Evaluation of Bowen Ratio Method for Evapotranspiration in Humid Forests Using Flux Tower Data

Evapotranspiration (ET) is a critical component of the hydrological cycle and surface energy budget, and its accurate measurement is essential for elucidating land-atmosphere interactions and the dynamics of water and energy exchange. The Bowen ratio energy balance (BREB) method is a widely utilized approach for observing ET, offering a more economical and less labour-intensive alternative to the eddy covariance (EC) technique, yet its accuracy in humid forested regions remains insufficiently studied. This study evaluates the performance of the BREB method in measuring ET in two subtropical humid forest sites in China (Dinghushan [DHS] and Qianyanzhou [QYZ]), using concurrent EC measurements at intraday and daily scales. The validity of the Bowen ratio (β) was analysed first, with over 90% of days meeting a data validity rate of > 90%. Results show strong correlations between BREB- and EC-derived daily ET (R² > 0.74), yet systematic BREB overestimation persists (MAE: 0.76–0.79 mm/day), particularly during summer high-radiation periods. XGBoost-SHAP analysis reveals that net radiation (Rn) dominates ET discrepancies (feature importance: 0.33–0.37). High Rn values cause the eddy diffusivity ratio (K_h/K_w) to exceed unity or become negative when β_EC > 0, thereby overestimating BREB measurements. Conversely, precipitation (P) usually induces β anomalies, resulting in underestimation of ET, especially in autumn–winter precipitation periods. During midday (11:00 AM–1:00 PM), the BREB method systematically overestimates 30-min mean ET by 0.05 mm (representing 33% overestimation) relative to flux tower measurements, whereas in evening periods (after 5:00 PM), it underestimates ET by ~0.025 mm (42% underestimation). These findings highlight the potential and limitations of the BREB method in humid forested regions, providing insights into its performance for measuring ET under varying climatic conditions.