Convert GeoTIFFs to Cloud Optimized GeoTIFF

This script batch converts standard GeoTIFF files into Cloud Optimized GeoTIFFs (COGs) using gdal_translate.

---

import os
import subprocess
from osgeo import gdal

input_folder = "input_folder"
output_folder = "output_folder"

os.makedirs(output_folder, exist_ok=True)

for filename in os.listdir(input_folder):
    if filename.endswith(".tif") and not filename.startswith("._"):
        input_path = os.path.join(input_folder, filename)
        output_path = os.path.join(output_folder, filename)

        print(f"Compressing: {input_path} -> {output_path}")

        command = [
            "gdal_translate",
            "-of", "COG", # Output format is Cloud Optimized GeoTIFF.
            "-a_nodata", "-9999", # Assigns nodata value (important for masking).
            "-co", "COMPRESS=LZW", # Uses LZW lossless compression (good general-purpose choice).
            "-co", "BLOCKSIZE=512", # Sets internal tile size (COGs need tiling; 256 or 512 are common).
            "-co", "RESAMPLING=AVERAGE", # For overviews (pyramid levels), resampling method is AVERAGE.
            "-co", "BIGTIFF=IF_SAFER", # Use BigTIFF only if file > 4GB.
            input_path,
            output_path
        ]

        subprocess.run(command, check=True)

Note

Compression methods
- COMPRESS=DEFLATE → Similar to LZW, often smaller files (slower to read).
- COMPRESS=ZSTD → Modern compression, faster and better ratios than LZW/DEFLATE (requires GDAL ≥ 3.1).
- COMPRESS=JPEG → Good for continuous imagery (satellite photos), lossy but smaller files.

Tiling & block size
- BLOCKSIZE=256 → Smaller tiles, better for web maps (faster random access).
- BLOCKSIZE=512 → Larger tiles, better compression efficiency but slower for small reads.

Predictor (for compression efficiency)
- PREDICTOR=2 → For integer rasters (uses horizontal differencing).
- PREDICTOR=3 → For floating-point rasters (better compression).

Overview (pyramid) settings
- RESAMPLING=NEAREST → For categorical data (LULC, classes).
- RESAMPLING=AVERAGE → For continuous data (climate, elevation).
- RESAMPLING=CUBIC → Higher-quality resampling for imagery.