https://orcid.org/0009-0001-0047-1967
Abstract
This study investigates the effectiveness of ponds as a nature-based solution (NBS) to concurrently ameliorate flood and drought impacts, emphasizing the need for an integrated response to multi-extreme hydrological events. We incorporate ponds into agricultural landscapes in the Bagmati River Basin of Nepal and assess their performance using the Soil and Water Assessment Tool (SWAT+). Six different scenarios are thoroughly explored to see how these interventions affect the main components of the water balance, such as surface run-off, lateral flow, percolation, and evapotranspiration. The spatial efficiency of the ponds, particularly in their immediate surroundings and downstream areas, has been proven to be a crucial factor in their overall efficacy in attenuating extremes, which increases with the size of the intervention area. Although the effects of ponds on floods and droughts are minor, they could be significantly magnified by a synergistic use of other NBS tactics, such as conservation tillage or soil conservation techniques. Future studies should establish the most appropriate sites and volumes for these interventions, as well as further investigate the possible advantages of several NBS, to optimize flood and drought management in the Bagmati River Basin and other similarly susceptible places.
HIGHLIGHTS
This research contributes to the studies that aim at understanding changes in river basins due to the implementation of ponds as an NBS.
Ponds have the potential to attenuate both floods and droughts.
Modeling of NBS is useful for assessing its effectiveness for river basin water management.
INTRODUCTION
Reducing the threat of severe events like floods and droughts is a growing priority for water resource development and management. These two catastrophic phenomena have caused unimaginable misery and economic loss for 3 billion people over the past two decades (The World Bank 2021). A flood event is an abnormally high river flow that exceeds a predefined threshold and is likely associated with damage (Prudhomme et al. 2023). Two distinct approaches can be used to classify flood events: annual maximum (AM) series and peak over threshold (POT), to study their occurrence frequency. The AM method allows the selection of AM discharge values, i.e., one event per year for the frequency analysis. There is a possibility that this method may exclude some multiple flood peaks that occur in the same year but may include some lesser peaks occurring in some other years (Bezak et al. 2014). On the other hand, the POT method selects all peak values above a certain threshold level. However, with the POT method, several complexities are associated with its application. According to Lang et al. (1999), the main complexity is the selection of threshold level as there is no universal value that has to be selected. Similarly, Bezak et al. (2014) mentioned that the independence criteria of the events are to be taken care of during the selection of peak values such that no two selected flood peaks relate to the same flood mechanism. Because the AM method is simple and well-established (Prudhomme et al. 2023), it is still the most widely used approach in the analysis of flood frequency (Lang et al. 1999). Its greatest advantage is that it ensures the selected peaks are independent of one another (Nagy et al. 2017).
A general definition of drought is provided by Tallaksen et al. (1997) as the drought being regional in nature and that leads to critical conditions in the event of prolonged water scarcity over huge areas. Different types of droughts are discussed in the literature, including meteorological droughts, soil moisture droughts, and hydrological droughts. These are classified according to different stages of hydrological cycles. Meteorological droughts are mostly the result of a deficiency in precipitation. Soil moisture drought occurs when there is a deficiency of soil moisture (primarily in the root zone), hence limiting the availability of moisture to vegetation (Van Loon 2013). Soil moisture droughts are sometimes referred to as agricultural droughts as they are closely related to crop failure. Hydrological drought, on the other hand, is related to a deficit in surface and sub-surface water (e.g., river discharge, groundwater levels, or water level in lakes) (Van Loon 2013). Droughts begin as meteorological droughts, progress to soil moisture droughts, and eventually become hydrological droughts (Fleig et al. 2006). The primary focus of this research is the streamflow characteristics based on the discharge time series. For characterizing the streamflow droughts, many authors have used the threshold level method (e.g., Zelenhasic & Salvai 1987; Kjeldsen et al. 2000; Fleig et al. 2006), where they used streamflow as the variable to identify the presence of a drought event. Here, drought occurs when the streamflow value falls below a predefined threshold. The primary advantage of this method is that it can characterize the events on a daily basis and also allows for detailed monitoring of the start and end of the event.
