Have significant changes occurred in extreme precipitation since the Three Gorges Reservoir impoundment? Open Access

Global environment change mainly includes atmospheric circulation change (e.g., monsoon) at a global scale and surface feature change (e.g., land use change and reservoir construction) at a regional scale (Wang et al. 2010; Shen 2020; Aalijahan et al. 2023; Baima et al. 2023; Rezaei et al. 2023). According to the Intergovernmental Panel on Climate Change, frequent extreme climate events, including extreme precipitation and extreme temperatures, have been increasingly triggered by global environmental change (Aalijahan et al. 2022; Jiang et al. 2022). These events, characterized by low occurrence probabilities and significant deviations from their average states (Yan & Yang 2000; Hu et al. 2007; Aalijahan et al. 2019), directly impact regional water resources and pose serious threats to economic development and human safety (Zhang et al. 2014). Furthermore, extreme precipitation serves as a crucial indicator for forecasting natural disasters such as floods, droughts, landslides, and mudslides (Xie et al. 2018; Sreejith et al. 2024). There was no significant temporal change in the average extreme precipitation during the 20th century globally, but spatial disparities are notable, particularly in middle and high latitudes where extreme precipitation has notably increased (Min et al. 2011; Liu et al. 2024). For instance, both the frequency and intensity of extreme precipitation in East Asia exhibit an increasing trend from inland to coastal areas. In China, numerous cities face flood and drought risks due to frequent extreme precipitation events (Kong et al. 2019; Lan et al. 2024). Consequently, research on extreme precipitation is garnering increasing attention from scholars and is pivotal in supporting natural disaster prevention efforts.

Surface feature changes can significantly influence atmospheric processes, potentially inducing regional climate variability. Among these changes, dam and reservoir construction can alter water surface area and regulate water vapor evaporation through seasonal impoundment (Zheng et al. 2017). Nonetheless, debates persist regarding the impact of reservoir construction on regional precipitation. The rapid expansion of the water surface due to reservoirs can lower surface temperatures, reduce upward airflow, increase water vapor divergence, and consequently lead to decreased precipitation (Miller et al. 2005). However, other scholars have argued that reservoir construction may reduce cloud cover, enhance net radiation, promote more water vapor evaporation, and result in increased precipitation (Yigzaw et al. 2013). Moreover, the precipitation changes induced by reservoir impoundment depend on both temporal and spatial scales. For instance, Wang et al. (2010) observed significant decreases in annual and flood season precipitation but increases in winter precipitation after construction of the Ankang Reservoir. Conversely, around the Xiaolangdi Reservoir in the Yellow River basin, annual precipitation slightly increased within the reservoir area but slightly decreased in the surrounding regions (Ma 2020). Therefore, the impacts of reservoir impoundment on precipitation represent complex outcomes accompanied by considerable uncertainty (Fu et al. 2018).

The Three Gorges Reservoir, the world's largest, commenced operations in 2003. Elevating the water level from 66 to 175 m, it expanded the water surface area to 1,084 km² in the Three Gorges Reservoir Area (TGRA) through anti-seasonal impoundment (Lyu et al. 2018). Previous studies have demonstrated that impoundment in the TGRA has altered both precipitation and temperature to some degree (Chen et al. 2009; Zhao et al. 2022; Wang et al. 2025). Chen et al. (2013) observed cooling effects on summer temperatures and warming effects on winter temperatures due to reservoir impoundment in the TGRA. Although no significant change occurred in annual precipitation before and after impoundment, noteworthy alterations are evident in intra-annual precipitation (Wu et al. 2006; Wu et al. 2012; Chen et al. 2013, 2022). Chen et al. (2022) noted decreased precipitation in summer and winter post-impoundment. Wu et al. (2006) and Wu et al. (2012) reported increased precipitation around the Qinling-Daba Mountains and decreased precipitation upstream of the TGRA following impoundment. It is evident that impoundment has had a discernible impact on the regional precipitation of the TGRA, albeit with inconsistent conclusions due to disparities in data sources and methodologies (Wu et al. 2023; Sun & Gu 2024). Further long-term observations and analyses are warranted.

To date, prevailing studies have primarily focused on variations in annual and seasonal precipitation amounts, with limited attention given to extreme precipitation events. However, as the confluence of climate change, land use alterations, and reservoir construction intensifies, the frequency of extreme climate events is escalating in the TGRA. For instance, the region experienced extreme high temperatures in August 2022 and substantial flooding in July 2023, resulting in losses of 200 million RMB and 20 fatalities (Sun et al. 2023; Zhou et al. 2024). Further investigation is imperative to ascertain whether significant changes in annual and seasonal extreme precipitation result from the impoundment of the Three Gorges Reservoir. Therefore, the objectives of this study are to: (1) analyze annual and seasonal trends of extreme precipitation before and after impoundment of the TGRA, (2) assess trends across different categories (low, medium and high) in extreme precipitation, and (3) discuss potential drivers of extreme precipitation variability.

