Selection of device to determine temperature gradients for estimating evapotranspiration using energy balance method

• Possibility of replacing the TMCs installed in the actual Bowen stations with RTDs.
• Results contrasted with lysimetric method.
• RTDs have a simpler installation and lower long-term maintenance costs.

Among the methodologies that analyze the energy and mass exchange during the evapotranspiration process, the micrometeorological methods provide one of the best tools to elaborate models and to understand the interaction process between crop and environment. Concretely, the Bowen ratio–energy balance (BREB) method has been used by several authors in the literature to estimate crop evapotranspiration in real time, obtaining accurate results. In order to calculate evapotranspiration, this method requires obtaining average temperature gradients over vegetation cover, which are measured in commercial equipment by the aid of thermocouples (TMCs). However, an alternative to the use of these thermometric gauges is the utilization of resistance temperature detectors (RTDs). These sensors present improvements of accuracy, tolerance and stability, among others, against TMCs. The present paper assess the employment of both thermometric gauges (TMCs and RTDs) when estimating evapotranspiration by using the BREB method. For this purpose, the evapotranspiration of a reference crop (Festuca arundinacea L. cv Schreb.) was determined by means of a weighing lysimeter and estimated with FAO-56 Penman–Monteith equation and BREB method, using for this last both TMC and RTD sensors for obtaining average temperature gradients. The results of the comparative analysis between the ET values measured by the lysimeter and the ET values obtained using the BREB method denote a greater similarity, obtaining a correlation coefficient of 0.958 when using RTDs and of 0.923 with TMCs. The ET values calculated using TMCs and RTDs were also compared with each other to estimate the degree of similarity between them. The results did not show significant differences, obtaining for all analyzed days a squared correlation coefficient above 0.985.