Inference of foliar temperature profile of a vineyard using integrated sensors into a motorized vehicle

Inferring foliar temperatures is motivated by the high cost related to measure each plant's temperatures in a vineyard. Foliar temperature is correlated with the plants water potential, which allows farmers to take irrigation decisions. In order to deal with this situation, we utilized information obtained from several temperature infrared sensors integrated into a motor vehicle that runs a vineyard taking measurements at different heights. This data was interpolated with Kriging and IDW methods. These models were evaluated through RMSE (rootmean-square error) and rRMSE (relative root-mean-square error) as performance measures. Main results indicate that the Krigning approach presents a better performance in comparison with IDW. It is possible to see that the sensors located at lower altitudes show less measured error as RMSE. It was also observed that the spatial distribution of errors is homogeneous, with most of these errors less than or equal to the average RMSE. Finally, no significant changes are observed in the model performance when halving the data employed for interpolation. From these assessments, optimum operation conditions for the monitoring equipment were proposed.