There are hundreds of rain gauges within the TGRA, which are designed and managed by different departments. To ensure data quality, only those belonging to the China Climate Reference Network (CCRN) are utilized in this study. Managed by the China Meteorological Administration, the CCRN comprises cutting-edge stations furnished with high-quality instruments and adheres to a rigorous observational protocol tailored for detecting climate indicators. There are 21 CCRN stations with observed records over the TGRA. To make the best use of available data with the best spatial coverage, daily data from 21 stations from 1983 to 2022 were selected. The data from these stations without missing values were acquired from the National Meteorological Information Center of the China Meteorological Administration (http://data.cma.cn). Daily precipitation exceeding 1 mm is considered an effective precipitation day. In this study, four seasons are defined as follows: spring (March–April–May), summer (June–July–August), autumn (September–October–November), and winter (December–January–February).

The World Meteorological Organization recommends 27 indices widely adopted in extreme climate studies (Karl et al. 1999). In this study, nine of these indices are chosen to assess extreme precipitation in the TGRA (Table 1). Among them, consecutive wet days (CWD) and consecutive dry days (CDD) measure consecutive precipitation days, indicating the duration of the most prolonged wet and dry spells, respectively. RX1day, RX3day, and RX5day offer insights into short-term precipitation intensity, serving as indicators for potential flooding. R10 and R20 characterize days with moderate and heavy precipitation, respectively. Additionally, R95p and PRCPTOT quantify precipitation amounts, with R95p denoting the magnitude of extreme precipitation events and PRCPTOT representing the annual total. These indices are computed using the RClimdexV3 software, developed by the Climate Research Branch of the Meteorological Service of Canada (Min et al. 2011).

The innovation trend analysis (ITA) method, as proposed by Şen (2012), is employed to identify monotonic trends in extreme precipitation indices (Table 1). The results are obtained through the following steps: (1) dividing the time series of each extreme precipitation index into two equal segments and arranging the two sub-series in ascending order; (2) plotting the first half series on the horizontal axis (x_i) and the second series on the vertical axis (y_i) of the Cartesian coordinate system to generate a scatter plot; (3) the plotted points falling above (below) the 1:1 line indicates a monotonic increasing (decreasing) trend; (4) calculating the trend slope (s) using Equation (1); and (5) conducting a significance test for the trend slope (s) following the approach outlined by Wang et al. (2020).

The monotonic trends in annual extreme precipitation from 1983 to 2022 detected by the ITA method are summarized in Table 2. While annual PRCPTOT shows an insignificant increasing trend with 0.41 mm/year, the remaining extreme precipitation indices are predominantly characterized by negative trends. Among them, the trend slopes of the consecutive precipitation indices (CDD and CWD) and short-term precipitation indices (RX1day, RX3day, and RX5day) exceed the 0.05 significance level. Notably, the results reveal negative trend slopes for both CWD and CDD, suggesting that effective precipitation days become more dispersed and random during the impoundment.

The seasonal trends in extreme precipitation, as determined by the ITA method, are presented in Table 2. Significant downward trends are observed for all extreme precipitation indices in summer (except CDD), distinguishing them from the patterns seen in other seasons. In spring and autumn, the trend slopes predominantly exhibit positive values. For instance, PRCPTOT decreases by −1.95 mm/year in summer, while it increases in the other three seasons, suggesting that the rise in annual PRCPTOT is primarily driven by spring and autumn, rather than summer. Furthermore, significant downward trends for CDD and upward trends for CWD are evident in winter, whereas opposing trends for both indices are observed in other seasons, indicating winter becomes wetter than before due to impoundment in the TGRA. Regarding R10 and R20, upward trends are noted in spring and autumn, while downward trends are observed in summer.

The change rates of annual and seasonal extreme precipitation before and after impoundment in the TGRA are depicted in Figure S5. PRCPTOT increases by more than 10% in spring and winter, while decreasing by 9% in summer. Significant seasonal variabilities are observed for CDD and CWD. CDD in summer and winter increased by 42% and decreased by 22%, respectively. CWD shows a downward trend in all seasons except winter, indicating drier spring and summer and wetter winter conditions following reservoir impoundment. Notably, a downward trend is prevalent across three short-term extreme precipitation indices, particularly for summer RX3day and RX5day, which exhibit the highest decreases of 14% and 16%, respectively. Furthermore, the most substantial increase (26%) and decrease (−25%) for R10 occur in spring and winter, respectively. It suggests that the rise in spring PRCPTOT was driven by heavy precipitation, while the increase in winter PRCPTOT was influenced by drizzle.

