Identification of Paddy Croplands and Its Stages Using Remote Sensors: A Systematic Review.

Area of Science:

• Agricultural remote sensing
• Geospatial analysis
• Food security monitoring

Background:

• Rice is a staple food for a significant portion of the global population.
• Accurate mapping and monitoring of rice paddies are crucial for food security, climate change adaptation, and land management.
• Population growth necessitates enhanced agricultural monitoring systems.

Purpose of the Study:

• To systematically review remote sensing methodologies for mapping paddy croplands.
• To analyze the evolution and classification of these methods from 2010 to 2022.
• To assess the impact of machine learning and common algorithms in paddy mapping.

Main Methods:

• Systematic literature review using the PRISMA protocol.
• Selection of 122 scientific articles (journals and conference proceedings) published between 2010 and October 2022.
• Classification of methodologies based on data source (multispectral, multisource, radar) and analysis of sensor usage and machine learning algorithms.

Main Results:

• Multispectral data (62%) dominates paddy mapping, followed by multisource (20%) and radar (18%).
• MODIS, Sentinel-2, Sentinel-1, and Landsat-8 are the most frequently used sensors.
• Random Forest, Support Vector Machine, and Isodata are the prevalent machine learning algorithms, with increasing use of Sentinel-1 data for its texture and cloud-penetrating capabilities.

Conclusions:

• Remote sensing, particularly with advancements in multisource data and cloud detection, offers robust solutions for paddy mapping across various crop maturity stages.
• The integration of machine learning, especially Random Forest, significantly enhances mapping accuracy.
• Future trends point towards more frequent and higher-resolution multisource solutions for comprehensive global paddy monitoring.