Calibration and Tests of Commercial Wireless Infrared Thermometers

Applications of infrared thermometers (IRTs) in large agricultural fields require wireless data transmission, and IRT target temperature should have minimal sensitivity to internal detector temperature. To meet these objectives, a prototype wireless IRT system was developed at USDA Agricultural Research Service, Bushland, Texas, and commercialized by Dynamax, Inc., Houston, Texas. The objective of this article was to calibrate and test the Dynamax, Inc. system. Following deployment in an irrigated field during the 2015 crop season, 26 IRTs were calibrated and tested in a temperature-controlled room. The IRTs measured a black body target temperature controlled at 15°C to 55°C in 5°C increments, and for each range of black body temperatures, ambient room temperatures were controlled at 15°C, 20°C, 25°C, 35°C, and 45°C under isothermal and steady-state conditions. Discrepancies between uncalibrated IRT and black body temperatures varied by IRT and had root mean squared errors (RMSE) between 0.25°C and 1.51°C, mean absolute errors (MAE) between 0.19°C and 1.17°C, and mean bias errors (MBE) between -0.66°C and 0.16°C. A calibration equation was derived from the longwave radiation energy balance of the IRT internal detector, and sensor-specific calibrations reduced discrepancies for all IRTs, with RMSE between 0.16°C and 0.28°C, MAE between 0.12°C and 0.21°C, and absolute MBE less than 0.05°C. A generic calibration was derived by pooling all sensor-specific calibrations, and reduced discrepancies for all but five IRTs, but these were very marginal compared with no calibration. Therefore, the generic calibration did not appear justified, but sensor-specific calibrations were justified for most IRTs. The IRTs were again deployed in the irrigated field and measured corn canopy temperature in 2016. Crop evapotranspiration (ETc) was calculated using measurements from one IRT and compared to ETc measured by a large weighing lysimeter. The choice of calibration (none, generic, or sensor-specific) had little impact on calculated ETc, which was likely related to a limited range of target and sensor body temperature differences in the field (mostly +10°C), in contrast to those in the temperature-controlled room (up to +40°C)