Potential flash flood dynamics in the humid tropics: insights from three basins in Sumatra

Abstract

Flash floods in humid tropical mountain basins are highly destructive hydrometeorological hazards, yet comparative studies that integrate geomorphological and hydrometeorological conditions remain limited. This study aims to identify and synthesize potential flash flood dynamics in three mountain basins in Sumatra, namely the Peusangan, Wampu, and Anai basins, through an integrated analysis of six key parameters: basin morphometry, high landslide susceptibility zones on steep slopes, land use related modeled runoff potential, event rainfall, antecedent wetness, and antecedent soil moisture. The results indicate that the three basins exhibit contrasting physical characteristics and hydrological response potential. The Wampu Basin has the highest drainage density, at 5.41 km/km², but also the longest time of concentration, at 28.8 h, whereas the Anai Basin has the shortest time of concentration, at 5.8 h. Spatial linkage between susceptible slope areas and the river network is stronger in the Peusangan and Anai basins, whereas in the Wampu Basin, 47.0% of susceptible slope areas are located more than 500 m from the drainage network. Land use related modeled runoff potential is most evident in the Peusangan Basin, where the runoff coefficient increases from 0.34 to 0.36 and runoff volume increases from 82.67 to 87.91 million m³, while changes are more limited in the Wampu and Anai basins. From a hydrometeorological perspective, event rainfall estimates range from 8.1 to 13.4 mm/h, while 5 day antecedent rainfall reaches 114.4 to 186.5 mm. Satellite and reanalysis based antecedent soil moisture indicators suggest wet to high moisture conditions before the events, with near saturated and saturated classes present within the basins. Overall, potential flash flood dynamics in these humid tropical mountain basins can be interpreted through the combined information from preconditioning factors and triggering rainfall. The comparative framework developed in this study provides a basis for preliminary flash flood assessment and disaster risk reduction in data scarce humid tropical regions.

1 Introduction

Flash floods are among the most destructive hydrometeorological hazards because they develop rapidly, provide very limited warning time, and generate high energy flows that can cause severe damage to infrastructure, settlements, and human safety (Marchi et al., 2010; Al-Rawas et al., 2024). Their significance becomes even greater in humid tropical mountain environments, where intense rainfall, steep relief, and rapid basin response create favorable conditions for destructive flood processes (Wohl et al., 2012; Quichimbo-Miguitama et al., 2024; Azizah et al., 2022). In this context, understanding the physical and hydrometeorological conditions associated with flash flood occurrence is essential not only for interpreting documented events, but also for strengthening the scientific basis of disaster risk reduction and flash flood preparedness (Borga et al., 2014; Stoffel et al., 2016).

Flash floods are not determined by event rainfall alone (Borga et al., 2011; Gaume et al., 2009). Previous studies have shown that flash flood response can be interpreted through the interaction among basin morphometry, drainage network efficiency, slope conditions, upstream instability, land use, soil infiltration capacity, antecedent wetness, and short duration triggering rainfall (Borga et al., 2014; Azizah et al., 2019; Andrieu et al., 2021). In this process based understanding, flash flood dynamics refers to the way preconditioning factors and triggering rainfall interact to shape flash flood response within a basin. Flash floods are therefore better understood as events associated with combined preconditioning and triggering factors rather than as the response to a single parameter (Liu et al., 2018; Zhao et al., 2019; Marchi et al., 2010). In this context, landslide susceptibility is relevant to flash flood analysis because susceptible hillslope areas may contribute material to river channels when source areas are hydrologically and geomorphologically connected to the drainage network. This process may increase sediment availability, reduce channel conveyance, and contribute to sediment laden or debris rich flash flood flow during intense rainfall events (Borga et al., 2014; Catane et al., 2012). Studies on rainfall thresholds for landslides further show that triggering rainfall needs to be interpreted together with antecedent rainfall, soil moisture, and local physical conditions because these factors influence slope and runoff response (Gonzalez et al., 2024; Yuniawan et al., 2022; Zhang et al., 2022).

However, much of the current conceptual understanding of flash floods has been developed from studies in subtropical and temperate regions (Marchi et al., 2010; Ma et al., 2018; Bae et al., 2018; Spitalar et al., 2014), whereas studies from humid tropical mountain basins remain relatively limited. This gap is important because humid tropical basins are characterized by high rainfall input, rapid hydrological response, intense weathering, steep relief, and abundant material supply (Wohl et al., 2012; Azizah et al., 2022; Muñoz-Villers and McDonnell, 2013; Beck et al., 2013; Labrière et al., 2015). Studies from tropical regions indicate that hydrological extremes and basin response in wet environments are influenced by the interaction between rainfall and basin physical characteristics, yet comparative analyses across mountain basins that integrate geomorphological characteristics, slope conditions, land use, and antecedent wetness within a single framework remain limited (Espinoza et al., 2022; Macalalad et al., 2021; Gonzalez et al., 2024; Nofita et al., 2025).

In Sumatra, previous studies from mountain basins indicate that flash floods and related hazards develop through the interaction of physiographic and hydrometeorological factors (Thoha et al., 2023; Azizah et al., 2022; Syofyan et al., 2025). These factors include rainfall and antecedent rainfall, steep slopes, geological conditions and landslides, soil infiltration capacity, land use change, and basin runoff response (Azizah et al., 2019; Ilhamni et al., 2023; Amri et al., 2023; Nofita et al., 2025). Findings from North Sumatra, Aceh, and West Sumatra consistently highlight the importance of topography, soil conditions, erosion, landslides, land use, and hydrological conditions in shaping the response of mountain basins (Azizah et al., 2021; Azmeri et al., 2016; Azmeri and Isa, 2018). Taken together, these findings suggest that flash flood dynamics in Sumatra are better understood through multiple interacting physical and hydrometeorological conditions rather than rainfall at the time of the event alone (Thoha et al., 2023; Azizah et al., 2022; Syofyan et al., 2025; Andriani et al., 2024).

