NetCDF Grid Generator

A desktop application that converts regular latitude-longitude tables in Excel or CSV format to coarser, georeferenced NetCDF files.

What it does

• Reads .xlsx, .xls, .csv, and .tsv files.
• Lets you select latitude, longitude, optional extra dimensions, and one or more variables.
• Detects and validates regular input-grid spacing.
• Supports nearest-neighbor and spherical area-weighted averaging.
• Rejects output resolutions that are equal to or finer than either input axis.
• Preserves missing values. Area-weighted output ignores missing contributors and remains missing when a target cell has no valid data.
• Writes compressed NetCDF4 with CF-1.8 metadata, WGS84 CRS information, coordinate bounds, and user-supplied descriptions.
• Writes through a temporary file, so a failed conversion does not leave a partial output.

Install and run

Double-click Run App.bat. It checks the default Python environment, installs any missing packages, and launches the application. Python 3.12 or 3.13 must be available in PATH.

For manual setup instead:

cd NetCDF Grid Generator Directory
python -m venv .venv
.venv\Scripts\python -m pip install -r Requirements.txt
.venv\Scripts\python Main.py

The batch launcher installs into the default Python environment as requested. If that location is protected, it retries using the current Windows user's package directory.

Input format

Use a flat table with one row per grid cell. Latitude and longitude must be numeric cell centres on a regular grid. Each combination of selected dimensions, latitude, and longitude must be unique. Variable cells may be blank, NaN, NA, N/A, null, none, or a numeric marker selected in the app.

The output resolution is a single square-cell size in degrees. It must be strictly larger than both input spacings and divide the input extent exactly.

Interpolation methods

• Nearest Neighbor selects the input cell centre nearest to each output cell centre. A missing selected value stays missing.
• Area-weighted average computes exact spherical latitude-longitude overlap areas. Missing source values are excluded from the denominator.

Sample data

The sample_data directory contains global 3.75°, 1°, and 0.25° grids. Regenerate them with:

.venv\Scripts\python scripts\generate_sample_data.py --include-excel

Tests

$env:QT_QPA_PLATFORM = "offscreen"
.venv\Scripts\python -m unittest discover -s tests -v

Run real conversions against all three bundled files with:

.venv\Scripts\python scripts\verify_samples.py