Application of the Random Forest Classifier to Map Irrigated Areas Using Google Earth Engine

Area of Science:

• Agricultural Science
• Remote Sensing
• Geographic Information Systems (GIS)

Background:

• Effective agricultural water management and land use planning require precise data on irrigated areas, especially in water-scarce regions.
• Understanding the spatial extent of current irrigation is crucial for food security initiatives and policy decisions regarding irrigation expansion.
• Existing data on irrigation extent is often insufficient, hindering informed decision-making.

Purpose of the Study:

• To improve the classification accuracy of irrigated areas using machine learning and satellite imagery.
• To assess the impact of the COVID-19 pandemic on agricultural irrigation patterns in smallholder farming areas.
• To provide accurate data for water resource management and land use planning in Mpumalanga Province, South Africa.

Main Methods:

• Applied a random forest machine learning algorithm to Landsat and Sentinel satellite imagery.
• Utilized Google Earth Engine (GEE) for automated big-data processing and R-programming for post-processing.
• Employed the Normalized Difference Vegetation Index (NDVI) to differentiate irrigated from rainfed areas during dry seasons, validated with ground-truth data and very high-resolution imagery.

Main Results:

• Achieved an overall classification accuracy of 88% for irrigated areas.
• Classified irrigated areas for 2020 and 2019 to analyze changes, particularly in smallholder farming regions.
• Demonstrated the effectiveness of integrating multiple data sources and local knowledge for enhanced classification.

Conclusions:

• The study successfully enhanced the accuracy of irrigated area classification through advanced remote sensing and machine learning techniques.
• The methodology provides a robust framework for monitoring irrigation dynamics and supporting sustainable agricultural water management.
• Findings offer critical insights for policymakers and stakeholders in water-scarce regions aiming to bolster food security through irrigation.