The Three Gorges Reservoir's impact on regional precipitation has sparked considerable debate since its inception. The analysis presented in Table 3 indicates that previous studies have explored the precipitation characteristics of the TGRA using datasets of varying lengths. While three studies have delved into extreme precipitation analysis, disparities in trends have been observed. The temporal trends show consistency across most indicators (CDD, CWD, RX5day, R20, and R95), while RX1day and PRCPTOT exhibit divergent patterns. These discrepancies likely stem from variations in series length and station coverage among the datasets. It is worth noting that previous studies were conducted from the aspect of the whole series, overlooking time-series non-stationarity (Sun & Gu 2024; Zeng et al. 2023). Given the alterations in land surface factors such as urbanization and water area after reservoir impoundment, we employed the ITA method to partition the entire dataset into two segments: 1983–2002 (pre-impoundment) and 2003–2022 (post-impoundment). The ITA method facilitates the identification of extreme precipitation trends and their variations across different grades. Our analysis revealed a slight increase (1.0%) in the annual average PRCPTOT, consistent with findings from prior studies (Sun & Gu 2024; Chen et al. 2022).

Moreover, considering the impact of intra-annual water-level fluctuations due to impoundment, this study also investigated seasonal characteristics of extreme precipitation. Significant increases were observed in spring and winter, while a notable decrease was noted in summer, highlighting the need for government attention to local disasters such as landslides and drought resulting from seasonal changes in extreme precipitation. The CDD exhibited pronounced strengthening in summer and weakening in winter. Additionally, it was observed that summer RX1day, RX3day, and RX5day experienced significant declines, particularly among higher grades. While the risk of summer flooding has diminished, potential threats from summer drought and water resource shortages should also be considered (Cui et al. 2022). In conclusion, the abnormal patterns of extreme precipitation in summer and winter are closely linked to the seasonal impoundment of the TGRA.

The variation in extreme precipitation within the TGRA is a complex process influenced by multiple factors, including reservoir impoundment, atmospheric circulation, and land use change (Ding et al. 2012; Li et al. 2019; Hu et al. 2023; Shao et al. 2024). As depicted in Figure S6, the monthly average water level and water area have undergone substantial changes, exhibiting a structural reversal due to impoundment regulation. Since the construction of the Three Gorges Reservoir, the water level has been maintained between 145 m in summer and 175 m in winter, while the water area has fluctuated from 382 to 863 km². The water area is contingent on the water level, thereby influencing regional evaporation and regulating the water vapor cycle (Saggai et al. 2016; Li et al. 2023). Furthermore, Li, Q. et al. (2021) utilized cross-wavelet transform analysis to identify a significant positive correlation between the monthly average water level and precipitation anomaly after 2011. Cui et al. (2021) highlighted the positive correlation between evaporation and precipitation, indicating that water surface evaporation maximizes water vapor in the TGRA during concentrated impoundment in winter. According to Wang & Lin (2022), there have been significant alterations in seasonal average evaporation, with a 9.6% decrease in summer and a 51.3% increase in winter since the Three Gorges Reservoir impoundment. Stagnant water vapor, facilitated by weak monsoons and complex terrain, readily converts into precipitation in winter (Li et al. 2019). Conversely, the decrease in summer precipitation can be attributed to reduced water vapor post-impoundment. Our findings align with these observations, as we noted a 9% decrease in PRCPTOT in summer and an 11% increase in winter, resulting in a drier summer and wetter winter since TGRA impoundment. Moreover, changes in extreme precipitation indices, such as consecutive precipitation days (CDD and CWD) and short-term precipitation (RX1day, RX3day, and RX5day), also correspond to impoundment regulation. From the above discussion, it can be inferred that impoundment regulation significantly impacts seasonal extreme precipitation in the TGRA.

Atmospheric circulations, indicative of global water vapor transport, play a pivotal role in the precipitation patterns of the TGRA (Li et al. 2019; Li, Y. et al. 2021). Studies by Zeng et al. (2023) and Dong et al. (2020) highlight the significant influence of the South Asian Summer Monsoon Index (SASMI) and East Asian Summer Monsoon Index (EASMI) on extreme precipitation indices within the TGRA. Contributions of water vapor to the TGRA precipitation from the Indian Ocean and the Pacific Ocean are reported at 29% and 4.8%, respectively (Shao et al. 2024). As depicted in Figure S7, EASMI exhibits a marginal trend, while SASMI demonstrates a decreasing trend after impoundment. The decline in SASMI may lead to reduced water vapor influx from the Indian Ocean to the TGRA during summer. Li, Y. et al. (2021) also attribute the decreased precipitation over the TGRA from 1979 to 2015 to a substantial reduction in moisture contribution from source regions southwest of the TGRA. Furthermore, Li et al. (2019) note that the Three Gorges Dam impedes low-level water vapor transported by EASMI into the TGRA. Consequently, the combined effects of the Three Gorges Dam and monsoon variations likely contribute to summer precipitation attenuation.