This study aims to identify and synthesize potential flash flood dynamics in three mountain basins in Sumatra, namely the Peusangan, Wampu, and Anai basins, through an integrated analysis of six key parameters: basin morphometry, high landslide susceptibility zones on steep slopes, land use related modeled runoff potential, event rainfall, antecedent wetness, and soil moisture. Although these parameters have been addressed in previous studies, this study applies them within a common comparative framework across three documented flash flood events in humid tropical mountain basins. The three basins were selected to provide a basis for examining variations in basin characteristics, hydrometeorological conditions, and flash flood response within a comparable regional setting. By doing so, this study provides a comparative basis for interpreting potential flash flood dynamics in humid tropical mountain basins and for informing preliminary disaster risk reduction assessment.

2 Study area

Our study sites are located on Sumatra Island, Indonesia, which lies within the humid tropical climatic zone. This study was conducted in three river basins, namely the Peusangan, Wampu, and Anai basins (Figure 1), which have experienced recurrent flash flood events. Administratively, the Peusangan Basin is located in Aceh Province, the Wampu Basin in North Sumatra Province, and the Anai Basin in West Sumatra Province. We selected the most devastating event in each basin, particularly in terms of fatalities and damage, as a case study. These events were considered to represent high impact flash flood occurrences and to provide an appropriate basis for comparative analysis (Ilhamni et al., 2023; Thoha et al., 2023; Andriani et al., 2024).

Figure 1

The three basins are situated along the Bukit Barisan mountain range, an elongated orogenic belt that forms the main physiographic backbone of Sumatra and originated from tectonic interaction between the Indo Australian and Eurasian plates (Hamilton, 1979; Barber et al., 2005). Each basin extends from upstream to downstream areas and exhibits marked elevational contrasts, reflecting the strong topographic gradients that characterize mountain river systems in humid tropical environments.

We identified the locations of the most devastating flash flood events for each basin. In the Peusangan Basin, the event occurred on 13 September 2015 in Damaran Baru, while the events in the Wampu Basin and the Anai Basin occurred on 2 November 2003 in Bukit Lawang and 11 May 2024 in Lembah Anai, respectively. Within all identified events, the damage occurred in downstream areas. These sites represent the downstream impact zones, whereas the triggering processes originated from the upstream parts of the basins, which are dominated by steep terrain. This upstream to downstream linkage reflects the geomorphic and hydrological connectivity that shapes flash flood development in mountainous catchments.

Substantial differences in basin size and geomorphological characteristics are evident among the Peusangan, Wampu, and Anai basins, as summarized in Table 1. The Wampu Basin is the largest basin, with an area of 4,173.4 km² and the longest main channel, whereas the Anai Basin is the smallest basin, with an area of 702.6 km² and the shortest time of concentration. The Peusangan Basin occupies an intermediate position in terms of basin area, main channel length, and hydrological response. These contrasts provide a suitable basis for comparing flash flood dynamics across mountain river systems in the humid tropics.

3 Materials and methods

3.1 Morphometric analysis

Morphometric analysis was conducted to quantify basin parameters including geometry, drainage network structure, relief conditions, and hydrological response indicators in the three study basins. The selected parameters were intended to represent geomorphological characteristics associated with flash flood response, particularly those related to basin shape, drainage network efficiency, and topographic elevation differences. Morphometric indices were used to evaluate the physical characteristics of the basins that are relevant to hydrological response to rainfall, especially in relation to runoff concentration and flow dynamics (Obeidat et al., 2021; Dutal, 2023; Dilip et al., 2025).

Basin boundaries and drainage networks were derived from a 30 m resolution Digital Elevation Model (DEM), which was downloaded from the USGS Earth Explorer and processed in a GIS environment. Basin delineation was carried out using flow direction and flow accumulation analysis with a consistent flow accumulation criterion applied across the three basins. The extracted drainage system was then used as the basis for calculating all morphometric parameters, allowing direct comparison among basins.

Relief conditions were analyzed through basin relief and relief ratio, which represent elevational differences and topographic gradients within each basin. Slope was derived from the DEM using the slope function and classified into five classes, namely flat, gentle, moderately steep, steep, and very steep. Basin hydrological response was further represented by the time of concentration, which was estimated using the empirical Kirpich method. Accordingly, the time of concentration was used as a comparative morphometric indicator of basin response potential, rather than as a direct reconstruction of observed flash flood timing.

3.2 Landslide susceptibility and slope analysis

This analysis was conducted using the official landslide susceptibility zonation map issued by the Centre for Volcanology and Geological Hazard Mitigation, PVMBG, together with a slope map derived from the 30 m resolution DEM. In this study, landslide susceptibility classes followed the PVMBG zonation as provided, which classifies the landscape into four categories: low, moderate, high, and very high. All spatial data were clipped to the boundary of each basin in order to maintain analytical consistency across the three study areas.

Areas classified as high and very high landslide susceptibility were then extracted and overlaid with the slope map. In this study, steep slopes were defined as areas with a slope gradient greater than 25°. The area of overlap between high landslide susceptibility zones and steep slopes was calculated for each basin to identify sectors where slope instability may be more likely.

