Author(s): Hoyoung Sun, soojin MOON, Eunji Kim, Boosik Kang
Linked Author(s):
Keywords: Complementary relationship; Evapotranspiration; Bayesian model averaging; Low flow regime;
Abstract: The moisture in the atmosphere implements a key role in evapotranspiration process among all meteorological variables governing hydrological energy and water cycle. In atmospheric dynamics, it functions to maintain the energy balance between land surface and atmosphere through precipitation process. Additionally, the evapotranspiration has significant effects on flow-duration curve of long-term streamflow particularly below ordinary and low flow regime. Most climate models have an evapotranspiration module, but lack of observations limits the validation of model performance. The purpose of this paper is to verify the simulation capability of the evapotranspiration module of the CMIP5 (Coupled Model Intercomparison Project Phase 5) climate models in the lack of observational data. The hypothesis of the complementary relationship between potential and actual evapotranspiration was verified in multi-purpose dam basin and then utilized for assessing the evapotranspiration outputs under reference scenario by the CMIP5 models. The 5 GCM datasets with the highest performance among the GCMs provided by CMIP5 in Korean region were collected and clipped into the study area of the Soyanggang dam basin in Han River, the South Korea. The actual evapotranspiration data of raw GCMs are shown to be overestimated during 1974~2000 of the reference period and the regional biases were corrected through the quantile mapping. The overall quantitative performance of individual GCM model was improved through the generic regional bias correction. However, the additional bias correction was required for balancing long-term mosture budget between precipitation and evapotranspiration expressed through the complementary relationship between the actual and potential areal evapotranspiration. In order to obtain representative scenario, the Bayesian multi-model ensemble averaging scheme was applied and resulted in sufficiently reliable performances in terms of absolute biases and reproducibility of complementary relationship. This methodology allowed reducing the uncertainty arising from the direct use of actual evapotranspiration of GCMs and consequently expected to contribute in improving reliability of the estimation of low flow during non-flood season for dam inflow projection.
Year: 2019