The study area is located in the downstream part of the Cambodian Mekong River basin (excluding the Tonle Sap Lake region), also known as the “Cambodian Mekong floodplain” (Fig. 1). The area is about 27 760 km2 and extends along the Mekong mainstream from Kratie Province to the Cambodia–Vietnam border. It covers parts of 12 provinces in Cambodia and one province in Vietnam (Tay Ninh) but does not extend into the Vietnamese Mekong Delta region (see division in Fig. 1).

A major part of the Cambodian Mekong floodplain is characterized by a flat terrace and low-lying grounds with gentle slopes that contain many depressions and lakes, except for the upper parts of the Prek Thnot and Prek Chhlong tributaries, which contain steeper terrain. Hydrological conditions within the area are dominated by the seasonality and year-to-year variability of the Mekong flow regimes. The wet season runs from June to October, and the dry season runs from November to May. During the wet season, the characteristics of the floodplain and Tonle Sap Lake play a vital role in flood peak attenuation and regulation, temporarily storing and later conveying water across the vast low-lying areas. During the wet season, water flows from the Mekong main stream into the Tonle Sap Lake, but this flow is then reversed in the dry season. This illustrates the highly complex hydrological system at play throughout the region and the seasonal variations that characterize the ecological and agricultural landscape.

Within our historic baseline period of 1971–2000, the annual average temperature across the study area varies from 26.9 to 28.2 ∘C, with mean monthly temperatures between 30 ∘C during the hottest months (April and/or May) and 26 ∘C in the coldest month (January). Average annual rainfall across the study area during the same period varies between 1100 and 1850 mm, with mean monthly rainfall ranging between 250 mm in the wettest months (May–June) and 10 mm in the driest (February).

We used a hydrological–floodplain model combination (Fig. 2), consisting of the distributed hydrological model IWRM-VMod (Lauri et al., 2006), the floodplain propagation model MIKE 11 (Dung et al., 2011), and the flood extent and duration model IWRM-Sub (MRC, 2018a) (Fig. 2). First, the IWRM-VMod model with a resolution of 5 km × 5 km (see extent and hydrological processes in Fig. 2a) was used to simulate the entire Mekong basin's flow response to hydropower developments, irrigation expansion, and climate change impacts around the year 2050. We used the model runs, both baseline and scenarios, from Hoang et al. (2019). From the hydrological model we derived the boundary condition discharges that were used to drive the 1D flood propagation model MIKE 11 (as constructed and employed in Triet et al., 2017, 2020) in order to obtain the initial floodplain conditions, water levels, and fluctuating discharge of the Tonle Sap River. MIKE 11 extends over the entire Mekong Delta down to the South China Sea, where sea level is used as another boundary condition. MIKE 11 also includes a detailed description of the channels, canals, and sluice gates in the delta (Triet et al., 2020). The results from MIKE 11 in turn were used as boundary conditions to the detail-scale (1 km × 1 km) floodplain hydrodynamic IWRM-Sub model. The IWRM-Sub model is a flood model that also has hydrological processes (i.e. precipitation, evaporation) in it, making it ideal for large floodplain modelling in monsoon climate. It uses the 2D depth-averaged Navier–Stokes and St. Venant equations to propagate a flood wave out into the floodplain from the water level points passed as boundary conditions (MRC, 2018a).

The IWRM-Sub model was applied to Cambodian floodplains for the Mekong River Commission's (MRC) Council Study (MRC, 2018a). It is based on the SRTM 90 m topographical map (Jarvis et al., 2008), a soil type map (FAO, 2009), and a land use map (GLC2000, 2003), all aggregated to 1 km × 1 km resolution (Table 1). Geospatial data and river cross-section data were retrieved and added from the Mekong River Commission (MRC). The future climate scenarios are based on an ensemble of five GCM projections of precipitation and temperature taken from the CMIP5 suite of models (ACCESS, CCSM, CSIRO, HadGEM2, and MPI). Whilst the CMIP6 collection has now superseded the CMIP5 model results, an analysis of the differences between model collections shows consistent mean values for both precipitation and temperature across our study area for both wet and dry seasons (Supplement Table S1).

Flood extent maps for calibration and validation were derived from Landsat images using a sophisticated water detection algorithm developed and optimized for the Lower Mekong region (Donchyts et al., 2016). All IWRM-Sub model inputs and their brief description are presented in Table 1, while input data for IWRM-VMod are detailed in Hoang et al (2019) and MIKE 11 in Triet et al. (2020).