The spatial relationship between these overlap areas and the drainage network was further examined using river buffers with three distance classes, namely 0 to 100 m, 100 to 500 m, and 500 to 1,000 m. This approach was used to evaluate the spatial proximity between susceptible slope areas and the drainage system as a basis for interpreting slope to channel linkage in each basin. The buffer analysis provided a basin scale proxy for potential slope to channel connectivity, rather than a reconstruction of event specific landslide initiation or sediment delivery during the flood events. Before the overlay and buffer analyses were performed, all spatial layers were processed in a GIS environment through reprojection, clipping, and rasterisation so that they shared a common coordinate system and spatial resolution.

3.3 Land use and hydrological analysis

Land use analysis was conducted using Landsat imagery representing pre event and event year land use conditions in each basin, namely Landsat 5 TM for 2006 and Landsat 8 OLI/TIRS for 2015 in the Peusangan Basin, Landsat 5 TM for 1990 and 2003 in the Wampu Basin, and Landsat 8 OLI/TIRS for 2013 and Landsat 9 OLI 2/TIRS 2 for 2024 in the Anai Basin. The image years were selected according to the timing of the documented flash flood event in each basin, so that land use conditions could be examined within the temporal context of each event. All images had a spatial resolution of 30 m and were processed to produce land use maps for each observation period.

Land use classification was performed using supervised classification with the Maximum Likelihood Classification algorithm. This method assigns each pixel to the class with the highest statistical probability based on the mean vector and covariance matrix of the training samples. Training samples were selected through visual interpretation of homogeneous land cover areas supported by high resolution imagery, with 30 to 45 samples for each class, equivalent to approximately 600 samples in total. Class separability was evaluated on the basis of spectral distribution before classification.

Classification accuracy was assessed using a confusion matrix based on approximately 250 independent validation points selected through stratified random sampling. The evaluation showed that overall accuracy ranged from 84 to 87%, with Kappa values ranging from 0.79 to 0.84.

The resulting land use maps were then clipped to the boundary of each basin and analyzed using a transition matrix derived from the overlay of land use maps from the two observation periods. To evaluate the hydrological implications of land use change, curve number values were calculated spatially in a GIS environment based on the combination of land use classes and hydrological soil groups.

Runoff coefficients and runoff volumes were estimated using the Soil Conservation Service Curve Number (SCS CN) approach as modeled indicators of runoff potential under the selected land use conditions. The standard SCS CN initial abstraction assumption, Ia = 0.2S, was used in the calculation. The estimated runoff coefficients and runoff volumes were interpreted as comparative modeled indicators rather than observed discharge responses. In this calculation, rainfall input was defined as the cumulative event rainfall up to the time of flash flood occurrence and was applied consistently to both the pre event and event year land use conditions within each basin.

The SCS CN approach was used as a parsimonious method to compare runoff potential among basins because it links rainfall, land use, hydrological soil group, and curve number within a simple event based runoff formulation (Soulis and Valiantzas, 2012; Hong et al., 2010). In this formulation, the storage parameter S is derived from curve number values and therefore reflects the selected CN assumptions rather than an independently calibrated basin parameter. Because event scale discharge observations were unavailable, the resulting runoff coefficients and runoff volumes were used as comparative modeled indicators of runoff potential. Hydrogeomorphic processes such as debris flow, sediment bulking, channel blockage, and sediment laden flow were not represented directly in the SCS CN calculation, but were considered separately in the interpretation of slope to channel linkage and flash flood dynamics (Borga et al., 2014; Catane et al., 2012).

3.4 Hydrometeorological analysis

Hydrometeorological analysis was conducted to characterize event rainfall and antecedent wetness conditions before and during the documented flash flood events in the Peusangan, Wampu, and Anai basins. Rainfall estimates were obtained from the Global Satellite Mapping of Precipitation (GSMaP) product developed by the Japan Aerospace Exploration Agency, JAXA. This product provides satellite based hourly rainfall estimates at a spatial resolution of 0.1° × 0.1° and has been widely used in hydrological and climatological studies because it offers spatially and temporally consistent rainfall information in regions with limited rain gauge observations (Li et al., 2024; Wu et al., 2026). In this study, GSMaP was selected because it provides consistent hourly rainfall estimates across the three event years analyzed, thereby allowing event rainfall characteristics to be examined among basins within a common analytical framework.

The reliability of GSMaP was evaluated against available rain gauge observations within the study area. Validation was carried out using seven rainfall stations, namely Malikul Saleh for the Peusangan Basin, Tuntungan for the Wampu Basin, and Padang Pariaman, Kandang IV, Kasang, Paraman Talang, and Canduang for the Anai Basin and its surrounding area. The evaluation was performed at hourly, daily, and monthly time scales using three statistical indicators, namely the Pearson correlation coefficient, percent bias, and root mean square error. The evaluation showed that the hourly correlation coefficient was 0.28, with RMSE of 2 mm/h and PBIAS of −53%. Daily correlation coefficients ranged from 0.20 to 0.31, with RMSE values of 13 to 28 mm/day and PBIAS ranging from −59 to 86%. Monthly correlation coefficients ranged from 0.60 to 0.85, with RMSE values of 83 to 314 mm/month and PBIAS remaining within the same range. These results indicate that GSMaP provides more consistent rainfall estimates at longer temporal scales, while its hourly estimates have greater uncertainty in representing short duration rainfall variability.

Because satellite rainfall products commonly contain systematic bias relative to ground observations, especially during intense rainfall events, GSMaP data were subsequently corrected using the linear scaling method before being used in the event rainfall and API5 analyses (Omay et al., 2025; Mashuri et al., 2025). However, linear scaling reduces systematic bias but does not correct rainfall timing errors or improve temporal correlation. Therefore, the event scale rainfall estimates in this study should be interpreted as bias corrected satellite based estimates with uncertainty at the hourly scale.

