Estimating actual evapotranspiration from an alpine grassland on Qinghai-Tibetan plateau using a two-source model and parameter uncertainty analysis by Bayesian approach

• Published in: Journal of hydrology (Amsterdam) Vol. 476; no. 7; pp. 42 - 51
• Main Authors: Zhu, Gaofeng, Su, Yonghong, Li, Xin, Zhang, Kun, Li, Changbin
• Format: Journal Article
• Language: English
• Published: Kidlington Elsevier B.V 07.01.2013; Elsevier
• Subjects: Alpine grassland; Bayesian analysis; Bayesian statistics; Bayesian theory; Canopies; canopy; China; Conductance; Earth sciences; Earth, ocean, space; eddy covariance; Estimating; Evapotranspiration; Exact sciences and technology; Grasslands; growing season; Hydrology; Hydrology. Hydrogeology; leaf area index; Marine; Mathematical models; model uncertainty; model validation; parameter uncertainty; Partitioning; Qinghai-Tibetan plateau; seasonal variation; Shuttleworth–Wallace model; uncertainty analysis; Qinghai China; Xizang China; Far East; Asia; China; Qinghai-Xizang Plateau; Alpine grassland; Bayesian statistics; Shuttleworth–Wallace model; Evapotranspiration; Qinghai-Tibetan plateau; models; vegetation; covariance; evapotranspiration; Shuttleworth-Wallace model; surface water; grasslands; plateaus; soils; evaporation; uncertainties; performances; seasonal variations; statistics
• ISSN: 0022-1694; 1879-2707
• DOI: 10.1016/j.jhydrol.2012.10.006

► The S–W model was calibrated using Bayesian approach for an alpine grassland. ► The seasonal fluctuation of parameters in the S–W model was identified. ► Seasonal variations of physiologic parameters have effects on ET estimates. ► The diurnal variation of E/ET was related to canopy conductance. ► The daily variation of E/ET was related to LAI and vegetation condition. A Bayesian method was used to fit the Shuttleworth–Wallace model to half-hourly measurements of evapotranspiration (ET) with the eddy covariance technique from an alpine grassland on the Qinghai-Tibetan plateau during the main growing season in 2008, and probabilistically estimated its parameters and predication uncertainties using both dataset-by-dataset and multi-data procedures. This enabled us to reveal the seasonal variations of some physiology-related parameters and the impacts of constant parameters on the performances of the model. Results indicated that the S–W model using the posterior mean parameter values obtained by different procedures all successfully reproduced the observed responses in ET. However, the seasonal variations in the canopy conductance parameter (gmax) should be counted in long-term ET estimating. From simulated results, the daily mean partitioning [i.e. the ratio of the estimated daily soil evaporation (E) over total evapotranspiration (ET); E/ET] was relative low (0.02–0.07 with a mean of 0.04) for the alpine grassland when leaf area index (LAI) was more than 3m2m−2, and was closed related to LAI and vegetation condition. At the diurnal timescale, the canopy conductance was the main factor control the partitioning of ET.