Author(s): Jia Wang; Nan Zhang
Linked Author(s):
Keywords: Decomposition method; Yellow River Basin; Unaccounted water; Water balance
Abstract: Estimating available water within a basin is crucial for water resource management. However, limitations in monitoring infrastructure make it challenging to fully account for all inflows and outflows, including unknown inflows and non-revenue water losses. This study focuses on uncounted water and proposes a decomposition-identification-regression framework to quantify the water volumes within a catchment. The goal is to understand the origins and destinations of river water volumes and to demonstrate that this framework is able to elucidate the influence mechanism of available water. Specifically, the framework involves decomposing the unaccounted water time series into regular and irregular components, identifying key factors, and simulating each component using a data-driven approach. Results show that: (1) There is unaccounted water loss about 5.78 billion m3/y of annual average unaccounted water of the Yellow River. Through decomposition, the unaccounted water flow series exhibits a certain periodic pattern and a gentle downward trend, indicating that water monitoring is becoming more accurate and comprehensive. (2) The unaccounted water flow shows sensitivity to local inflow, with an inverse relationship, while it is directly proportional to diversion flow. Uncontrolled inflow and diversion flow significantly contribute to unaccounted water during flood seasons and irrigation periods. (3) The proposed framework is demonstrated to be effective in unaccounted water simulation. By comparing the regression performance of multiple scales, it is found that simulating monthly water flow is superior that of daily flow due to mitigation of the time lag effect.
Year: 2026