The Bagmati River Basin of Nepal is one of the major river basins that support much of the country's socioeconomic activity (Babel et al. 2012). This basin often faces the problem of flooding, resulting in enormous destruction every year. Although numerous structural solutions have been implemented to mitigate floods, flood-related losses are not diminishing. Dixit et al. (2008) concluded that when structural barriers keep floodwater from draining rapidly, they benefit a few settlements but harm downstream and vulnerable places. Although droughts are not widely studied in Nepal as much as flooding, few studies indicate that they cannot be excluded. Bagale et al. (2021) analyzed the drought over Nepal using standard precipitation index (SPI) for a period between 1977 and 2018 for 107 rainfall stations. According to the occurrence of droughts, they classified the years as summer drought years (8 years) and annual drought years (10 years). Shrestha et al. (2018) conducted a future drought assessment for the Bagmati River Basin and found that drought events with different severity and extent are likely to occur in 10 of the 24 years between 2030 and 2053.
The increasing frequency of floods and droughts necessitates the development of novel solutions for the sustainable management of water resources. Recent focus has shifted towards some natural solutions that mimic the natural process, are sustainable, and can mitigate the risk without producing any environmental risk. This solution is usually termed as nature-based solution (NBS). NBSs harness the power of natural processes to address issues including climate change, water resources, food security, or disaster risk management (Pauleit et al. 2017). NBS slows down the rate of run-off in a catchment by boosting interception, infiltration, or storage for flood water, hence mitigating the risk. Modeling of NBS can be very useful to understand the changes they bring in various water balance components. Several studies have analyzed the impact of NBS on water resources. Mwangi et al. (2016) assessed the influence of agroforestry on water balance using the soil and water assessment tool (SWAT) where they observed an increase in evapotranspiration, a decrease in baseflow, surface run-off, and overall water yield. Similarly, Spyrou et al. (2021) implemented a flood storage reservoir using the TUFLOW hydraulic model and the MIKE- Système Hydrologique Européen (MIKE-SHE) hydrological model. They concluded that the maximum flooding depth and velocity were reduced, particularly around and downstream of the NBS implementation area. The flooded area was also decreased, especially for more regular occurrences. Ruangpan et al. (2020) provided a review of the literature where NBSs were used for hydro-meteorological risk reduction. The study identified numerous papers on using NBS to reduce flood peaks (Liao et al. 2015; Ercolani et al. 2018; Mei et al. 2018; Yang & Chui 2018) but only three articles on the drought reduction (Radonic 2019; Wang et al. 2019; Lottering et al. 2020). Also, the implementation of NBS in large-scale catchments (river basin, rural, and regional levels) are underexplored compared to that in small-scale urban settings due to their complexity in representation (Ruangpan et al. 2020).
Regardless of the numerous applications of NBS in hydrological models in the world, very little research on the usage of hydrological models has been conducted in Nepal (Manjan & Aggarwal 2014; Dahal et al. 2016), and the impacts of NBS are still not examined. The primary objective of this study was to assess the impact of NBS on floods and droughts in the Bagmati River Basin of Nepal. Rainwater harvesting (referred to as ponds in this research) were used as NBS in order to assess their impacts on floods and droughts in this basin. Globally, in situ rainwater harvesting (IWRH) is commonly used as a strategy for dealing with both excessive rains and dry spells (Hofman & Paalman 2014). Some examples of rainwater harvesting reported in the literature include the evaluation of the impact of rainwater harvesting on streamflow values by Masih et al. (2011), assessing the influence on crop yield, evaporation, and river flow by Andersson et al. (2011), and analyzing the impact on different water balance components by Wambura et al. (2018) and Welderufael et al. (2013). However, their study did not analyze the impacts on floods and droughts. Thus, in this research, variations produced by NBS were analyzed in terms of change in the number of flood and drought occurrences, contrary to the analysis of just one of these events commonly performed in other studies. SWAT+ was used as a modeling tool for representing the current hydrological characteristics of the basin and to model the ponds. SWAT+ is widely used to evaluate the impact of different land use change scenarios on water resources (Chanasyk et al. 2003; Mapfumo et al. 2004; Lin et al. 2007; Ouyang et al. 2008). Additionally, this tool permits intervention in the hydrological response units (HRUs), which enables the analysis of effects throughout the basin because extreme events and the consequences of interventions are not limited to the river channel.
