Estimation of evapotranspiration in a mid-latitude forest using the Microwave Emissivity Difference Vegetation Index (EDVI)

We developed an algorithm to estimate evapotranspiration (ET) from dense vegetation covered area from the first principle of surface energy balance model by using satellite retrieved Microwave Emissivity Difference Vegetation Index (EDVI). This algorithm can be used under both clear sky and cloudy sky conditions. Long term seasonal trend of EDVI is linked to variance of canopy resistance due to the interrelationship among leaf development, environmental condition and microwave radiation. Short term changes of EDVI caused by synoptic scale weather variations is used to parameterize the responds of vegetation resistance to the quick changes of environmental factors including water vapor deficit, water potential and others. The performance of this algorithm was test at the Harvard forest site by using satellite measurements from the SSM/I F13 and F14 sensors. Validation at the site with 169 samples shows that the correlation coefficient (R2) between estimated and observed ETs is 0.83 with a mean bias of 3.31 Wm− 2 and a standard deviation of 79.63 Wm− 2. The overall uncertainty of our ET retrieval is ~30%, which is within the uncertainty of current ground based ET measurements. Furthermore, the estimated ET in different local times (up to 4 times per day) successfully captured the diurnal cycle of ET. It is the first time that the diurnal variations of vegetation–atmosphere interactions were directly monitored from space. This study demonstrates that the technique reported here extends the current satellite capability of vegetation property and ET flux remote sensing from daytime, clear-sky conditions to day and night times and from intermediate leaf area index (LAI) to all range of vegetation states.