How variable is the variability in annual evapotranspiration?

Evapotranspiration is a key component of the hydrologic cycle, which is often simplified as precipitation = streamflow - evapotranspiration ± change in storage. While streamflow and precipitation are relatively easily quantifiable, both evapotranspiration and storage are more difficult and/or expensive to measure. For that reason, the storage term is usually assumed to be negligible on annual time scales for many hydrologic applications and is subsequently removed from the water balance. Evapotranspiration can then be estimated by subtracting streamflow from precipitation. However, recent research suggests that the variability in evapotranspiration may be overestimated and that, in fact, evapotranspiration may vary less than previously assumed. This would have major implications for the storage term of the water balance. Consequently, recent studies have highlighted the importance of the storage component and the associated carry-over effect of stored water on the streamflow of the following year, further calling into question the practice of calculating evapotranspiration as the residual of a water balance without storage changes. We address the assumed variability in annual evapotranspiration through a comprehensive analysis of evapotranspiration measurements from the worldwide FLUXNET network of eddy covariance towers. We evaluated annual evapotranspiration for different climate zones and land cover classes. There are significant differences in the evapotranspiration variability between different climate zones, with arid regions exhibiting the greatest variability and tropical sites the lowest annual variability. However, for all sites the annual variability in evapotranspiration is much less than the variability in annual precipitation. Further, we examined relationships between evapotranspiration and meteorologic variables commonly used for modeling (potential) evapotranspiration. The lack of strong correlations across the different climate zones suggests potential ramifications for the application of established evapotranspiration models in certain landscapes and climates. A better understanding of evapotranspiration will subsequently translate into more accurate management of water resources.