CASE STUDY
The Bagmati River Basin extends between 27°10′ and 27°50′ N latitude and between 85°02′ and 85°58′ E longitude in central Nepal. The Bagmati River begins from mountain springs north of Kathmandu and flows south through the Kathmandu valley and Terai plain, eventually connecting with the Ganges River system in India. River discharge increases throughout the wet season, reaching a peak in July–August and a minimum in January–April. On average, this basin receives 1,800 mm of annual rain, with the monsoon season accounting for 80% of the total.
Location map of Bagmati River Basin.
METHODOLOGY
General methodology.
Model input data
Precipitation, temperature, relative humidity, wind speed, and solar radiation were the climatic variables used in this study. Daily precipitation data were gathered from four separate stations (Dhap, Kathmandu Airport, Hariharpurgadhi, and Ramolibariya) for the period between 1994 and 2013. For the observed data, the daily discharge measured at the Pandheradobhan station was taken and utilized to calibrate and validate the hydrological model. Daily data for temperature, relative humidity, wind speed, and solar radiation (1994–2013) were downloaded from climate forecast system reanalysis (CFSR) (https://swat.tamu.edu/data/cfsr) and used for this study.
DEM with a spatial resolution of 30 m from shuttle radar topography mission (SRTM) was used in this study. The majority of the catchment's area is occupied by hills and mountains, and the elevation ranges from as low as 121 m to as high as 2,787 m (Figure 1). Compared to the upper and middle parts of the catchment, the lower part is relatively flat.
Land use land cover.
Soil types.
Model setup
The model setup started with the delineation of the catchment using the raster DEM of the catchment. The thresholds for the area of the channel and the streams were set to 7 and 70 km2, respectively. The outlet point was drawn at Pandheradobhan station, and the delineation of the catchment was done, which resulted in a number of sub-basins. Landscape units (a division of sub-basins into floodplain and upslope) were created using the DEM inversion method with a ridge threshold of 70 km2 and slope position threshold of 0.10 (default value).
Having calculated the landscape units, the subdivision of these into a series of HRUs was done. Raster maps of land use land cover and the soil are the inputs needed for this stage. A total of 163 channels and 23 sub-basins were created after delineating the catchment, which resulted in an area of 2,822 km2. These sub-basins were further discretized into 2,269 Hydrological Response Units (HRUs).
For the run-off estimation, the Soil Conservation Service (SCS) curve number method was used, the Muskingum routing method for routing of flow, and the Penman–Monteith method for the calculation of evapotranspiration. Meteorological inputs (precipitation, temperature, relative humidity, solar radiation, and wind speed) from four stations (Figure 1) were imported and the model was run for a period of 20 years (1994–2013), taking the initial 2 years as warming-up period.
Model calibration and validation
An initial model was developed that can simulate the rainfall-run-off process in the catchment to a certain extent. This model was considered as the baseline scenario based on which analysis of NBS was done. In order to develop the model, SWAT + , a restructured version of SWAT, was used. A total of 5,842 daily records were taken for calibration and validation of the model. The data were split into calibration (1994–2005) and validation (2006–2013) sets. The first 2 years of both calibration and validation sets were considered as a warming-up period. For the calibration process and the sensitivity analysis for streamflow simulation, the automatic calibration method with SWAT+ toolbox was applied.
