The exchange rate is often characterized by spatio-temporal heterogeneity, but the spatio-temporal patterns of exchange rate have rarely been quantified, especially in water transfer lakes. This study was conducted from March to July 2021. The tracer data of δ 2 H, δ 18 O (n = 121), 222 Rn (n = 522), Cl (n = 151), TDS (n = 155) in lake water, shallow groundwater (7-10 m), deep groundwater (25-40 m), and an improved single-well radon model were applied in 3 (A, B and C) typical areas ($1 km 2 ) of the Hongze lake. The results show that during the water transfer period (March to May) the rising lake level from normal water level (13 m asl) to the storage level (13.5 m asl), caused the exchange rate to increase from À6.3 Â 10 À7 to 33.2 Â 10 À7 m/s. All tracers in groundwater of A and C were continuously diluted by lake water, but shown a better mixing of the lake water, shallow and deep groundwater in area B with a water transfer channel/river ($À100 m 3 /s). In rainstorm season (June and July), the exchange rate changed from 3.4 Â 10 À7 to À44.8 Â 10 À7 m/s due to the high groundwater table (13-15 m asl) caused by flood and rainstorm. The rainstorm imposed the inflow of both shallow and deep groundwater into lake in river areas (A and B), but only shallow groundwater recharged lake in non-river area (C).Additionally, the exchange rate of the whole lake was estimated by the water balance equation, which varied between À1011 and 458 m 3 /s with an average of À26 m 3 /s.Finally, a conceptual model of exchange rate among lake, shallow and deep groundwater under spatio-temporal heterogeneity is proposed. The findings offer better understanding of the spatio-temporal heterogeneity of lake-groundwater interaction and the effects on lake water balance and recharge systems.