Defining a new hydrological normal for the Vietnamese Mekong Delta: distinguishing upstream regulation from climate variability

River deltas represent some of the most dynamic, biologically diverse and economically vital landscapes on Earth (Giosan et al., 2014; Syvitski et al., 2009). Acting as the critical interface between terrestrial and marine environments, these systems support dense human populations and serve as global agricultural powerhouses (Bui and Dang, 2025; Rogozhina, 2022). However, their sustainability relies heavily on a delicate equilibrium between sediment deposition, which counters subsidence, and the erosive forces of the ocean (Xue et al., 2012; Zoccarato et al., 2018; Minderhoud et al., 2019). This balance is increasingly threatened by anthropogenic interventions, particularly the construction of mega-dams that fundamentally alter riverine hydrology and sediment transport (Kondolf et al., 2014; Schmitt et al., 2017).

The Vietnamese Mekong Delta (VMD), often described as the “rice bowl” of Southeast Asia, exemplifies this precarious existence (Syvitski et al., 2009; Giosan et al., 2014). Home to nearly 20 million people and accounting for a significant portion of Vietnam’s rice exports and aquaculture, the VMD’s ecological health has evolved over millennia in sync with the Mekong River’s natural flood pulse (Bui and Dang, 2025; Kummu and Varis, 2007). This unique hydrological rhythm – characterized by a distinct monsoon-driven flood season and a dry season – is the engine of the delta’s productivity (Bui and Dang, 2025; MRC, 2019). The annual floods deliver essential nutrient-rich sediments to the floodplains, replenish groundwater aquifers and, critically, provide the hydraulic pressure necessary to flush saline water out of the intricate canal network (Chau et al., 2026; Quang et al., 2026; Wang et al., 2011). Consequently, the livelihoods, culture and biodiversity of the region are intrinsically linked to the predictability and variability of this natural flow regime (Bui and Dang, 2025; MRC, 2019).

Over the past three decades, however, the Mekong River Basin has undergone a radical transformation, becoming a focal point for one of the world’s most intensive hydropower development campaigns (Lu et al., 2021; Ly et al., 2023; Grumbine and Xu, 2011). The construction of a cascade of large-scale dams on the Upper Mekong (Lancang) in China, followed by mainstream and tributary dams in the Lower Mekong Basin (Laos PDR, Cambodia), has created a massive active storage capacity capable of reshaping the river’s hydrology (Lauri et al., 2012; Hecht et al., 2019). The operational logic of these reservoirs is consistent: store water during the wet season to mitigate floods and generate power, and release water during the dry season to maintain electricity production. While this regulation offers theoretical benefits such as flood control and dry-season irrigation, it inadvertently disrupts the natural flood pulse that the VMD ecosystem depends upon (Richter et al., 1996; Li et al., 2017).

Despite the growing body of literature documenting flow alterations in the Mekong (Binh et al., 2020; Räsänen et al., 2017), a scientific debate persists regarding the primary driver of these changes. Is the observed hydrological shift predominantly a consequence of upstream reservoir operations, or is it merely a reflection of basin-wide climate variability, such as changing precipitation patterns driven by El-Niño-Southern Oscillation (ENSO)? Critics of the “dam-driven” hypothesis argue that analyzing discharge solely at the delta’s gateway (e.g. Tan Chau and Chau Doc stations) is insufficient to rule out climatic drivers, as local rainfall anomalies could skew the interpretation. Furthermore, existing studies often lack a rigorous, basin-wide attribution framework that quantitatively disentangles the effects of rainfall variability from flow regulation.

This ambiguity presents a significant challenge for policymakers in the VMD. Without a definitive attribution of causes, developing effective adaptation strategies is fraught with uncertainty (Binh et al., 2020). If the changes are climate-driven, they might be cyclical or random; if they are dam-driven, they represent a structural, permanent shift – a “new normal” that requires a fundamental overhaul of water management paradigms. The lack of a clear, quantified baseline for this regulated era hampers the ability to model salinity intrusion accurately, plan agricultural calendars and negotiate transboundary water release protocols.

The study encompasses the Mekong River Basin, stretching from the Upper Mekong cascade down to the VMD. The VMD is a vast, low-lying coastal plain highly susceptible to the interplay between river discharge and marine tides (Nguyen et al., 2025b; Zoccarato et al., 2018). To ensure spatial representativeness and capture the downstream propagation of hydrological signals, we utilized daily discharge and rainfall data from five key stations along the mainstream: Luang Prabang (Laos, immediately downstream of the Lancang cascade), Mukdahan (Thailand), Pakse (Laos), Kratie (Cambodia – the primary control station before the delta) and Tan Chau (Vietnam – the primary gateway to the VMD). Figure 1 illustrates the study area and the geographical distribution of these hydrological and meteorological stations. The basin’s morphology plays a crucial role, with the mean slope decreasing significantly from the mountainous upper reaches to the flat deltaic plains, facilitating sediment deposition and flow attenuation (Bussi et al., 2021; Wolman and Miller, 1960).