Event rainfall characteristics were then analyzed using the bias corrected hourly GSMaP estimates. The analysis focused on three flash flood events, namely 13 September 2015 in the Peusangan Basin, 2 November 2003 in the Wampu Basin, and 11 May 2024 in the Anai Basin. Three main parameters were extracted from these data, namely rainfall intensity at the time of the event, cumulative rainfall during the 6 hours before the event, and cumulative rainfall during the 5 days before the event. The rainfall intensity values reported for each basin were derived from basin averaged bias corrected hourly GSMaP estimates. These values should be interpreted as basin scale satellite rainfall estimates, not as point gauge observations or rainfall from a single GSMaP grid cell at the documented flash flood location. In addition, 24 h hyetographs were constructed for each event to describe the temporal distribution of rainfall, the sequence of rainfall pulses, and the concentration of rainfall accumulation prior to flash flood occurrence. Rainfall intensity classes followed the categories used by the Indonesian Agency for Meteorology, Climatology, and Geophysics, BMKG.

Antecedent wetness conditions were first evaluated using the 5 day Antecedent Precipitation Index, API5, calculated from the bias corrected hourly GSMaP estimates. API was computed following the formulation of Fedora and Beschta (1989), which accounts for the declining influence of antecedent rainfall through a decay constant, as expressed by: API_i = k(API_i-1) + P_i, where API_i is the antecedent precipitation index at time i, P_i is rainfall at time i, and k is the decay constant. To account for uncertainty in the API decay constant, API5 was calculated using four representative k values, namely 0.80, 0.85, 0.90, and 0.95, based on ranges reported in previous API applications (Hong et al., 2010; Li et al., 2021; Li et al., 2026). The same set of k values was applied across the three basins to maintain comparability in the API5 sensitivity assessment. API5 was treated as a rainfall based indicator of antecedent wetness conditions at the basin scale rather than as a direct measurement of soil moisture, and it has been used in hydrological and landslide studies to represent wet conditions before extreme events (Bainbridge et al., 2022). This formulation provides a rainfall based estimate of antecedent wetness, although evapotranspiration, deep percolation, and spatial variation in soil water storage are not explicitly represented.

Antecedent wetness conditions were also complemented by soil moisture data. In the Peusangan and Anai basins, pre event soil moisture conditions were derived from Soil Moisture Active Passive, SMAP, with a spatial resolution of 9 km. In the Wampu Basin, because the flash flood event occurred in 2003, before the SMAP observation period, antecedent soil moisture was represented using ERA5 Land. SMAP and ERA5 Land were used according to the availability of soil moisture products for the respective event years, and the resulting soil moisture information was interpreted within the context of each basin and product. Soil moisture values were then classified into five interpretative classes, namely very dry (0.00 to 0.15), dry to moderate (0.15 to 0.30), wet (0.30 to 0.40), near saturated (0.40 to 0.45), and saturated (>0.45), in order to evaluate the position of flash flood locations relative to the spatial distribution of antecedent wetness conditions in each basin. The near saturated and saturated classes were defined with reference to the concept that runoff response may increase under high antecedent soil moisture conditions (Penna et al., 2011).

Accordingly, the hydrometeorological analysis in this study was built on three complementary components, namely event rainfall, API5, and antecedent soil moisture. These components were used jointly to describe rainfall forcing and antecedent wetness conditions before flash flood occurrence. Event scale discharge measurements were not available for the three documented flash flood events. Ground based soil moisture observations were also unavailable, so antecedent soil moisture was interpreted from SMAP and ERA5 Land products rather than from in situ measurements. Uncertainty also remains because satellite and reanalysis products may not fully resolve convective rainfall cores and local orographic effects in mountainous humid tropical terrain. Because the analysis was conducted at the basin scale, the resulting indicators should be interpreted as basin scale conditions rather than as direct descriptors of event specific contributing areas. The available historical event information did not allow reliable delineation of contributing areas associated with precise upstream initiation zones and flow paths. Consequently, morphometric indicators, modeled runoff estimates, and antecedent soil moisture classes may not fully represent the local conditions within the actual areas that contributed to each flash flood.

4 Results and discussion

4.1 Morphometric characteristics

The morphometric characteristics of the three basins are presented in Table 1. All three basins show elongated shapes, although with clear differences in drainage configuration, relief, and hydrological response potential. In general, morphometric characteristics are important for interpreting basin response to rainfall, particularly through their influence on runoff concentration and flow dynamics (Obeidat et al., 2021; Bashir and Alsalman, 2024).

Drainage network development differs markedly among the three basins. The Wampu Basin shows the highest total stream length and drainage density, whereas the Peusangan Basin has the lowest drainage density among the three basins. These differences reflect variation in branching pattern and flow connectivity, which in turn influence runoff concentration processes. Similar relationships between drainage configuration and basin response have also been reported in humid tropical basins, including those in Indonesia (Indarto and Hidayah, 2019).

Relief characteristics further indicate that all three basins have high topographic energy, but with contrasting slope conditions. The Anai Basin is characterized by steeper terrain, whereas the Wampu Basin is relatively gentler. These differences are relevant to overland flow velocity, as steeper slopes tend to accelerate surface runoff and increase the potential for sediment transport. Such topographic effects are commonly observed in humid tropical mountain basins (Bashir and Alsalman, 2024).