We adapted and applied the existing IWRM-VMod (Hoang et al., 2019), MIKE11 (Triet et al., 2017), and IWRM-Sub (MRC, 2018a) models to assess the smaller-scale cumulative impacts of future development plans and climate change on the Cambodian Mekong floodplain. Here we enhanced the reliability of these existing models, particularly in the Cambodian Mekong floodplain, by advancing the predictive accuracy of the hydrology (recalibration), accounting for multiple calibration stations (four stations), and validating flood extents against satellite imagery, as described below.

Our initial model setup describes the current state of the floodplain for the historic baseline period of 1971–2000, which we further calibrated and validated against observations of water discharge and water level taken at Kratie, Kampong Cham, Chroy Changvar, and Neak Loeung hydrological stations (see locations in Fig. 1). The model performance was systematically quantified and evaluated based upon the Nash–Sutcliffe efficiency (NSE), percent bias (PBIAS), ratio of the root mean square error to the standard deviation of observed data (RSR), and coefficient of determination (R2).

The use of 1971–2000 as our baseline represents the hydrological state of the basin before major alterations were introduced well (Soukhaphon et al., 2021). Including years after 2000 in our baseline would introduce significant hydrological and irrigation influences that would prohibit a thorough examination of these in isolation as part of our simulations.

Flood extent maps generated from the IWRM-Sub model were validated for the same period against satellite-based flood extent maps generated by the Surface Water Mapping Tool (SWMT). The SWMT is a Google Appspot-based online application developed by Donchyts et al. (2016). A stack of Landsat (4 and 5) data were generated using SWMT from 1984–2000. This stack of images were then used to generate a water index map using the modified normalized difference water index (MNDWI) (Xu, 2006) to distinguish between water and non-water areas, which were then adjusted to account for dark vegetation and hill shadows using a height above nearest drainage (HAND) map (Rennó et al., 2008). Figure S1 in the Supplement illustrates all procedures of the Surface Water Mapping Tool.

To evaluate the model performance for flood inundation maps, we applied three indices: recall, precision, and the ratio between simulated and observed flood extent areas. Recall evaluates what proportion (0–1) of the flood derived from remote sensing images is identified by the simulation. Precision evaluates what proportion of the simulated extent agrees with the remote sensing data. If the simulated extent overlaps the observed extent area perfectly, recall, precision, and the ratio of extents become 1.

Once the IWRM-Sub model was successfully calibrated and validated, we modulated the inflow at Kratie and at the confluence of the Tonle Sap River with the main Mekong channel to represent the upstream impacts of multiple development and climate change scenarios (see Sect. 2.4). We then simulated the Cambodian Mekong floodplain's hydrological and flood conditions (flood extent, flood depth, and flood duration) for each scenario.

The scenario setup that we adopted for our study is the same as that described in Hoang et al. (2019). The baseline (1971–2000) represents the Mekong basin at a time before significant alterations to the hydrological functioning of the catchment have occurred through infrastructural development. We then defined 11 development scenarios that cover each of the three main drivers of hydrological change in isolation (hydropower, irrigation, and climate change), as well as combinations of these together. For future scenarios, we used climate data from an ensemble of five GCMs (ACCESS, CCSM, CSIRO, HadGEM2, and MPI) for the years 2036–2065 and considered Representative Concentration Pathway (RCP) levels 4.5 and 8.5. These GCMs were selected based on their performance in reproducing historic temperature, seasonal precipitation, and climate extremes in the Mekong region. The GCM data were downscaled using bilinear interpolation and statistically bias corrected using a quantile mapping method. For full details see Hoang et al. (2016, 2019). The sea level boundary condition was adjusted by 43 cm for future scenarios to account for the combined effects of sea level rise and deltaic subsidence, taken as the average of the range estimated by Manh et al. (2015), i.e. 22–63 cm. This value was used for both RCP4.5 and RCP8.5 as the climate change component of sea level rise for our study period taken from IPCC (2014) is relatively consistent across RCP scenarios (RCP4.5: 19–33 cm; RCP8.5: 22–38 cm). Our hydropower development scenario includes 126 dams on both main streams (N= 16) and tributaries (N= 110) of the Mekong, equivalent to a total active storage of 108 km3, all of which is planned to be active between 2036 and 2065. Dam simulation was based on the optimization scheme developed by Lauri et al. (2012), which calculates each dam's operating rules separately in a cascade, aiming to maximize productive outflows (i.e. outflows through the turbines), thus maximizing hydro-power production. The optimized dam operation rules were later validated against observations by Räsänen et al. (2017). We also included two irrigation scenarios, a high- and low-expansion version, using the global projected irrigation expansion scenarios by Fischer et al. (2007) applied to the baseline irrigation extent taken from the MIRCA – “Global Dataset of Monthly Irrigated and Rain-fed Crop Areas around the Year 2000” (Portmann et al., 2010). A list of scenarios and their notation are presented in Table 2, and a thorough description and justification for these scenarios can be found in Hoang et al. (2019).