The objective function used for calibration was to maximize the Nash–Sutcliffe efficiency (NSE). The calibration and validation were done at a daily time step using the observed discharge data at Pandheradobhan. Nash–Sutcliffe efficiency (NSE) and percent bias (PBIAS) were used to measure the statistical model performance for both calibration and validation of the streamflow.
NSE and PBIAS are widely used for the model performance analysis of SWAT (e.g., Leta et al. 2021; Tayebzadeh Moghadam et al. 2021). Similarly, the rating of performance measures was based on the evaluation of NSE and PBIAS given by Leta et al. (2021).
Scenarios assessed
Division of catchment for the implementation of ponds.
Example of hydrographs and different thresholds for flood peak identification.
The intervention with ponds was carried out in the baseline model to assess the catchment's response to this intervention. In this research, the ponds were represented by increasing the available water capacity (awc) of soil by 20%, adopting the methodology of Masih et al. (2011). They mentioned a similar study done by Faramarzi et al. (2010), who examined the effect of a 20% increase in awc on irrigation requirement and discovered a considerable improvement in irrigation water use. The awc parameter is related to the water retention capacity of soil and it is assumed that the ponds increase the retention capacity of soil by 20%.
Different scenarios were implemented based on different locations in the catchment as listed below – each scenario was named after the initials of where the ponds were implemented:
Scenario U: Ponds in Upper catchment
Scenario M: Ponds in Middle catchment
Scenario L: Ponds in Lower catchment
Scenario UM: Ponds in Upper + Middle catchment
Scenario ML: Ponds in Middle + Lower catchment
Scenario UML: Ponds in Upper + Middle + Lower catchment
As the stored water is mostly used for crop production, these ponds were only implemented in the agricultural HRUs in the catchment.
Based on the scenarios mentioned above, the following analysis is done at points P1, P2, and P3:
- (a)
Change in flood peaks
- (b)
Change in number and duration of drought events
- (c)
Mean monthly variation of the streamflow
Flood frequency analysis
Flood frequency analysis (FFA) is the estimation of how frequently a specific event occurs. It is vital to understand the FFA for determining the magnitude of flood at various return periods and thereby minimizing damage. FFA in this research was done at the basin's outlet using the AM series method.
Discharge values were computed corresponding to return periods 2, 5, and 10 years for the baseline scenario. Three different ranges were set for the discharges: (a) Q2–-Q5, (b) Q5–Q10, and (c) greater than Q10. The peaks within these ranges were used to compare the baseline scenario to the changes caused by the interventions. Figure 6 provides an illustration of hydrographs and different thresholds for flood peak identification.
Drought analysis
The value of tc was chosen to be 5 days. This was determined to be suitable for perennial streams following Fleig et al. (2006)'s sensitivity study with various tc values. Also, it is necessary to exclude minor drought events, for which several methods that are being used are specified by Fleig et al. (2006). Among those, this study employed excluding minor drought events less than 9 days (30% of a month).
RESULTS
Model calibration and validation
Table 1 shows the results of the calibration process with the optimum values of the calibrated parameters. The negative signs in the optimum values are the percentage change or the relative change in the default parameter values. These optimum values were fed to the initial model in order to calibrate the model.
