As distortion of the hydrologic cycle intensifies, the importance of water resource management has increased; however, quantitative evaluation criteria and analytical frameworks for water loss factors remain insufficient. In particular, existing studi...
As distortion of the hydrologic cycle intensifies, the importance of water resource management has increased; however, quantitative evaluation criteria and analytical frameworks for water loss factors remain insufficient. In particular, existing studies on water loss mainly rely on reservoir-based empirical equations or limited observations, which restrict their applicability to river conditions and introduce uncertainty in estimation.
This study aims to establish a methodology to quantitatively evaluate water loss occurring within river flow systems. Evaporation and infiltration are defined as the dominant components of water loss. Evaporation is estimated using the Penman Combination Equation (PCE) and the Penman Wind Function (PWF). To account for parameter uncertainty, a statistical ensemble framework is constructed by combining Kernel Density Estimation (KDE) and Copula. Infiltration is estimated using a water balance approach based on upstream–downstream discharge differences within a defined control volume along a river reach. This approach provides a practical way to quantify infiltration losses directly from observations without explicitly modeling groundwater flow or saturated–unsaturated processes. The methodology is applied using reservoir observations, surface evaporation data, and full-scale channel experiment data.
The proposed framework has several distinguishing features. First, parameter calibration and uncertainty quantification are treated separately, allowing structural representation of parameter uncertainty. Second, the “reservoir-to-river transfer problem” is explicitly addressed by evaluating the applicability of empirical equations derived under still-water conditions to flowing river systems. Third, the framework is designed to be extensible under varying hydrometeorological conditions.
The results demonstrate that (1) reservoir-based empirical equations can be applied to river conditions when parameter uncertainty is properly considered, and (2) the combined physical–statistical approach improves the reliability of water loss estimation. In particular, representing water loss as a probabilistic range rather than a single deterministic value provides a basis for quantifying uncertainty in interpretation.
The proposed methodology can be applied as a practical framework for quantifying water loss in rivers with limited observations. It can also support basin-scale water balance analysis and water resource management by providing quantitative decision-making criteria that incorporate uncertainty. Furthermore, the approach is expected to contribute to the advancement of hydrological modeling and integrated water cycle analysis.