The Mekong-basin-wide IWRM-VMod hydrological model was calibrated and validated against discharges in various stations, with very good performance: validation period NSE at Nakhon Phanom station of 0.74 and at Stung Treng station of 0.64 (Hoang et al., 2019). MIKE 11 application to the entire Mekong delta was, in turn, validated against two flood events in 2000 and 2011 in Triet et al. (2017), also with good correspondence to the observations, achieving NSE of observed water levels of between 0.72 and 0.97 across 19 different gauging stations.

Here we validated the IWRM-Sub model for the Cambodian Mekong floodplain against water levels and discharge in four stations and flood extent based on Landsat imagery (see Methods). Based on the validation measures (Table 3), a good model performance is obtained at all stations (both water discharge and water level) with the values of NSE between 0.69 and 0.87, PBIAS between -14.4 % and +9.8 %, RSR between 0.37 and 0.55, and R2 between 0.89 and 0.93. It should be noted that the statistical model performance with NSE and R2 greater than 0.5, PBIAS between ±25 %, and RSR less than 0.7 is indicated as the decision guideline for hydrologic model studies (Benaman et al., 2005; Setegn et al., 2010). A time series comparison between the simulated and observed water discharge and water level (1985–2000) at four hydrological stations can be found in Figs. S2 and S3. It is apparent that the simulated water discharge among these stations is in line with the observed data throughout the 15-year hydrological record available for comparison.

Results of the flood extent comparison between the IWRM-Sub model and SWMT observations over the time horizon 1985–2000 show an equally good agreement. The model underestimates the total flooded area by just 0.1 % as the ratio of simulated to observed flooded extent areas is 0.99. However, the overlapping flooded area only constituted 71 % of the observed (SWMT) extent (which constitutes the recall) and 72 % of the simulated (IWRM-sub) extent (which is the precision) (Fig. 3). Part of this discrepancy may be accounted for by the inclusion of rivers and lakes in the extent of the simulation, yet not in the SWMT-derived extents. Using multiple models in succession can have the negative effect of compounding errors; however these results demonstrate that this has not unduly impacted our methodology as our estimations closely match the observations of flood extent.

Having run the model for each of the development scenarios (S1–S12; see Table 2), we obtained the corresponding daily time series of water discharge and water level at each station and compared them with the baseline scenario. We then calculated the mean monthly water discharge and water level across the study period. Finally, we computed the percentage change in mean monthly water discharge and water level for each scenario at each station. The results at Kratie, Kampong Cham, and Chroy Changvar were virtually indistinguishable from one another, so to avoid unnecessary repetition, we have presented results from only Kampong Cham (as the midway station) and Neak Loeung, which differs significantly from the other stations for being downstream of the Tonle Sap River confluence (Fig. 1) and the Bassac River distributary (Fig. 4).

All scenarios that contain an element of hydropower development follow the same pattern of increasing both water discharge and water level during the dry season (November–May), whilst reducing water discharge and water level during the early and middle wet season (June–September) (Fig. 4). The impact of climate change appears to fluctuate during the months of January to June between Kampong Cham (and Kratie and Chroy Changvar) and Neak Loeung, as there is a slight increase in discharge and water levels at the upstream stations yet a slight decrease at the downstream station, though the magnitude of any alteration is only small. From July to December, however, the climate change impact is much stronger and increases discharge and water levels at all stations. The larger magnitude of the climate change impacts during the wetter months counteracts the impact of hydropower and irrigation (which slightly reduces flows and water levels in all months), which can be seen in the difference between scenario S2 (hydropower solo) and scenarios S7–S12 that incorporate multiple drivers (Fig. 4; scenario description in Table 2). This is most evident at Kampong Cham station in October, where climate change impacts are large enough to offset hydropower impacts, so that only those scenarios that incorporate the additional impact of irrigation are strong enough to reduce flows and water levels. Whilst the largest magnitude impacts are in the wetter months of July to September, the proportional impacts are far larger in the dry season, where the impact of hydropower development dominates the flow regime and increases water levels up to 150 % in April at Kampong Cham, compared to a maximum decrease of < 25 % in July.

