Maximising Value of Frugal Soil Moisture Sensors for Precision Agriculture Applications

Abstract

Rugged soil moisture sensors with stable measurement profiles are usually expensive for a common farmer. The moisture readings for frugal, inexpensive, and often resistive, sensors are usually jittery where the sensor health tends to degrade over a period of time. Failing to catch the reduced reliability due to degraded sensor health would lead to imprecise irrigation decisions. We propose a soil moisture calibration and health management system that adds a layer of reliability to a distributed IoT-edge solution involving a frugal soil moisture sensor to help make its adoption pervasive for precision farming applications. Our approach offers a multi-step process based on artificial intelligence that maximizes the value of a low-cost soil moisture sensor. The sensor is first calibrated to give volumetric water content (a derived irrigation-related parameter) equivalent to a rugged sensor with a 5% root mean square error (RMSE). A classification model is then developed to predict the health of the sensor based on the sensor values and image analytics with an overall accuracy of 93%. We believe the outcomes would significantly help increase the adoption of precision agriculture, especially in emerging geographies, by making technology-driven intelligent solutions more affordable.