An Experimental Pilot Study of Cranberry Crop Yield Estimation Using Near-Field Microwave Sensing

We present a pilot investigation of near-field microwave sensing for cranberry crop yield estimation using an open-ended rectangular waveguide suspended above the cranberry canopy surface. We compare the performance of two estimation techniques applied to 1.0-1.3 GHz reflection coefficient data measured at 20 unique test sites in a bed in central Wisconsin. The maximum cranberry yield at these test sites was 900 cubic centimeters (cc) per square-foot surface area. A linear discriminant analysis (LDA) with principal component analysis (PCA) and error-correcting-output codes applied to the measured microwave data was able to correctly estimate the volume within 125 cc (roughly 50-60 berries) of the true canopy berry volumes in 71% of the test cases.