These morphometric differences are also reflected in the variation in time of concentration, which indicates contrasting hydrological response potential among basins. Smaller basins with steeper slopes tend to respond more rapidly, whereas larger basins show a slower response. Overall, these results indicate that basin hydrological response can be interpreted through the interaction between basin shape, drainage network development, and relief conditions, as has also been reported for humid tropical basins elsewhere (Obeidat et al., 2021; Indarto and Hidayah, 2019).

4.2 Landslide susceptibility, slope, and drainage proximity

The spatial distribution of areas with high landslide susceptibility on steep slopes shows clear differences among the three basins (Figure 2). The Peusangan Basin has the largest proportion of overlap area relative to basin area, reaching 28.8%, followed by the Wampu Basin at 18.8% and the Anai Basin at 10.8%. This pattern indicates that areas combining high landslide susceptibility and steep slopes occupy the largest proportional extent in the Peusangan Basin, whereas in the Wampu Basin the overlap area is concentrated mainly in the western to southwestern part of the basin and contributes less to the overall basin area.

Figure 2

The analysis of proximity to the drainage network reveals contrasting spatial connectivity among basins. In the Peusangan Basin, almost the entire overlap area lies within 500 m of the river network, with 48.0% located within 0 to 100 m and 50.9% within 100 to 500 m. A similar pattern is observed in the Anai Basin, where 69.8% of the overlap area falls within 0 to 100 m and 30.1% within 100 to 500 m. These patterns indicate that susceptible slope areas in both basins are located close to the drainage network.

In contrast, the Wampu Basin shows a different pattern. Only 13.1% of the overlap area lies within 0 to 100 m of the river network, whereas a substantial proportion is located more than 500 m away, reaching 47.0%. This indicates that many susceptible slope areas in the Wampu Basin are located farther from the river system.

These differences reflect variation in spatial linkage between susceptible slope areas and the drainage network among basins. In the Peusangan and Anai basins, the close proximity between overlap areas and the river network suggests stronger slope to channel linkage. In contrast, in the Wampu Basin, the greater distance between overlap areas and the drainage network suggests lower spatial connectivity. This finding is consistent with the view that spatial proximity between sediment source areas on slopes and the channel network is important for interpreting material transfer efficiency within a basin (Bracken et al., 2015; Heckmann and Schwanghart, 2013).

4.3 Land use and hydrological response

Comparatively, land use change across the three basins shows different magnitudes and directions, with modeled runoff implications that are likewise not uniform. The Peusangan Basin experienced the most pronounced land use change, the Wampu Basin underwent more moderate and directed change, whereas the Anai Basin remained relatively stable. These differences are reflected not only in the changes in land use classes, but also in the spatial distribution of curve number values (Figure 3) and the estimated runoff coefficients and runoff volumes prior to flash flood occurrence. These contrasts were interpreted within the event specific observation period of each basin.

Figure 3

In the Peusangan Basin, the most pronounced changes were the decline in shrubland from 32.8 to 8.9%, the increase in mixed dryland agriculture with shrubs from 25.0 to 50.1%, and the increase in settlement area from 0.7 to 2.4%. The two largest transitions were the conversion of shrubland to mixed dryland agriculture with shrubs over 627.7 km² and the conversion of secondary forest to the same class over 30.3 km². As shown in Figure 3, the spatial distribution of curve number in 2015 indicates that areas with moderate to high modeled runoff potential became more extensive than in 2006, whereas areas with low curve number decreased. This pattern is consistent with the estimated increase in runoff volume from 82.67 million m³ to 87.91 million m³ and the increase in runoff coefficient from 0.34 to 0.36. These results indicate that land use change in the Peusangan Basin was associated with higher modeled surface runoff potential prior to the flash flood event.

In the Wampu Basin, land use change was more moderate. The three most visible changes were the decrease in shrubland from 3.95 to 0.40%, the increase in plantation forest from 0.40 to 3.89%, and the increase in plantation area from 7.08 to 8.38%. Accordingly, the two most prominent directions of change in this basin were the conversion of transitional vegetation into plantation forest and plantation land. The spatial distribution of curve number in Figure 3 shows that the moderate class remained dominant across most of the basin, while the low class persisted mainly in the western part. This spatial shift was relatively limited, which is reflected in the small change in modeled runoff indicators, with runoff volume increasing only from 253.84 million m³ to 254.66 million m³ and the runoff coefficient from 0.452 to 0.454. Thus, land use change in the Wampu Basin was associated with only a limited change in modeled runoff potential compared with the Peusangan Basin.

The Anai Basin shows the most limited land use change. The three most relevant changes were the decline in dryland agriculture from 24.79 to 14.58%, the increase in shrubland from 1.35 to 7.47%, and the increase in secondary forest from 16.75 to 18.00%. The two main directions of change in this basin indicate a shift from cultivated land toward more semi natural land cover. This pattern is consistent with Figure 3, which shows that the curve number distribution in 2024 remained dominated by low to moderate classes, while high values appeared only in limited locations. As a result, the change in modeled runoff indicators was very small, with runoff volume changing only from 7.96 million m³ to 7.98 million m³ and the runoff coefficient remaining around 0.17. This indicates that, during the observation period, land use change in the Anai Basin was not associated with a substantial increase in modeled runoff potential prior to the flash flood event.

4.4 Hydrometeorological characteristics

4.4.1 Event rainfall and antecedent wetness conditions

The characteristics of event rainfall and antecedent wetness conditions indicate that the three flash flood events developed under a combination of short duration triggering rainfall and basin conditions that had already become wet before the events. In flash flood studies, rainfall occurring within a few hours before the event is particularly important because small to medium sized basins commonly respond rapidly over hourly time scales, while the effect of triggering rainfall becomes stronger when it occurs over a basin that is already wet. In addition, the role of multi day rainfall accumulation before an event has also been reported as an important factor in the development of major floods during extreme rainfall events in South Asia (Marchi et al., 2010; Norbiato et al., 2008).