Ranges and the optimum value for the calibrated parameters
| Name | Definition | Group | Change type | Range | Optimum |
|---|---|---|---|---|---|
| cn2 | SCS curve number for moisture condition II | HRU | Percent | 35–95 | −2.72 |
| cn3_swf | Pothole evaporation coefficient | HRU | Replace | 0–1 | 0.28 |
| perco | Percolation coefficient | HRU | Replace | 0–1 | 0.72 |
| epco | Plant uptake compensation factor | HRU | Replace | 0–1 | 0.22 |
| ovn | Overland Manning's roughness | aqu | Percent | 0.01–30 | 6.48 |
| alpha | Baseflow alpha factor (1/days) | aqu | Replace | 0–1 | 0.01 |
| flo_min | Minimum aquifer storage to allow return flow (mm) | aqu | Replace | 0–5,000 | 1,478.54 |
| k | Saturated hydraulic conductivity (mm/h) | sol | Replace | 0.0001–2,000 | 1,992.04 |
| bf_max | Maximum baseflow (mm) | aqu | Replace | 0.1–2 | 0.82 |
| chk | Effective channel hydraulic conductivity (mm/h) | rte | Replace | −0.01 to 500 | 268.99 |
| chn | Manning coefficient for main channel | rte | Replace | −0.01 to 0.3 | 0.21 |
| awc | Available water capacity of the soil layer (mm H2O/mm soil) | sol | Relative | 0.01–1 | −0.12 |
| Name | Definition | Group | Change type | Range | Optimum |
|---|---|---|---|---|---|
| cn2 | SCS curve number for moisture condition II | HRU | Percent | 35–95 | −2.72 |
| cn3_swf | Pothole evaporation coefficient | HRU | Replace | 0–1 | 0.28 |
| perco | Percolation coefficient | HRU | Replace | 0–1 | 0.72 |
| epco | Plant uptake compensation factor | HRU | Replace | 0–1 | 0.22 |
| ovn | Overland Manning's roughness | aqu | Percent | 0.01–30 | 6.48 |
| alpha | Baseflow alpha factor (1/days) | aqu | Replace | 0–1 | 0.01 |
| flo_min | Minimum aquifer storage to allow return flow (mm) | aqu | Replace | 0–5,000 | 1,478.54 |
| k | Saturated hydraulic conductivity (mm/h) | sol | Replace | 0.0001–2,000 | 1,992.04 |
| bf_max | Maximum baseflow (mm) | aqu | Replace | 0.1–2 | 0.82 |
| chk | Effective channel hydraulic conductivity (mm/h) | rte | Replace | −0.01 to 500 | 268.99 |
| chn | Manning coefficient for main channel | rte | Replace | −0.01 to 0.3 | 0.21 |
| awc | Available water capacity of the soil layer (mm H2O/mm soil) | sol | Relative | 0.01–1 | −0.12 |
Results of the performance measures
| Performance criteria | Calibration | Validation |
|---|---|---|
| NSE | 0.53 (Satisfactory) | 0.66 (Good) |
| PBIAS | 4.6 (Very Good) | −6.95 (Very Good) |
| Performance criteria | Calibration | Validation |
|---|---|---|
| NSE | 0.53 (Satisfactory) | 0.66 (Good) |
| PBIAS | 4.6 (Very Good) | −6.95 (Very Good) |
Observed and simulated discharges in calibration (left figure) and validation (right figure).
Observed and simulated discharges in calibration (left figure) and validation (right figure).
The NSE value of 0.53 was obtained for calibration, which can be considered a satisfactory result; however, for the validation, performance slightly improved with an NSE of 0.66. Similarly, the result showed that the model slightly underestimated flow during calibration as indicated by PBIAS of +4.6 and slightly overestimated in validation with PBIAS of −6.95. The mean observed and simulated flows during the calibration period were 129.96 and 123.98 m3/s, respectively, whereas, during validation, they were 109.59 and 117.11 m3/s.
When comparing the simulated and observed hydrographs for both calibration and validation, the overall hydrology of the catchment was well represented; however, some of the peaks were not well represented. Although it is not a perfect representation of reality, it is more or less a good representation of the processes taking place in the catchment. As such, this model was used as a baseline for this research in order to compare the changes that the NBS intervention would produce.
FLOOD FREQUENCY ANALYSIS
Hydrographs at the main outlet before (baseline) and after the implementation of ponds (NBS).
Hydrographs at the main outlet before (baseline) and after the implementation of ponds (NBS).
Table 3 summarizes the number of flood events in the baseline scenario over these various return periods. It can be observed that more than 60% of the total number of peaks lie in the range Q2–Q5.