Comparing results from upstream stations with those at Neak Loeung, we see that the magnitude of climate change impacts is larger downstream both absolutely and proportionally. This is evident in the greater differences between the solo hydropower scenario (S2) and the combined hydropower and climate change scenarios (S7–S12) here than observed at the upstream stations. Nevertheless, hydropower impacts still dominate the flow regime, especially during the drier months where discharges increase > 100 % in April.

Our results suggest that planned hydropower developments will drastically alter the hydrology of the Mekong main channel and far outweigh the effects of irrigation or climate change impacts in either counteracting or enhancing these alterations.

Here we present the quantitative results together with the spatial analysis of flood conditions throughout the entire study area. The comparisons between each scenario and their justifications are described in the analysis at the provincial level because of the similarity in patterns. Under the baseline scenario (S1), the modelling results between 1971 and 2000 show that the yearly flooded area ranges from 7785 to 11 525 km2. Its mean annual value is estimated at 9370 km2, about 34 % of the whole study area.

We compared year to year the impact of each development scenario against the S1_baseline (1971–2000) on the total flooded area across the study area (Fig. 5). Scenarios S2–S4 use the same driving climate data as the baseline scenario (S1), and so the variability in the impact shown is significantly reduced to produce consistent impacts for all years. Conversely scenarios S5–S12 are driven by future climate data projections, so that the variability in comparing year to year is significant. Nevertheless, there is a clear pattern that emerges, once again showing the dominance of hydropower development in significantly reducing the yearly flooded area. The impacts of both irrigation development scenarios (S3 and S4) also reduce the yearly flooded area, though to a lesser extent. Climate change impacts in isolation (S5 and S6) increase the flooded area overall, though there are some years in which the area is reduced compared to the baseline. The proportional magnitude of these effects is most evident in the solo hydropower development, with a median reduction of > 20 % year on year, yet the combined impact of irrigation, hydropower, and climate change did reduce flooded areas by up to 40 % in some years (Fig. 5).

The spatial distribution of flood inundation and depth across the Cambodian Mekong floodplain varies greatly between scenarios of planned developments and climate change (Fig. 6). The floodplain is characterized spatially by a high fluctuation of flood depth and flood duration alteration of over ±100 % in almost all scenarios, especially in the southeast and the southwest parts of the study area. Whilst the magnitude of these fluctuations is large across all scenarios, it is most evident in hydropower (S2) (reductions of depth and duration) and climate change RCP8.5 (S6) scenarios (increase in depth and duration). Though even in these most extreme cases, there are areas that run contrary to the general pattern of change, highlighting the hydrological complexity of the region. The low-irrigation scenario (S4) has the least impact (Fig. 6), though even this level of development may significantly impact the lower-lying regions in the southwest and southeast where much of the rice cultivation is concentrated. Our results suggest that all scenarios will cause heterogeneous impacts across the region that may effectively shift flood impacts from one area to another rather than completely dispel the associated risks.

We examined the change in flooded area, flood depth, and flood duration for 10 provinces that have a considerable part of their area within the study area (Kampong Speu and Kampot provinces and Tay Ninh Province in Vietnam were not included; see Fig. 1). Each scenario was compared to the baseline period at the provincial level (Fig. 7). Under the baseline scenario (S1), the modelling results show that the average flooded area ranges from a minimum of 188 km2 in Phnom Penh Province to a maximum of 2308 km2 in Prey Veng Province, which represents 43 % of the provincial territory. Whilst the average flood depth ranges from 0.54 m in Svay Rieng Province to 2.4 m in Kratie Province, the average flood duration ranges from 10 d in Svay Rieng Province to 79 d in Kampong Chhnang Province.

Except for the Svay Rieng region, which appears anomalous, Kampong Chhnang and Kratie are least affected by the impacts of climate change, whilst Prey Veng and Takéo are most affected (Fig. 7). The development scenarios have the least effect in Prey Veng, where flood area and depths are almost unaffected in comparison to the other provinces.

Svay Rieng displays an extreme reduction in flood duration for all scenarios, including climate change scenarios, which is also true of the flooded area except for the RCP4.5 climate impact scenario (S5). Depths, however, increase in all scenarios, suggesting that flooding in this province is reduced in extent and duration to a shorter, more intense (and deep) flood event.