The API5 sensitivity curves in Figure 4 show that higher k values produced higher API5 estimates and slower antecedent wetness recession, while the temporal pattern of wetness build up remained generally consistent across the tested k scenarios.

Figure 4

Based on Table 2, the flash flood in the Peusangan Basin occurred at 16:30 WIB with an event rainfall estimate of 8.1 mm/h. Rainfall accumulated during the 6 hours before the event reached 18.0 mm, whereas cumulative rainfall during the preceding 5 days reached 186.5 mm. The hyetograph and API5 sensitivity curves in Figure 4a show that rainfall had already occurred since 11 September and increased again shortly before the event. This pattern indicates that antecedent wetness conditions had already developed before the triggering rainfall occurred.

In the Wampu Basin, the flash flood occurred at 21:30 WIB with an event rainfall estimate of 11.2 mm/h. Rainfall accumulated during the 6 hours before the event reached 33.4 mm, whereas cumulative rainfall during the preceding 5 days reached 114.4 mm. The hyetograph and API5 sensitivity curves in Figure 4b show that rainfall was concentrated within the hours immediately before the event, while antecedent wetness remained evident across the tested k scenarios. This condition reflects a combination of short duration triggering rainfall and high antecedent wetness conditions before flash flood occurrence.

In the Anai Basin, the flash flood occurred at 22:00 WIB with an event rainfall estimate of 13.4 mm/h. Rainfall accumulated during the 6 hours before the event reached 54.6 mm, whereas cumulative rainfall during the preceding 5 days reached 186.2 mm. The hyetograph and API5 sensitivity curves in Figure 4c show that rainfall increased from late afternoon into the evening and coincided with the timing of the event, while API5 remained high prior to flash flood occurrence under the tested k scenarios. This pattern indicates that short duration rainfall and antecedent wetness conditions were present together before the flash flood in the Anai Basin.

4.4.2 Antecedent soil moisture

The analysis of antecedent soil moisture reveals clear differences in the spatial distribution of soil moisture among the three basins before flash flood occurrence (Figure 5). In the Peusangan Basin, the wet class dominates 72.3% of the basin area, whereas the near saturated and saturated classes account for 8.6 and 4.8%, respectively. In the Wampu Basin, antecedent conditions are characterized by extensive near saturated and saturated areas, with the near saturated class covering 73.8% of the basin and the saturated class 20.0%. In contrast, the Anai Basin is dominated by the wet class, which covers 56.7% of the basin area, although the near saturated and saturated classes still occupy substantial proportions, reaching 32.3 and 10.7%, respectively. These patterns indicate that all three flash flood events were preceded by high antecedent soil moisture conditions before event rainfall occurred. In steep basins, antecedent soil moisture conditions of this kind are important for interpreting runoff response under short duration triggering rainfall (Penna et al., 2011; Norbiato et al., 2008; Wasko and Nathan, 2019).

Figure 5

In the Peusangan Basin, shown in Figure 5, soil moisture at the flash flood location in Damaran Baru reached 0.33 before the event, placing it within the wet class. The flash flood location does not coincide with the highest soil moisture zone, but the central to southern part of the basin shows a clear development of near saturated to saturated conditions. This distribution indicates that, although the event location itself was classified as wet, parts of the basin had already reached very wet conditions before the flash flood occurred.

In the Wampu Basin, soil moisture at the flash flood location in Bukit Lawang reached approximately 0.45 before the event, placing it near the boundary between the near saturated and saturated classes. As shown in Figure 5, much of the central part of the basin is dominated by near saturated to saturated conditions, whereas lower soil moisture classes appear only in limited areas in the north. This pattern indicates that the flash flood in the Wampu Basin occurred under antecedent conditions that were already very wet before the triggering rainfall occurred.

In the Anai Basin, shown in Figure 5, soil moisture at the flash flood location in Lembah Anai reached 0.42 before the event, placing it within the near saturated class. The event location coincides with a near saturated zone, while most of the remaining basin is still dominated by the wet class. Saturated conditions are also present, but they are spatially limited to the southwestern part of the basin. Thus, antecedent soil moisture around the flash flood location in the Anai Basin remained high, with a more localized spatial extent within the basin.

4.5 Synthesis of flash flood dynamics

A cross basin analysis of the Anai, Wampu, and Peusangan basins indicates that potential flash flood dynamics in humid tropical environments can be interpreted through the combined information from physical basin characteristics and hydrometeorological conditions. In general, flash flood occurrence in the three basins was not characterized by a single parameter, but by the combination of pre event conditions and triggering rainfall (Wohl et al., 2012; Filho et al., 2021; Salgado and Nájera, 2022).

In this study, the synthesis of flash flood dynamics was developed from six main parameters, namely basin morphometry, high landslide susceptibility zones on steep slopes, land use related modeled runoff potential represented by curve number and runoff coefficient, event rainfall, antecedent wetness conditions, and soil moisture. These six parameters were used to characterize and compare flash flood dynamics across the three study basins. Some of these parameters are also consistent with previous findings from the Tamiang Basin, where antecedent precipitation index and curve number were used to interpret flash flood potential in a humid tropical setting (Azizah et al., 2022).

