Singh et al. (2025) Remote Sensing and Machine Learning for Irrigation Management in Potato Cultivation: A Systematic Review

Identification

• Journal: American Journal of Potato Research
• Year: 2025
• Date: 2025-12-01
• Authors: Ravinder Singh, Lakesh Sharma, Hardeep Singh, Lincoln Zotarelli
• DOI: 10.1007/s12230-025-10029-3

Research Groups

• Soil, Water, and Ecosystem Sciences Department, University of Florida, Gainesville, FL, USA
• West Florida Research and Education Center, University of Florida, Jay, FL, USA
• Horticultural Sciences Department, University of Florida, Gainesville, FL, USA

Short Summary

This systematic review synthesizes research from 49 peer-reviewed studies (2000–2025) to evaluate the application of remote sensing and machine learning for irrigation management in potato cultivation, highlighting current trends, limitations, and future research directions to enhance sustainable water use.

Objective

• Identify and classify studies utilizing remote sensing and/or machine learning for potato irrigation management.
• Synthesize the technological and methodological approaches used in these studies.
• Highlight emerging trends and knowledge gaps in the field.
• Propose future research directions to enhance sustainable water management in potato production.

Study Configuration

• Spatial Scale: Variable, ranging from controlled experimental plots and greenhouses to field-scale and regional applications.
• Temporal Scale: January 2000 to April 2025 (for reviewed studies).

Methodology and Data

• Models used:
  • Machine Learning: Random Forest (RF), Support Vector Machine (SVM), Partial Least Squares Regression (PLSR), XGBoost, Backpropagation (BP) neural networks.
  • Remote Sensing/ET Models: Surface Energy Balance Algorithm for Land (SEBAL), Mapping Evapotranspiration at high Resolution with Internalized Calibration (METRIC), Two-Source Energy Balance (TSEB-PT), FAO Penman-Monteith equation, Soil Water Balance (SWB).
  • Indices: Normalized Difference Vegetation Index (NDVI), Normalized Difference Water Index (NDWI), Normalized Difference Moisture Index (NDMI), Land Surface Temperature (LST), Crop Water Stress Index (CWSI).
• Data sources:
  • Databases: Scopus, Web of Science (following PRISMA guidelines).
  • Remote Sensing Platforms: Unmanned Aerial Vehicles (UAVs), Satellites (MODIS, Landsat, Sentinel, SMAP, SMOS, AMSR2), Proximal sensors.
  • Sensor Types: Multispectral imaging, thermal imaging, hyperspectral imaging, microwave sensors (L-band radiometers), RGB digital cameras.
  • In-situ measurements: Gravimetric sampling, neutron probes, lysimeters, tensiometers, time domain reflectometry (TDR), frequency domain reflectometry (FDR), sap flow, leaf-level measurements.
  • Other: Wireless sensor networks, IoT-enabled tools, meteorological data.

Main Results

• A total of 49 studies published between 2000 and 2025 were analyzed, showing a marked surge in publications since 2020.
• Evapotranspiration (ET)-based approaches were the most prevalent (47%), followed by plant water stress detection (39%), soil moisture monitoring (8%), and crop coefficient (Kc) approaches (6%).
• UAV-based multispectral and thermal imaging are increasingly utilized for high-resolution monitoring of ET, soil moisture, and canopy temperature.
• Machine learning algorithms (RF, SVM, PLSR, XGBoost, BP neural networks) demonstrated strong predictive capabilities for crop water needs and stress detection, with R² values up to 0.90 for Leaf Water Content (LWC) estimation.
• Proximal sensors were most frequently used for plant water stress detection, while satellite-based remote sensing primarily supported ET estimation. UAV platforms showed balanced application across various irrigation parameters.
• Key research gaps include limited field-scale validation of models, underutilization of UAV-based hyperspectral imaging, insufficient research on dynamic Kc estimation, challenges in root-zone moisture monitoring, and issues with system interoperability and scalability.

Contributions

• Provides the first comprehensive and up-to-date systematic review specifically on remote sensing and machine learning for irrigation management in potato cultivation.
• Synthesizes a broad range of studies (49 peer-reviewed articles from 2000-2025), offering a consolidated evaluation of methodological frameworks, sensor applications, and implementation barriers.
• Identifies critical knowledge gaps and limitations, such as the need for more extensive field-scale validation, improved root-zone moisture modeling, and dynamic crop coefficient (Kc) estimation.
• Proposes clear future research directions, including sensor fusion, robust model development across diverse environments, and user-focused decision support tools, to foster sustainable and resilient potato irrigation systems.

Funding

This study was conducted without financial support from any public, private, or non-profit funding sources.

Citation

@article{Singh2025Remote,
  author = {Singh, Ravinder and Sharma, Lakesh and Singh, Hardeep and Zotarelli, Lincoln},
  title = {Remote Sensing and Machine Learning for Irrigation Management in Potato Cultivation: A Systematic Review},
  journal = {American Journal of Potato Research},
  year = {2025},
  doi = {10.1007/s12230-025-10029-3},
  url = {https://doi.org/10.1007/s12230-025-10029-3}
}