Abstract: The hydrological regime in East Dongting Lake has changed greatly in recent years; it needs an evaluation on the contributions of different driving factors to the regime change. This study constructs a long short-term memory neural network model to simulate the lake’s water level variations, and uses the indicators of hydrological alteration (IHA) to evaluate the existing hydrological regime. Four scenarios are examined for the major contributing factors: changes in the flow in the Yangtze mainstream and the lake’s four tributaries, and changes in its topographic conditions. The results show that the Yangtze flow changes are the main factor driving the drop in flood season water levels in the lake and the shortened average duration of its low water level pulses. The tributary flow changes are the main factor driving the increase in the number of the lake’s water level reversals. The topographic changes have significantly lowered its water level, and this effect is significant in the dry season. The results help manage the lake’s water resources and evaluate hydrologic regime change factors for other river systems.

Key words: East Dongting Lake, hydrologic regime changes, driving factor, contribution, long short-term memory neural network