The model performance metrics achieved by our hydrological simulation of water discharge and water level for the baseline period of 1971–2000 at all four monitoring stations (Kratie, Kampong Cham, Chroy Changvar, and Neak Loeung) exceed existing studies within the same region (Västilä et al., 2010), with the exception of Dang et al. (2018), who recorded an NSE value of 0.98 compared to our value of 0.80 at Kampong Cham station. Whilst there are studies of flood extent within our study area that only focus on a single event rather than a multi-year analysis that slightly surpass our own in terms of performance metrics (Fujii et al., 2003), our continual analysis of annual flood patterns comprising a 30-year time horizon is comparable to, and often exceeds, other such multi-year analyses done in the region (Try et al., 2020a, b). The relative success of our baseline simulations allows us to have a high degree of confidence in our future projections of the Cambodian Mekong floodplain's hydrological response to planned infrastructural development and future climate changes. All future projections of scenarios containing multiple drivers that we considered within our analysis followed the same generic pattern of alterations to both the expected discharge and river water level, increasing during the dry season (November–May) and decreasing during the early and middle wet season (June–September). Such a general pattern of alteration is due to the overwhelming dominance of the hydropower development impacts, which overcome any counteraction that might be applied by either irrigation development schemes (counteracts in dry season) or climate change impacts.

These general trends are in line with the majority of previous research in the region (Dang et al., 2018; Kallio and Kummu, 2021; Lauri et al., 2012; Piman et al., 2013; Räsänen et al., 2012; Västilä et al., 2010). The degree of alteration to these hydrological indicators is most pronounced in the upstream areas of Kratie, Kampong Cham, and Chroy Changvar stations and diminishes downstream of the confluence with the Tonle Sap River towards Neak Loeung station, which is also consistent with earlier findings (Dang et al., 2018).

Our findings clearly demonstrate the homogenizing effect that the planned hydropower developments would have on the Mekong River's hydrograph, which would go far beyond simply contracting the impacts of other drivers and would reshape the expected flow regime, massively increasing dry season low flows and significantly reducing wet season high flows.

The future projections of flood conditions suggest that most provinces will see an increase in depth, duration, and area under climate change scenarios but that these alterations are counteracted by the combined development scenarios reflecting the flood prevention benefit afforded by irrigation and hydropower scenarios. These findings are supported by other studies that look at the impact of isolated drivers of hydrological change in the region (Fujii et al., 2003; Try et al., 2020a) and studies that look at multiple drivers in nearby regions (Hoanh et al., 2010; Pokhrel et al., 2018).

Our provincial level assessment shows that Prey Veng Province is most vulnerable to the largest flooded area (Fig. 7), as its large territory is entirely located in the low-lying area adjacent to the Mekong River. Kampong Thom Province receives the largest flood prevention benefit provided by the planned hydropower developments, whilst Kampong Chhnang receives the least in terms of flooded area and flood duration, most likely because the flood regime is strongly controlled by the Tonle Sap Lake system and receives less influence from the upstream flow alterations. Svay Rieng Province is drastically impacted by all the scenarios. This is most likely due to the extremely low ground surface elevation (majority less than 8 m), meaning that slight alterations have proportionally large impacts. The region may also be affected by changes to the hydrological conditions on the Vietnamese Mekong Delta, some of which were represented in this study by means of the boundary conditions supplied by Triet et al. (2020) that considered the whole delta region.

Changes in hydrological and flood conditions in the Cambodian Mekong floodplain could imply both positive and negative consequences to various sectors such as water resource management, agricultural productions, and ecosystem services (Arias et al., 2012; Kummu and Sarkkula, 2008). In addition, the direction, magnitude, and frequency of impacts will be varied from one location to another.

The beneficial consequences associated with the impact of planned developments are derived from increased water availability in the dry season and reduced flood prevalence in the wet season. The reduction in flood risk due to the decline in the wet season flows and water levels would be a large socio-economic benefit of these development plans, potentially reducing the duration and extent of affected regions by more than 20 % (Fig. 5). In addition, increased dry season flow would greatly enhance agricultural productivity, enhance water security, and minimize conflicts between consumers. Environmental flow could also be secured, which may help some aspects of ecosystem productivity. Increases in water levels might also reduce energy costs associated with water pumping and better facilitate dry season navigation.