Data were collected for the 25-year period from 2000 to 2024, sourced from the Mekong River Commission and the Vietnam National Center for Hydro-Meteorological Forecasting. Table 1 summarizes the specific characteristics and data duration for these stations. The data set was divided into two periods based on the timeline of major dam commissioning (e.g. Xiaowan in 2010, Nuozhadu in 2014):

1. Growth-dam period (2000–2009): This 10-year period represents the initial phase of major dam construction. While some regulation effects were emerging from dams, the river’s flow regime retained significant natural variability, making it a suitable baseline for a “pre-mega-dam” state.

2. Mega-dam period (2010–2024): This 15-year period covers the full operational phase of the Lancang cascade’s largest reservoirs and subsequent downstream developments, representing the current, highly regulated state of the river.

We employed the Mann–Kendall test and Sen’s slope estimator to detect monotonic trends in both discharge and rainfall time series (Lee and Dang, 2019; Nguyen et al., 2025a). This non-parametric approach is widely accepted in hydro-climatic studies as it is robust against outliers and non-normally distributed data (Lee and Dang, 2019; Huy et al., 2025). A trend is considered statistically significant if the p -value is less than 0.05 (Kendall, 1975; Mann, 1945).

To move beyond simple trend detection and quantify hydrological alteration, we applied the Range of Variability Approach (RVA) of Richter et al. (1996). The target range was defined as the 25th–75th percentile of the baseline period (2000–2009) and post-regulation observations (2010–2024) were assessed against this range. The HAF was interpreted as a magnitude metric, whereas the Mann–Whitney U test was used as a statistical test of distributional differences.

Hydromorphometric descriptors such as basin centre of gravity, drainage density and catchment form were not included as inputs to the RVA/HAF calculation because RVA/HAF is a time-series alteration framework based on observed flow and rainfall regimes. These relatively static basin descriptors are instead used conceptually to interpret spatial attenuation and lag-time effects along the river continuum:

where observed frequency is the percentage of post-regulation years falling within the RVA target range, and expected frequency is 50% because the target range is defined by the 25th–75th percentiles of the baseline period. |HAF| < 0.33: low alteration 0.33 ≤ |HAF| < 0.67: moderate alteration; |HAF| ≥ 0.67: high alteration. Attribution was based on the combined evidence from discharge HAF, rainfall HAF, Mann–Whitney U tests and spatial consistency across stations, rather than on HAF magnitude alone.

The Mann–Kendall analysis reveals a clear decoupling between discharge and rainfall, but the strength of this signal varies by station (Figure 2 and Table 2). Dry-season discharge increased significantly at Luang Prabang (Sen’s slope = +110.0 m³/s/year, p = 0.002), Pakse (+36.8 m³/s/year, p = 0.038) and Mukdahan (+393.4 m³/s/year, p  < 0.001). Kratie showed a non-significant dry-season trend (p = 0.072) but a significant flood-season decrease, while Tan Chau showed no significant trend in annual, dry-season or flood-season discharge. These results support a regulated-flow signal that is strongest upstream and becomes attenuated toward the delta gateway.

Rainfall patterns do not show a corresponding basin-wide increase (Figure 3 and Table 3). Recomputed Mann–Kendall tests indicate no significant increasing trend in dry-season rainfall at Luang Prabang, Kratie, Pakse or Tan Chau (Figure 3(A)). Annual and rainy-season rainfall decreased significantly only at Tan Chau, while the remaining rainfall series were statistically non-significant (Figure 3(B)). Thus, the observed dry-season discharge increases in the upstream and mid-basin stations cannot be explained by a general wetting of the basin.

To further isolate the anthropogenic footprint from natural climate variability, we analyzed flood-pulse dynamics during selected ENSO years, noting that ENSO-monsoon interactions can substantially modulate Southeast Asian rainfall (Webster and Yang, 1992). Table 4 has been regenerated to reconcile station identities and percentage changes. In the La Niña year 2021, the flood peak at Pakse was 21,993 m³/s, 41.3% below the 2000–2008 baseline, while the Kratie flood peak was 32,655 m³/s, 25.5% below the baseline. The previously reported value of 14,617 m³/s was therefore removed because it did not correspond consistently to Kratie in the verified discharge table. Mukdahan was excluded from the ENSO-year comparison in Table 4 because rechecking showed physically implausible event-year values in the available source series, suggesting a source, unit or station-labelling problem that should not be used as evidence.

