In recent years, there has been frequent occurrence of extreme weather due to global climate change (Liu et al. 2020; Shi et al. 2020). Watersheds have been heavily modified by engineering measures (reservoir operation, pumping, diversions) as a result of rapid urbanization, hydropower production and irrigation for agricultural uses. The evolution of flood risk under climate change is a critical issue for sustainable development (Reynard et al. 2017). The flood that occurred in July/August 2020 in China’s Yangtze River Basin (YRB) provides insights into this issue.

After the 1998 historic flood in the YRB that caused enormous casualties and property losses (Zong and Chen 2000), China greatly improved its engineering capacity to reduce peak discharges in the main stream of the Yangtze River (YR), especially after the Three Gorges Reservoir (TGR) was completed in 2009. In general, a smaller discharge should correspond to a lower water level at a given station if there is little or no change in the cross section. Nevertheless, it is interesting that during the first flood in 2020, the high water level at Hankou station (28.77 m, July 12th 2020) was the same as that in 1998, even though the observed discharge (56,000 m3/s) was much smaller than that in 1998 (70,700 m3/s) (HBMWR and HBYRWRC 2002). It should be noted that the TGR reduced the discharge by 12,800 m3/s (~ 40%, Fig. 1). Without the TGR, the discharge at Hankou station would have reached 68,800 m3/s (= 56,000 + 12,800 m3/s, close to that in 1998), with a water level much higher than 28.77 m, posing an unprecedented risk of overtopping and levee failure. An important implication is that the destructiveness of a given discharge is now greater than before. There are multiple reasons for this.

Fig. 1
figure 1

Comparison of 1998 and 2020 floods in the YRB

Full size image

First, the backwater effect from downstream cannot be ignored. In early July 2020, the middle and lower reaches of the YR experienced heavy rainfall with a total of about 505 mm (1.6 times than the historic average), causing the elevated water level to expand from downstream to upstream simultaneously with the movement of the rain band. In addition, the backwater effect of the astronomic tides in early July might be another vital factor. Consequently, the highest water levels occurred earlier at the downstream stations and later at the upstream stations on 12th July.

Second, pump drainage measures can suddenly raise the water level in the main stream of the YR. Electric pumping can be employed to drain water in urban areas to mitigate waterlogging during heavy rains. The total pumping capacity within the basin has increased greatly in the past two decades, and the proportion of pumping capacity draining stormwater directly into the main stream of the YR almost doubled from 1998 to 2016 (HBMWR and HBYRWRC 2002; ICMWR and HBYRWRC 2019). Therefore, in extreme events, the volume of stormwater pumped into the main stream in 2020 is likely twice as large as that in 1998 (Fig. 1). Such additional water could cause an increase in the water level. Moreover, pump drainage measures may speed-up the flow of water from urban areas to the main stream (from several days to only several hours), which is an important reason why the highest water levels of different stations along the YR all occurred on the same day.

Third, the buffering effects of lakes on flood detention have been weakened due to human activities in recent decades. The total water storage of all the lakes within the YRB declined by approximately 20% (Fig. 1) from 1998 (37.2 km3) to 2018 (29.8 km3), mainly due to lake reclamation (involving the conversion of water bodies to agricultural/built-up lands) (Hou et al. 2020). Without the regulation of lakes during flood periods, greater volume of water flows into the main stream of the YR in a relatively shorter time, inducing a rapid rise in water level.

In summary, in the YRB, reservoir operation can lower flood risk by reducing river discharge, while pump drainage measures and lake reclamation increase flood risk during heavy rainfall. This may provide a reference for other river basins facing the similar problems to the YRB (e.g., backwater effects, pump drainage measures, and buffering effects) (van Noordwijk et al. 2017; Stevaux et al. 2020). Future extreme rainfall events will prove costly to life and property. Therefore, up-to-date technologies should be developed and utilized to better address this issue. For instance, the ongoing sponge city development in the urban areas along the main stream should put more effort into enhancing infiltration capacity in urban areas, such that waterlogging and fast flow towards the main stream can be mitigated simultaneously (Nguyen et al. 2019). Additionally, accurate flood control methods, such as soil moisture prediction based on deep learning (Fang et al. 2019), rainfall data crowdsourcing (Yang et al. 2019), and streamflow forecasting based on computer vision (Jiang et al. 2018), can provide new avenues for coping with such floods.