Number of peaks in different return periods in baseline scenario
| Return period (years) | Number of peaks (baseline) | ||
|---|---|---|---|
| P1 | P2 | P3 | |
| Q2–Q5 | 8 | 6 | 6 |
| Q5–Q10 | 1 | 2 | 2 |
| >Q10 | 1 | 1 | 2 |
| Total | 10 | 9 | 10 |
| Return period (years) | Number of peaks (baseline) | ||
|---|---|---|---|
| P1 | P2 | P3 | |
| Q2–Q5 | 8 | 6 | 6 |
| Q5–Q10 | 1 | 2 | 2 |
| >Q10 | 1 | 1 | 2 |
| Total | 10 | 9 | 10 |
Number of peaks reduced (bars) and maximum flow reduced (markers) under different scenarios at (a) P1, (b) P2 and (c) P3.
Number of peaks reduced (bars) and maximum flow reduced (markers) under different scenarios at (a) P1, (b) P2 and (c) P3.
CHANGE IN NUMBER AND DURATION OF DROUGHT EVENTS
Table 4 shows the number of drought events under the baseline and other scenarios as described in the methodology (scenarios assessed). For each scenario, the impact of the intervention on all three points, P1, P2, and P3, was investigated. At P1, the number of drought events in all scenarios remains the same as in the baseline scenario. At P2, the scenarios involving the middle catchment reduce the number of events by 1 while the other cases remain unchanged. At P3, scenario U could not reduce any drought event, while others could reduce it by 1 or 2. In total, out of 245 drought events, all the scenarios could reduce them to 243, except for scenario U.
No. of drought events at P1, P2, P3 in the baseline and under different scenarios
| Location | Baseline | Scenarios | |||||
|---|---|---|---|---|---|---|---|
| U | M | L | UM | ML | UML | ||
| P1 | 79 | 79 | 79 | 79 | 79 | 79 | 79 |
| P2 | 78 | 78 | 77 | 78 | 77 | 77 | 77 |
| P3 | 88 | 88 | 87 | 86 | 87 | 87 | 87 |
| Total | 245 | 245 | 243 | 243 | 243 | 243 | 243 |
| Location | Baseline | Scenarios | |||||
|---|---|---|---|---|---|---|---|
| U | M | L | UM | ML | UML | ||
| P1 | 79 | 79 | 79 | 79 | 79 | 79 | 79 |
| P2 | 78 | 78 | 77 | 78 | 77 | 77 | 77 |
| P3 | 88 | 88 | 87 | 86 | 87 | 87 | 87 |
| Total | 245 | 245 | 243 | 243 | 243 | 243 | 243 |
Number of drought events at different durations and scenarios at (a) P1, (b) P2, (c) P3.
Number of drought events at different durations and scenarios at (a) P1, (b) P2, (c) P3.
MEAN MONTHLY VARIATION OF THE STREAMFLOW AT POINTS P1, P2, P3
Percentage change in mean monthly discharge at points P1, P2, P3: (a) implementation of ponds in upper catchment-scenario U, (b) implementation of ponds in middle catchment-Scenario M, (c) implementation of ponds in lower catchment-Scenario L, (d) implementation of ponds in upper + middle catchment-Scenario UM, (e) implementation of ponds in middle + lower catchment-Scenario ML, (f) implementation of ponds in upper + middle + lower catchment-Scenario UML, the 0 value is for the baseline, positive values are the percentage increase, and the negatives are the decrease).
Percentage change in mean monthly discharge at points P1, P2, P3: (a) implementation of ponds in upper catchment-scenario U, (b) implementation of ponds in middle catchment-Scenario M, (c) implementation of ponds in lower catchment-Scenario L, (d) implementation of ponds in upper + middle catchment-Scenario UM, (e) implementation of ponds in middle + lower catchment-Scenario ML, (f) implementation of ponds in upper + middle + lower catchment-Scenario UML, the 0 value is for the baseline, positive values are the percentage increase, and the negatives are the decrease).