Dry-season dynamics also indicate regulation, but the magnitude is station-specific. For example, Pakse dry-season discharge was 37.5% above the baseline in 2019 and 4.4% above the baseline in 2024, whereas Kratie was 26.5% above the baseline in 2019 and 0.9% above it in 2024. These revised values support elevated dry-season baseflow in parts of the mainstream, but they do not justify a blanket statement that all stations increased strongly in all recent years.

Although Kratie is not displayed as a separate panel in Figure 4 to maintain figure readability, it is retained in the inferential scope and is fully analysed in Tables 2, 4 and 5.

The RVA analysis provides a statistical quantification of this regime shift (Table 5). At Luang Prabang, discharge during February, March and April exhibits high alteration (HAF = −0.87 to −1.00) with significant Mann–Whitney U results. Pakse, Mukdahan and Kratie also show high dry-season discharge alteration in several months, whereas Tan Chau shows mainly low-to-moderate alteration, with only April significant. This spatial pattern is consistent with strong upstream regulation and downstream attenuation.

Applying the same RVA framework to rainfall (Table 6) requires a distinction between HAF magnitude and statistical significance. Several rainfall months have high-magnitude HAF values, but most Mann–Whitney U tests are not significant; the main significant rainfall change occurs at Pakse in March. The defensible contrast is therefore not that rainfall has no high HAF values, but that rainfall alterations are generally less statistically consistent than discharge alterations. Physically, this means that month-to-month rainfall redistribution can occasionally fall outside the baseline interquartile range, yet it does not reproduce the coherent, multi-station dry-season discharge increase created by reservoir storage and release operations. This contrast supports, rather than proves on its own, the interpretation that the observed discharge alteration is not primarily explained by rainfall variability.

Our findings directly address the attribution question raised in previous studies. Rainfall does not show a basin-wide increase capable of explaining the dry-season flow rise at Luang Prabang, Pakse and Mukdahan, whereas discharge exhibits a coherent regulation signal in the upper and middle basin. At Tan Chau, the absence of a significant trend indicates attenuation rather than absence of upstream influence, because tributary inflows, Tonle Sap storage and delta-channel hydraulics can buffer the upstream signal before it reaches the VMD gateway.

The analysis of stations from upstream to downstream (Luang Prabang, Mukdahan, Pakse, Kratie and Tan Chau) confirms that the regulated signal is strongest in the upper and middle Mekong and becomes less statistically distinct toward the VMD gateway. This revision avoids overstating Tan Chau: the delta gateway does not show a significant monotonic trend in Table 2, but it remains exposed to a regulated upstream hydrograph whose effects are moderated by intervening storage, tributary inflows and travel time.

It is crucial to note the phenomenon of lag time in this propagation. Water released from upstream reservoirs takes days to weeks to reach the VMD, depending on season, flow velocity and intervening storage. The weaker statistical signal at Tan Chau should therefore be interpreted as downstream buffering rather than a contradiction of the upstream-regulation finding. This lag time has practical value because transboundary reservoir-operation information could improve early warning and adaptive water management in the delta.

The results define a “new normal” characterized by: Elimination of the natural flood pulse: The “beautiful floods” that historically brought sediment and fisheries wealth to the VMD are disappearing (Table 4), threatening the delta’s geomorphic stability and aquaculture productivity (Anthony et al., 2015; Uy et al., 2025).

Artificial dry-season flows: While beneficial for salinity repulsion in the short term, permanently elevated low flows disrupt ecosystems adapted to seasonal drought cues (Arias et al., 2014) and may encourage unsustainable agricultural expansion.

Unpredictability: Flash floods in 2019 and 2024 (Table 4) suggest that while average floods are lower, sudden releases from dams can cause unexpected surges, creating a “hybrid” hazard profile that is difficult to predict with traditional statistical methods.

This study provides a revised quantitative baseline for the Mekong mainstream entering the regulated era. By distinguishing discharge and rainfall responses across multiple stations, we conclude that upstream reservoir regulation is the dominant driver of the altered hydrograph in the upper and middle basin, while the signal at Tan Chau is attenuated and not statistically significant as a monotonic trend. The VMD should therefore be managed under a new hydrological normal defined by dampened flood pulses, altered dry-season baseflows and greater dependence on upstream operational decisions.

This reality demands a strategic pivot in management: from relying on historical statistical probabilities to operating within a regulated, albeit less predictable, framework. Future adaptation must prioritize sediment management, transboundary data sharing on dam operations, and flexible agricultural models that do not rely on the predictable flood pulses of the past.