Thermal remote sensing of plant water stress in natural ecosystems

Abstract Thermal indices based on canopy temperature (Tc) measurements have been widely used to monitor plant water stress. Such applications have mainly been on homogeneous agriculture crops. Few studies with continuous observations have been conducted on plants in natural ecosystems. Here we examine if the commonly used crop water stress index (CWSI) applies for vegetation in non-managed ecosystems. Two types of parameterization, determined empirically from vapour pressure deficit (CWSI_e) and theoretically from Penman-Monteith equation (CWSI_t), are evaluated on the potential of indicating plant water stress based on Tc and concurrent air temperature (Ta) measurements on three sites under a Mediterranean climate. Results show that although requiring more weather data inputs, CWSI_t does not perform better than CWSI_e, which only requires Tc, Ta and vapor pressure deficit. Weather conditions of wind speed less than 3.0 m s−1 and solar radiation over 600 W m−2 are recommended for Tc for the index calculations. Midday is the optimal time for Tc measurements to estimate water stress for most of the trees. An exponential relationship is found between CWSI_e and soil water content (θ), and a linear relationship exists between CWSI_e and soil water potential (ψ). Values of θ and ψ inverted from CWSI_e agree well with the observations (R2 is about 0.7), indicating that with appropriate measurements of Tc, CWSI_e provides a reliable and non-contact manner for assessing plant water stress in natural ecosystems.