Morphometric characteristics reveal clear differences in basin physical characteristics and hydrological response potential among the three basins. Variation in drainage density and time of concentration indicates that each basin has a distinct flow configuration. The Wampu Basin has the highest drainage density, at 5.41 km/km², but also the longest time of concentration, at 28.8 h, whereas the Anai Basin has the shortest time of concentration, at 5.8 h. The Peusangan Basin occupies an intermediate position, with a time of concentration of 14.8 h. These results indicate that morphometric differences help interpret variation in basin response potential (Obeidat et al., 2021; Dutal, 2023).

Slope conditions and landslide susceptibility are reflected in the distribution of areas with high landslide susceptibility on steep slopes and in their proximity to the drainage network. The Peusangan and Anai basins show greater spatial proximity between susceptible slope areas and the river network, whereas in the Wampu Basin many susceptible slope areas are located farther from the channel system. This pattern indicates variation in spatial linkage between susceptible slopes and the drainage network among the three basins, which helps indicate potential slope to channel linkage (Bracken et al., 2015; Catane et al., 2012).

Land use was represented through curve number and modeled runoff indicators. The Peusangan Basin shows an increase in runoff coefficient from 0.34 to 0.36, the Wampu Basin changes only slightly from 0.452 to 0.454, whereas the Anai Basin remains low and relatively stable at around 0.17. These results indicate that land use related modeled runoff potential was not uniform among basins, but it remains an important component for interpreting differences in hydrological response potential (Muñoz-Villers and McDonnell, 2013).

The hydrometeorological factors in this study consist of two components with different roles. Antecedent wetness conditions and soil moisture were interpreted as preconditioning indicators, because both indicate that the basins were already in a wet state before the events occurred. In all three basins, this condition is reflected by high API5 values and by soil moisture classes ranging from wet to saturated. In contrast, event rainfall was interpreted as the triggering component because rainfall occurring within the event timescale preceded flash flood occurrence. These results indicate that the three flash flood events can be understood as rainfall events occurring over basins with high antecedent wetness conditions (Marchi et al., 2010; Norbiato et al., 2008; Borga et al., 2014).

Conceptually, the six parameters can be grouped into two main phases, namely the preconditioning phase and the triggering phase. The preconditioning phase includes basin morphometry, high landslide susceptibility zones on steep slopes, land use represented by curve number and runoff coefficient, antecedent wetness conditions, and soil moisture. The triggering phase is represented by event rainfall. This conceptual framework is summarized in Figure 6 and represents the main contribution of this study, namely the integration of six parameters into a single comparative framework for interpreting flash flood dynamics in humid tropical environments.

Figure 6

From the perspective of disaster risk reduction, these findings indicate that flash flood assessment is insufficient if it relies only on event rainfall. Instead, a multi parameter approach that considers both preconditioning and triggering information can support flash flood analysis and preparedness in humid tropical regions.

5 Conclusion

This study indicates that potential flash flood dynamics in the Peusangan, Wampu, and Anai basins can be interpreted through the combined information from geomorphological setting, land use related modeled runoff potential, antecedent wetness, antecedent soil moisture, and event rainfall. The three cases indicate that event rainfall alone does not fully characterize flash flood occurrence in humid tropical mountain basins. Instead, the events were associated with preconditioning conditions that reflected basin morphology, slope to channel linkage, land use related modeled runoff potential, antecedent wetness, and soil moisture, while short duration event rainfall represented the immediate triggering component.

The main contribution of this study lies in the integration of six parameters into a comparative framework that distinguishes between preconditioning and triggering phases of flash flood dynamics. This framework provides a comparative basis for interpreting documented flash flood events in humid tropical mountain basins, consistent with the descriptive design and data constraints of the study. The runoff coefficients and runoff volumes should therefore be read as modeled indicators of runoff potential, not as observed flood magnitude. From a disaster risk reduction perspective, the findings suggest that preliminary flash flood assessment in data scarce humid tropical regions should consider both antecedent basin conditions and event rainfall, rather than relying on rainfall information alone.

References

• 1

  Al-RawasG.NikooM. R.JanbehsarayiS. F. M.HassaniM. R.ImaniS.NiksokhanM. H.et al. (2024). Near future flash flood prediction in an arid region under climate change. Sci. Rep.14, 1–18. doi: 10.1038/s41598-024-76232-0,

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 2

  AmriU.AzizahC.ErnawitaRoboS.NuraidaIsmyR.et al. (2023). Lake cliff landslide mitigation – a case study of Lut Tawar Peusangan Lake, Aceh, Indonesia. J. Ecol. Eng.24, 165–172. doi: 10.12911/22998993/156699

  • CrossRef
  • Google Scholar

• 3

  AndrianiS.BambangI.RahmadY.MahdiM.Al GiffariM.ArianD.et al. (2024). Post-disaster engineering strategy for Anai River debris flow management on Anai Valley National Road West Sumatra Indonesia. J. Appl. Eng. Technol. Sci.6, 365–379. doi: 10.37385/jaets.v6i1.5845

  • CrossRef
  • Google Scholar

• 4

  AndrieuH.MoussaR.KirstetterP. (2021). The event-specific geomorphological instantaneous unit hydrograph (E-GIUH): the basin hydrological response characteristic of a flood event. J. Hydrol.603:127158. doi: 10.1016/j.jhydrol.2021.127158,

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 5

  AzizahC.PawitanH.DasantoB.RidwansyahI.TaufikM. (2019). Soil physical properties and their relationship with the infiltration capacity of Tamiang River basin. Jurnal Tanah dan Iklim.43, 167–173. doi: 10.21082/jti.v43n2.2019.167-173

