Technical Note: Smartphone-based evapotranspiration monitoring

Abstract. Evapotranspiration plays a key role in the terrestrial water cycle, climate extremes and vegetation functioning. However, the understanding of spatio-temporal variability of evapotranspiration is limited by a lack of measurement techniques that are low-cost, and that can be applied anywhere at any time. Here we show that evapotranspiration can be estimated accurately using only observations made by smartphone sensors. Individual variables known to effect evapotranspiration generally showed a high correlation with routine observations during a multi-day field test. In combination with a simple ML-algorithm trained on observed evapotranspiration, the smartphone-observations had a mean RMSE of 0.10 and 0.05 mm/h when compared to lysimeter and eddy covariance observations, respectively. This is comparable to an error of 0.08 mm/h when estimating the eddy covariance ET from the lysimeter or vice versa. The results suggests that smartphone-based ET monitoring could provide a realistic and low-cost alternative for real-time ET estimation in the field.