeps only the data within the specified time range (default: 1960–2020).
  4. Overwrite Handling: If a preprocessed file (final.nc) already exists in the output directory, the user is prompted to overwrite or skip.

• Output:

  • Preprocessed data is saved as final.nc in the corresponding folder within data/preprocessed/ERA5.