Reply on RC2

Abstract. Understanding the impacts of climate change and human activities on large endorheic lakes is crucial for sustainable water management, yet quantitative attribution remains a significant challenge. This study introduces the Hydrological Attribution and Analysis Framework (HAAF), a novel three-stage methodology, to provide a comprehensive explanation for the nearly-centennial (1931–2024) water balance dynamics of Lake Balkhash. The HAAF first establishes a high-fidelity hydrological reconstruction using a Physics-Informed Machine Learning (PIML) model, then employs the Budyko framework to attribute runoff changes, and finally links these catchment-scale drivers to the lake's terminal water balance. Our results confirm the robustness of the PIML model in simulating historical runoff (KGE > 0.75). The attribution analysis then reveals a complex interplay of competing forces. During the intensive intervention period (1970–1990), a substantial human-induced runoff reduction of -9.21 km³ completely masked a significant climate-driven wetting potential (+6.13 km³), triggering the lake's sharp decline. In the recent period (1991–2024), the basin's hydrology has been governed by a fragile stalemate in which a massive, climate-driven potential for increased runoff (+10.80 km³) was almost entirely neutralized by the persistent negative impact of human water use (-11.36 km³). At the lake level, this translated into an apparent stability sustained only by a favorable climatic subsidy. Future projections under various climate scenarios indicate that this climatic buffer is transient and unlikely to persist, exposing the lake to a high risk of rapid decline. We conclude that the recent stability of Lake Balkhash is not a sign of systemic recovery but a "masked vulnerability." This highlights the urgent need for sustainable and forward-looking water management strategies that account for these underlying, competing drivers.