Deriving Daily Transpiration from Instantaneous Measurements in Almond Orchards with Varied Water Stress and Production Systems

For the purpose of managing irrigation water and raising agricultural water yield, the daily transpiration (T_d) monitoring is essential. The estimation of evapotranspiration (ET) and its components, which include crop transpiration (T) and soil evaporation (E), for a variety of crops has shown to be robust when using remote sensing energy balance models. However, as measurements from remote sensing are instantaneous, daily upscaling methods are required in order to estimate T_d from remote sensing models. Although upscaling methods for daily ET have been validated by multiple studies, those techniques have not been validated for estimating T_d separately in woody crops. The purpose of this study is to assess upscaling methods for recovering T_d in almond crops with varying water status and production systems. Sap flow sensors were used to monitor the T in-situ (T-SF), allowing for a continuous measurement for each plant every 15 minutes. The stem water potential (Ψs), stomatal conductance (g_s) and leaf transpiration (E_leaf) were also measured at 7:00, 9:00, 12:00, 14:00 and 16:00 solar time for two days in the same trees where sap flow sensors were installed. High-resolution images were used to estimate hourly T using the two-source energy balance model (TSEB). The upscaling methods were evaluated with in situ sap flow data and then implemented to the TSEB estimations. The evaluated upscaling methods were the simulated evaporative fraction variable (EFsim), irradiance (Rs) and potential evapotranspiration (ETp) methods. As a results, the EFsim and ETp methods were more correlated with T-SF, reducing the observed potential underestimation using the Rs method. The improvement was especially important at midday in the tress subjected to severe water stress, where the EFsim reduced the error by 17.61% and the ETp reduced it by 10.6% compared to the Rs method, respectively. Nevertheless, the daily T-SF revealed significant differences across production systems that the daily upscaling methods used in the TSEB were unable to identify. One issue in determining T_d on surfaces with discontinuous architectural features was the insufficient sensitivity of daily TSEB between production systems. This issue might be resolved by applying more sophisticated ETp models or enhanced ETp as an upscaling parameter, since ETp can account for variations in canopy structures that have an impact on daily T curves.