When the monsoon season is considered, the discharge decreases in all the scenarios. However, for the dry season, in the majority of the cases, the discharge increases from October through December and drops from January. This means that, with this intervention, the flood events are reduced, which mostly occur in the rainy season, and drought events are reduced from October to December but increase from January to May.
Additional details about each scenario are presented below.
Implementation of ponds in upper catchment (Scenario U)
The results of this scenario are illustrated in Figure 11(a). Due to the fact that this intervention is limited to the upper catchment, the closest downstream point, P1 exhibits a greater impact of change than P2, and the least influence occurs at P3, the furthest downstream point.
Implementation of ponds in middle catchment (Scenario M)
Figure 11(b) shows the results for this scenario. The novel aspect of this situation is that, because the intervention occurs in the middle catchment, a greater change in P2 is observed. However, another thing that can be inferred is that there is no change in the streamflow on the upstream part of the intervention, i.e., no change occurs in P1.
Implementation of ponds in lower catchment (Scenario L)
As the intervention occurs in the lower catchment, change is only observed in the lowermost point, P3, as both P1 and P2 are at 0 (Figure 11(c)). In contrast to all other scenarios where the discharge decreases in the months January to March, it increases in this case. This indicates that if we deploy ponds exclusively in the lower portion of the catchment, we may be able to alleviate drought in the lower part of the catchment even during the initial dry months of the year.
Implementation of ponds in upper + middle catchment (Scenario UM)
Additionally, the results of this scenario (Figure 11(d)) indicate that expanding the area of intervention increases the influence downstream. For example, when the intervention is done in upper and middle catchments, the percentage changes in the values in P2 and P3 were greater than they were individually: the reduction percentage in P2. However, P1 remains unchanged from when only the upper catchment was intervened.
Similar results could be interpreted from Scenario ML (Figure 11(e)) and Scenario UML (Figure 11(f)).
DISCUSSION
Mean monthly variation in water balance components considering all HRUs of the upper catchment (implementation of ponds is done only in the upper catchment, the 0 value is for the baseline, positive values are the percentage increase, and the negatives are the decrease).
Mean monthly variation in water balance components considering all HRUs of the upper catchment (implementation of ponds is done only in the upper catchment, the 0 value is for the baseline, positive values are the percentage increase, and the negatives are the decrease).
CONCLUSION
This study looks at the potential of ponds as an NBS in the complex context of flood and drought control in the Bagmati River Basin, Nepal. The use of the hydrological model SWAT+ allowed for a thorough examination of the impact of incorporating ponds into agricultural landscapes on water balance components and, as a result, the change in flood and drought events.
Addressing floods and droughts concurrently is inherently challenging due to the antithetical nature of these two phenomena. Yet, this study unveils a modest but meaningful potential of ponds in mitigating both. By retaining water and reducing surface run-off, ponds contribute to lowering flood intensity – a significant finding given the extensive damage floods cause annually in the region. Similarly, water retention by ponds aids in enhancing soil moisture content, providing some relief during periods of drought.
Notwithstanding, the research underscores that while post-monsoon soil moisture improvement was evident, a prolonged dry season could present challenges. Moreover, it was found that the influence of ponds remains confined to their immediate surroundings and downstream areas, and they have minimal impact on upstream regions. Their effectiveness increases in tandem with the scale of the intervention area.
The investigation was comprehensive, encompassing six different scenarios for NBS implementation, spanning all agricultural HRUs. Yet, in practice, it is recommended to study field conditions meticulously to optimize the location and number of interventions.
While ponds on their own exerted a minor impact on floods and droughts, their effect could be substantially amplified when combined with other NBS strategies, such as conservation tillage or soil conservation techniques. These synergistic interactions offer promising avenues for future research.
DATA AVAILABILITY STATEMENT
All relevant data are included in the paper or its Supplementary Information.
CONFLICT OF INTEREST
The authors declare there is no conflict.
