Remote Sensing Tree Physiology

Abstract

Transpiration, or plant water loss, is a critical component of water balance and flux for terrestrial systems worldwide, yet uncertainty in large-scale estimates create significant challenges for water resource forecasting (Jasechko 2013, Coenders-Gerrits 2014). Effectively measuring changes in transpiration over time can provide insight into water availability issues facing a tree, a stand, or a landscape. As temperature and precipitation patterns change in the next century and cause shifts in water availability, forest health around the world is likely to change, too, and should be monitored. Currently, transpiration is often monitored at research sites with specialized eddy covariance flux towers, or sap-flux sensors installed in trees that measure the transpiration of individual trees or stands. To measure transpiration frequently across the Earth, a different approach is required. One possible tool for this is ECOSTRESS (Ecosystem Spaceborne Thermal Radiometer Experiment on Space Station). ECOSTRESS is an experiment run by NASA’s Jet Propulsion Laboratory that measures elements of plant transpiration worldwide (Fischer 2020). Attached to the International Space Station, the ECOSTRESS radiometer measures latent heat flux and estimates plant water loss occurring around the world every day. The objective of this project is to investigate and compare ECOSTRESS to more traditional methods of measuring transpiration at three temperate forest research sites in the US and France. Comparing transpiration and evapotranspiration data at each site since the inception of ECOSTRESS in 2018 reveal significant but weak correlations between the two measurements (all r2 < 0.4). ECOSTRESS generally measured much higher transpiration for the days of overpass compared to sap-flux measurements. Time of day of overpass had a significant relationship with the difference between the two measurements, but sub-setting the data to exclude times of day with low transpiration or winter months generally did not improve the correlations, except for excluding morning observations. Based on these observations, ECOSTRESS should be used with caution for non-spatial time-series-type studies. Additionally, there remains a significant spatial difference between the two approaches, as one ECOSTRESS pixel covers about half a hectare.