Land use changes have the potential to induce regional climate variations in precipitation by modifying the energy exchange processes between the surface and the atmosphere (Tse et al. 2018). Analysis depicted in Figure S8 reveals that the overall alterations in land use within the TGRA between 2000 and 2020 are not substantial. There is an increase in construction land (2.32%), forest area (1.24%), and water bodies (0.56%), while grassland (−2.57%) and cropland (−1.57%) exhibit decreases. The newly added construction land and forests are predominantly situated in the upstream and downstream regions of the TGRA, attributed to urbanization and the return of farmland to forest. The expansion of construction land may diminish latent heat flux, leading to a reduction in upstream water vapor, whereas increased forest coverage can elevate latent heat flux, thereby enhancing downstream humidity (Li, Y. et al. 2021). The findings of Wu et al. (2023), showing statistically insignificant increases in annual precipitation due to land use changes, align with our results (ITA slope s = 0.41 for PRCPTOT). Our study also revealed strong temporal coherence between seasonal extreme precipitation variations and seasonal water storage area fluctuations. However, the spatial patterns of extreme precipitation changes associated with other land use types require further investigation.

The preceding discussion indicates that extreme precipitation patterns have changed since the construction of the Three Gorges Dam, primarily due to multiple influencing factors, including reservoir impoundment, monsoon circulation, and urbanization. Their changes did not manifest as an overall increase or decrease but exhibited pronounced seasonal variability, particularly in summer and winter. It suggests that earlier concerns regarding large-scale and high-intensity climatic impacts from the dam may have been overstated (Wu et al. 2006; Wu et al. 2012). We also found that both rainfall amount and rainy days in summer have decreased significantly since reservoir impoundment, a trend that may help mitigate flood risks in smaller watersheds within the TGRA. However, it is important to note that reservoir river operations typically maintain lower water levels during summer. If these conditions coincide with prolonged high temperatures and droughts, such as the extreme heat events in July 2022 and August 2024 (Zhou et al. 2024; Yu & Sun 2025), water resource shortages could disrupt residential water use and agricultural irrigation. To address these challenges, optimizing the allocation of water resources in summer within the TGRA operational framework is essential.

This study utilized the ITA method to analyze the annual and seasonal characteristics of extreme precipitation before and after reservoir impoundment in the TGRA. The ITA method visually presents results and trends in different categories (low, medium, and high), yielding valuable insights. Furthermore, the study investigated the driving factors behind changes in extreme precipitation, considering intra-annual impoundment, atmospheric circulation, and land use changes. The conclusions and remarks derived from this study are as follows.

• (1) At the annual scale, significant negative trends were observed in five extreme precipitation indices, including two consecutive precipitation indices and three short-term precipitation indices. However, an insignificant positive trend was noted in the annual PRCPTOT, with a rate of 0.41 mm/year. At the seasonal scale, a noteworthy decrease of −1.95 mm/year was identified in summer PRCPTOT, while significant increases were observed in the other seasons. Among all indices, the most pronounced decreases were observed in summer short-term precipitation indices and winter CDD. These findings suggest that the TGRA experienced drier summers and wetter winters following impoundment.

• (2) The periodic changes in intra-annual water areas emerge as the primary driving factor influencing seasonal extreme precipitation in the TGRA, resulting in reduced water vapor in summer and increased water vapor in winter. The weakened SASMI, which carries less water vapor, also significantly contributes to the attenuation of summer extreme precipitation.

Y.W. conceptualized the work, supervised the project, and wrote the draft. Z.S. wrote the original draft, validated the process, investigated the work. Y.S. developed the methodology and contributed in software. L.F. developed the methodology and contributed in software. L.C., W.C., W.Y., and Q.L. contributed in data analysis.

This work was sponsored by the National Natural Science Foundation of China (No.42201045), Natural Science Foundation of Chongqing, China (No. cstc2021jcyj-msxmX0692; CSTB2023NSCQ-MSX0632), the Science and Technology Research Program of Chongqing Municipal Education Commission (No. KJZD-K202400501; KJZD-K202300507; KJQN202300551), Belt and Road Special Foundation of The National Key Laboratory of Water Disaster Prevention (Grant No. 2023490911)

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.