Algorithm for Predicting Rice Yield Potential at Different Sensor Head Orientation.

Canopy reflectance in the visible and near infrared regions of the spectrum has been used to develop vegetation indices for non-destruction estimation of crop biomass .  In rice production,  exposed water background prior to canopy closure can affect canopy reflectance readings.  Therefore, a study was established to determine the effect of sensor head orientation on canopy reflectance of three rice varieties: Catahoula, Neptune and Clearfield.  Sensor readings were taken once every week for five weeks starting at panicle initiation from plots applied with preflood N rates of 0, 34, 68, 102, 136, 170, 204 and 238 kg N ha^-1.  A GreenSeeker^TM handheld sensor was used to measure normalized difference vegetation index (NDVI).  Three different sensor head orientations were investigated: nadir, twisted, and tilted.  Regression analysis was conducted to identify at which sensor head orientation a rice yield potential algorithm with the highest coefficient of determination (r²) can be obtained.  In the early growing season, the twisted orientation had the highest r² value of 0.66. Across the sensing dates, the tilted orientation achieved the most consistent r² values that ranged from 0.55 to 0.58.  At panicle differentiation, all three sensor head orientations obtained relatively similar r² values.  Improving prediction of rice grain yield potential using canopy reflectance at early- or mid-season can be achieved by reducing the influence of exposed water background.