Improving continuous monitoring of evapotranspiration with future thermal infrared missions

Remote Sensing (RS) in the thermal infrared (TIR) provides useful information on evapotranspiration (ET). However, current satellites that are available for monitoring ET at a spatial resolution lower than 100 m have long revisit intervals (16 days for Landsat). Cloud occurrence also reduces the number of available images hindering the accuracy of continuous monitoring of ET (requiring interpolation between available ET estimations from RS data). Future satellite missions providing high spatial resolution data every 1 to 4 days are under study : TRISHNA by CNESand ISRO (France/India) and the Land Surface Temperature Monitoring mission (LSTM) by ESA/COPERNICUS.

We analyzed the impact of satellite revisit on the uncertainty in monitoring ET over Europe by considering combinations of climate, land use, revisit characteristics and errors in estimating ET from RS data. We analyzed a large range of crop types / soil / climate / revisit combinations by using synthetic data of ET as simulated using the ISBA-A-gs land surface model (while previous studies were based on flux tower measurements and considered only a limited range of situations). Revisit scenarios were defined from the orbital characteristics of TRISHNA and LSTM in comparison to nominal scenarios with revisit between 1 day and 16 days. We also introduced errors in ET estimation from RS data at the time of acquisition (depending on the surface energy balance model used to derive ET and on the accuracy of RS measurements).         

As expected, the uncertainty in monitoring ET increased significantly with the revisit period when cloud occurrence increased (Figure). However, the impact of cloud regime was lower at higher latitudes because the frequency of image acquisitions increased with the latitude. The impact of the uncertainty in estimating ET at the time of image acquisition was the main driver of the accuracy, in particular in southern Europe.

Figure: uncertainty in ET estimation (ΔET) depending on revisit, crop type, location and day of the year (DOY) as calculated for an error in estimating ET from RS data of 0.6 mm d^-1.