Afridi (2025) Real-Time Monitoring and Prediction of Evapotranspiration and Organic Carbon in Soil using IoT Sensors

Identification

• Journal: Macquarie University
• Year: 2025
• Date: 2025-11-19
• Authors: Afridi, Waqas Ahmed Khan
• DOI: 10.25949/30323653

Research Groups

• Macquarie University, School of Engineering (Author: Waqas Ahmed Khan Afridi; Supervisors: Subhas Mukhopadhyay, Bandita Mainali)

Short Summary

This research develops a cost-effective, IoT-enabled system integrating smart sensors and machine learning for real-time monitoring and prediction of evapotranspiration and soil organic carbon. The system demonstrates high accuracy in field deployments, offering scalable solutions for precision agriculture and sustainable water resource management.

Objective

• To develop a cost-effective, scalable, and self-sustainable IoT-enabled system that utilizes smart sensing technologies and advanced machine learning models for real-time monitoring and prediction of evapotranspiration and soil organic carbon, thereby improving precision agriculture and environmental monitoring.

Study Configuration

• Spatial Scale: Field scale, specifically in agricultural and pastoral regions of New South Wales (NSW), Australia, including an agricultural farm and areas around Sydney.
• Temporal Scale: Real-time, continuous monitoring and prediction over an unspecified period (typical for PhD research, implying months to years of data collection).

Methodology and Data

• Models used:
  • Evapotranspiration (ET) Prediction: Support Vector Machine (SVM), Extreme Learning Machine (ELM), M5P Regression Tree.
  • ET Benchmarking: Penman-Monteith, Soil Water Balance (empirical models), and climate ET data from the Australian Bureau of Meteorology.
  • Soil Organic Carbon (SOC) Prediction: Model derived from sensor impedance characterization correlated with SOC percentage, accounting for environmental factors.
• Data sources:
  • Custom IoT-enabled Smart Sensor Nodes: Multi-depth, microcontroller-based sensor nodes for soil moisture (SM), soil temperature (ST), rainfall precipitation (P), air temperature (T), relative humidity (RH), wind speed (u2), and net radiation (Rn).
  • Novel Electromagnetic Sensor: Designed using interdigital and spiral inductance-capacitance geometries on an FR-4 printed circuit board for detecting SOC.
  • Field Data: Real-time environmental data collected from an agricultural farm (air temperature, barometric pressure, wind speed, relative humidity, rainfall, solar exposure, soil temperature, changes in soil moisture).
  • Laboratory Analysis: Walkley–Black (WB) method for validating SOC percentage content in soil samples.
  • Soil Samples: Collected from pastoral regions around Sydney for SOC sensor training and validation.

Main Results

• The developed IoT-enabled smart sensor system was validated for monitoring capabilities, revealing strong dependencies: soil moisture varied significantly with precipitation and net radiation, while soil temperature was largely influenced by air temperature.
• For evapotranspiration prediction, the Support Vector Machine (SVM) model achieved the highest accuracy (R² = 0.97, Root Mean Square Error (RMSE) = 0.19 mm/d, Mean Absolute Error (MAE) = 0.14 mm/d), outperforming Extreme Learning Machine (ELM) and M5P Regression Tree models.
• The novel electromagnetic sensor for soil organic carbon (SOC) demonstrated good overall performance (R² = 0.92, MAE = 0.45) with an accuracy of ±0.5% across 1000 runs, successfully predicting SOC content while accounting for confounding environmental factors like temperature, humidity, and soil moisture.

Contributions

• Development of a cost-effective, scalable, and self-sustainable IoT-enabled smart sensing system for real-time monitoring of soil and environmental parameters, addressing the need for alternatives to expensive commercial tools.
• Integration of advanced machine learning techniques (SVM, ELM, M5P) for highly accurate real-time prediction of evapotranspiration, providing high-resolution insights into soil-water dynamics.
• Introduction and validation of a novel electromagnetic sensor for accurate, real-time, in-situ detection of soil organic carbon, a critical indicator of soil quality.
• The system effectively mitigates uncertainties in existing water monitoring networks and provides a robust solution for precision agriculture and sustainable water resource management, particularly in the agricultural context of New South Wales, Australia.

Funding

Not specified in the provided text.

Citation

@article{Afridi2025RealTime,
  author = {Afridi, Waqas Ahmed Khan},
  title = {Real-Time Monitoring and Prediction of Evapotranspiration and Organic Carbon in Soil using IoT Sensors},
  journal = {Macquarie University},
  year = {2025},
  doi = {10.25949/30323653},
  url = {https://doi.org/10.25949/30323653}
}