ARTIFICIAL NEURAL NETWORK MODEL FOR SOIL MOISTURE ESTIMATION AT MICROWAVE FREQUENCY

This paper reports a neural-network-based methodology to estimate the amount of moisture content in soil at L, S and C frequency bands. A multilayered artificial neural network, using the Levenberg-Marquardt algorithm, is used as the ANN model. The input training data comprise the measured values of dielectric constant of soil in the dry and moist states. Dielectric constant is measured using microwave free-space transmission technique. Measurement has been performed using Vector Network Analyzer (VNA), microstrip patch antenna and soil sample holder. One great advantage with this method is that there is no need to test the pH value of the soil sample, and hence all the associated pre-processing steps, such as drying, pulverizing, can be avoided.