A moisture-based triangle approach for estimating surface evaporative fraction with time-series of remotely sensed data

Evapotranspiration (ET) is a primary process for water and heat transfer between the land and atmosphere. The spatial contextual information-based surface temperature versus vegetation index triangle (temperature-based triangle) is one of the most famous and widely applied methods for regional ET estimation from remotely sensed data. However, the determination of temporally variable satellite scene-specific dry and wet edges in this traditional triangle is often largely biased due to the limited range of variability of surface soil moisture availability and fractional vegetation cover (or the nonexistence of end-member pixels), or must rely on the auxiliary ground-based measurements, leading to a large uncertainty or difficulty in regional evaporative fraction estimation. This short communication presents a practically operational moisture-based triangle to address the deficiency of the temperature-based triangle, where temporally universal dry and wet edges in this new triangle are determined by making use of the time-series slope of remotely sensed surface temperature versus vegetation index negative relationship over a small window centering the target pixel. Our validation results show that this new triangle outperformed the temperature-based triangle, reducing the root mean square error from 0.19 to 0.16 and increasing the coefficient of determination from 0.44 to 0.53, when the model-estimated evaporative fractions were validated against ground-based eddy covariance measurements at 34 sites across the globe. • A moisture-based triangle is developed with time-series of remotely sensed data. • It avoids using absolute temperature to determine the limiting edges of triangle. • It enhances practicability and operationality of triangle method for EF estimate. • The new triangle outperforms the traditional temperature-based triangle.