However, there are many negative consequences to the reduction in flood extent and duration associated with the planned development scenarios. Hydropower projects in the Mekong are projected to trap considerable parts of the sediments and the nutrients it contains in the reservoir behind the dam wall, reducing their transportation downstream and subsequent distribution across the floodplain (Kondolf et al., 2018; Kummu et al., 2010; Schmitt et al., 2017, 2018). The reduction in sediment transport rates associated with reduced wet season flows and sediment trapping upstream inevitably leads to sediment-starved water flow downstream. This in turn leads to increased rates of channel incision and accelerating riverbank erosion as river waters gain in situ material for transportation up to carrying capacity (Darby et al., 2013; Morris, 2014). The drop in soil fertility (nutrient bound to sediment) throughout the downstream floodplains would result in a great challenge for ecosystem productivity (Arias et al., 2014), rice production (Boretti, 2020), and the sustainability of flooded forests (rich habitats for fish and other species) (Arias et al., 2014). Dams also act as barriers disturbing fish migration between upstream and downstream sections essential for feeding and breeding, resulting in fishery losses (Ziv et al., 2012). In addition, the increasing dry season water levels will disturb various river works – for instance, the low-water-level condition is favourable to river channel maintenance (dredging) and constructions of water infrastructure, usually starting and very active during the dry season months.

Whilst higher economic damages from flood disasters are proportional to extended flooded areas, intensifying flood depths, and prolonging flood durations, there are counteracting positive impacts associated with floods, including the transport of nutrients and increased fishery productivity. Increasing flood extents widen the coverage of fertile agricultural land (Lamberts, 2008), which implies a more extensive production of rice – the most important agricultural activity in the Cambodian Mekong floodplain. In contrast, a substantial reduction in flooded area would lead to a fall in flooded forest, a rich habitat for fish and other species (Arias et al., 2014; Kummu and Sarkkula, 2008), leading to a decline in fishery and ecosystem productivity in general. These benefits from an extended flood extent need to be balanced against the detrimental impacts of deep flood depths and long flood durations, which can be catastrophic to crop yields across the floodplains. Therefore, suitable flood conditions should be well determined for a better trade-off with the developmental impacts.

Several studies have been conducted to understand hydrologic processes within the Cambodian Mekong floodplain, Tonle Sap Lake basin, and Vietnamese Mekong Delta. Different considerations have been taken into account for the analysis in previous research; they include but are not limited to (1) water infrastructure development, (2) climate change, (3) sea level rise, (4) land use and land cover change, (5) population growth, and (6) climate-related phenomena. However, the present study is targeted to gain insight into how the combination of upstream hydropower development, irrigation expansion, and climate change will affect the Cambodian Mekong floodplain in terms of hydrological and flood patterns. Under climate change scenarios, the future rainfall and temperature were assumed to respectively be wetter and warmer.

Future research should employ finer-resolution climate models and newer CMIP6 scenarios, although according to our analysis, basin-wide mean precipitation and temperature do not differ greatly between these two climate change modelling phases (Table S1). In addition, a small-scale decision support tool, as well as satellite-based image analysis, was set up to assist in evaluating a comprehensive study of the flood vulnerability in the Cambodian Mekong floodplain or the wider implications for the water–energy–food nexus for present and future conditions.

Another relevant research direction is the prediction of future land use and river morphological changes. This could generate a key input for a more realistic assessment of hydrological and flood alterations. River sand mining has been very active in the Cambodian Mekong River and its main tributaries as rapid and on-going urbanization requires a massive amount of sand, which is an important material not only for construction but also for backfill (Boretti, 2020; Hackney et al., 2020). Riverbank collapses, directly or indirectly associated with excessive sand extraction, have been very severe. Moreover, many floodplains and wetlands have been filled by sand and transformed into urban areas, resulting in a critical change in river morphology and landscape along the river channels and throughout the floodplains. More importantly, these alterations are still being perpetuated without the full impact of their occurrence being understood or accounted for.

Floods are an essential component of the landscape for both the people and the ecosystem of the Mekong Basin, but they also pose significant hazards and losses when the magnitude is too great to handle effectively. As the development of water infrastructure could cause a decrease in flood conditions and climate change may reverse such impacts, it is still unknown what the desired flood water level and flood duration should be. This has led to a great difficulty in proposing optimum flood protection measures while maximizing dam benefits. Therefore, another potential research topic is the determination of the ideal flood conditions for maximum productivity from both the agricultural and ecosystem perspectives.

The intended purpose of this future research is to provide valuable information and assist governments, policymakers, and water resource engineers in foreseeing future threats of different intensities. Moreover, their results would be helpful in formulating better water resource management strategies and in elevating all living things' resilience to the future challenges for the sustainability of resources within the floodplain.