A Workflow for Automated Satellite Image Processing: from Raw VHSR Data to Object-Based Spectral Information for...

Area of Science:

• Remote Sensing
• Agricultural Science
• Geospatial Information Systems

Background:

• Earth Observation (EO) data is crucial for land use/land cover services.
• Increasing satellite resolutions and open access policies generate vast amounts of remote sensing data.
• Timeliness and efficiency are key challenges in processing large EO datasets.

Purpose of the Study:

• To develop a fully automated workflow for processing very high spatial resolution (VHSR) satellite images.
• To generate actionable information for smallholder farming applications.
• To monitor crop development using multi-temporal and multi-sensor imagery.

Main Methods:

• Sequential image processing of VHSR satellite data.
• Extraction of statistical information from agricultural parcels.
• Creation of a crop spectrotemporal signature library.
• Utilized free and open-source software (R, Python, GDAL, FORTRAN, C++, GNU Make).
• Tested on over 270 VHSR images (WorldView, QuickBird, GeoEye, RapidEye) across five study areas.

Main Results:

• A robust, automated workflow for processing large archives of satellite imagery.
• Generation of a crop spectrotemporal signature library for agricultural monitoring.
• Demonstrated capability to follow crop development through the season.
• Successful application in smallholder farming contexts.

Conclusions:

• The developed workflow effectively processes large volumes of EO data for smallholder agriculture.
• Open-source tools enable efficient and timely generation of actionable insights from remote sensing data.
• This approach enhances the utility of EO for agricultural monitoring and decision-making.