Impact of calibrating a low-cost capacitance-based soil moisture sensor on AquaCrop model performance

Area of Science:

• Agricultural Science
• Hydrology
• Sensor Technology

Background:

• Effective irrigation management relies on integrating sensor data with agro-hydrological models.
• Data-scarce regions require parsimonious crop water models and low-cost, maintainable soil moisture sensors for practical application.
• Site-specific calibration of soil moisture sensors is crucial for reliable irrigation management.

Purpose of the Study:

• To calibrate low-cost capacitance-based soil moisture sensors using various methods.
• To assess the impact of sensor calibration on the FAO AquaCrop Open Source (AquaCrop-OS) model's performance.
• To recommend best practices for sensor calibration in crop modeling.

Main Methods:

• Calibration of Spectrum Inc. SM100 soil moisture sensors using multiple least squares and machine learning models with laboratory and field data.
• Field-based piece-wise linear regression was identified as the optimal calibration technique.
• The calibrated sensor data was used to adjust soil hydraulic parameters in the AquaCrop-OS model.

Main Results:

• Field-based piece-wise linear regression yielded the best calibration results (r²=0.76).
• Calibrated low-cost sensor data significantly improved AquaCrop-OS soil moisture simulations and water productivity.
• Machine learning models showed poor field validation due to overfitting; using literature values for calibration was a viable alternative if field calibration was not feasible.

Conclusions:

• Calibrating low-cost soil moisture sensors is essential for improving crop water productivity through enhanced agro-hydrological modeling.
• Field calibration against a reference sensor is recommended for optimal results.
• In the absence of calibration, using literature values for soil hydraulic parameters can be a cost-effective alternative without compromising model performance.