Estimation of Specific Gravity of Potato Tubers Using Dielectric Properties

Abstract

Potatoes are an economically important crop, and their quality is closely related to the starch content, which is typically inferred from specific gravity (SG). Although microwave sensing technologies have been increasingly developed for underground potato detection and quality assessment in recent years, no accurate model has yet been established to link the dielectric properties of potatoes with their key agronomic traits. To address this gap, we developed a model for estimating potato tubers' SG based on their dielectric constant. To construct and validate the model, we conducted SG measurements and dielectric spectroscopy measurements in the frequency range of 0.3 GHz to 3.0 GHz on 250 potatoes of five different types (red, russet, yellow, purple, and chipping potatoes, with 50 samples per type). Out of the 250 data sets, 200 data sets were used for model development, and 50 data sets were used for model validation. A linear regression model was used to summarize the relationship between SG and dielectric constant, where the regression coefficients are expressed as fourth-order polynomial functions of frequency. Experimental results on the 50 validation data sets show that the model achieves high estimation accuracy with mean absolute errors (MAE) less than \(4.8\times10^{-3}\) and mean absolute percentage errors (MAPE) less than 0.45\%. The study of the dielectric properties of potatoes, along with the derived SG estimation model, provides a foundation for the future development of microwave sensing technologies for agronomic trait assessment in the potato production and processing industries. All measured data will be made publicly available upon acceptance of the paper.