  • CrossRef
  • Google Scholar

• 6

  AzizahC.PawitanH.DasantoB.RidwansyahI.TaufikM. (2022). Risk assessment of flash flood potential in the humid tropics Indonesia: a case study in Tamiang River basin. Int. J. Hydrology Sci. Technol.13, 57–73. doi: 10.1504/ijhst.2022.119236

  • CrossRef
  • Google Scholar

• 7

  AzizahC.PawitanH.NuraidaSatriawanH.AbbasR.RoboS.et al. (2021). Hydrological characteristics and its impact on flood Jambo Aye Basin in Aceh Indonesia. J. Watershed Manag. Res.5, 171–184. doi: 10.20886/jppdas.2021.5.2.171-184

  • CrossRef
  • Google Scholar

• 8

  AzmeriHadihardajaI.VadiyaR. (2016). Identification of flash flood hazard zones in mountainous small watershed of Aceh Besar regency, Aceh Province, Indonesia. Egypt. J. Remote Sens. Space Sci.19, 143–160. doi: 10.1016/j.ejrs.2015.11.001

  • CrossRef
  • Google Scholar

• 9

  AzmeriIsaA. (2018). An analysis of physical vulnerability to flash floods in the small mountainous watershed of Aceh Besar regency, Aceh Province, Indonesia. JAMBA10, 1–6. doi: 10.4102/jamba.v10i1.550

  • CrossRef
  • Google Scholar

• 10

  BaeD. H.LeeM. H.MoonS. K. (2018). Development of a precipitation-area curve for warning criteria of short-duration flash flood. Nat. Hazards Earth Syst. Sci.18, 171–183. doi: 10.5194/nhess-18-171-2018

  • CrossRef
  • Google Scholar

• 11

  BainbridgeR.LimM.DunningS.WinterM. G.Diaz-MorenoA.MartinJ.et al. (2022). Detection and forecasting of shallow landslides: lessons from a natural laboratory. Geomat. Nat. Hazards Risk13, 686–704. doi: 10.1080/19475705.2022.2041108

  • CrossRef
  • Google Scholar

• 12

  BarberA. J.CrowM. J.MilsomJ. S. (2005). Sumatra: Geology, Resources and Tectonic Evolution. London: The Geological Society.

  • Google Scholar

• 13

  BashirB.AlsalmanA. (2024). Morphometric characterization and dual analysis for flash flood hazard assessment of Wadi Al-Lith watershed, Saudi Arabia. Water16:3333, 1–21. doi: 10.3390/w16223333

  • CrossRef
  • Google Scholar

• 14

  BeckH. E.BruijnzeelL. A.Van DijkA. I. J. M.McVicarT. R.ScatenaF. N.SchellekensJ. (2013). The impact of forest regeneration on streamflow in 12 mesoscale humid tropical catchments. Hydrol. Earth Syst. Sci.17, 2613–2635. doi: 10.5194/hess-17-2613-2013

  • CrossRef
  • Google Scholar

• 15

  BorgaM.AnagnostouE. N.BlöschlG.CreutinJ. D. (2011). Flash flood forecasting, warning and risk management: the HYDRATE project. Environ. Sci. Pol.14, 834–844. doi: 10.1016/j.envsci.2011.05.017

  • CrossRef
  • Google Scholar

• 16

  BorgaM.StoffelM.MarchiL.MarraF.JakobM. (2014). Hydrogeomorphic response to extreme rainfall in headwater systems: flash floods and debris flows. J. Hydrol.518, 194–205. doi: 10.1016/j.jhydrol.2014.05.022

  • CrossRef
  • Google Scholar

• 17

  BrackenL. J.TurnbullL.WainwrightJ.BogaartP. (2015). Sediment connectivity: a framework for understanding sediment transfer at multiple scales. Earth Surf. Process. Landf.40, 177–188. doi: 10.1002/esp.3635

  • CrossRef
  • Google Scholar

• 18

  CataneS. G.AbonC.SaturayR. M.MendozaE. P. P.FutalanK. M. (2012). Landslide-amplified flash floods-the June 2008 Panay Island flooding, Philippines. Geomorphology169–170, 55–63. doi: 10.1016/j.geomorph.2012.04.008

  • CrossRef
  • Google Scholar

• 19

  DilipN.VegadU.MishraV. (2025). Drivers of flash floods in the Indian sub-Continental River basins. Nat. Hazards2, 1–12. doi: 10.1038/s44304-025-00121-3,

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 20

  DutalH. (2023). Using morphometric analysis for assessment of flash flood susceptibility in the Mediterranean region of Turkey. Environ. Monit. Assess.195:582. doi: 10.1007/s10661-023-11201-0,

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 21

  EspinozaJ. C.MarengoJ. A.SchongartJ.JimenezJ. C. (2022). The new historical flood of 2021 in the Amazon River compared to major floods of the 21st century: atmospheric features in the context of the intensification of floods. Weather Clim. Extremes35:100406. doi: 10.1016/j.wace.2021.100406

  • CrossRef
  • Google Scholar

• 22

  FedoraM. A.BeschtaR. I. (1989). Storm runoff simulation using an antecedent precipitation index (API) model. J. Hydrol.112, 121–133. doi: 10.1016/0022-1694(89)90184-4

  • CrossRef
  • Google Scholar

• 23

  FilhoG. M. R.CoelhoV. H. R.FreitasE. D. S.XuanY.AlmeidaC. D. N. (2021). An improved rainfall-threshold approach for robust prediction and warning of flood and flash flood hazards. Nat. Hazards105, 2409–2429. doi: 10.1007/s11069-020-04405-x

  • CrossRef
  • Google Scholar

• 24

  GaumeE.BainV.BernardaraP.NewingerO.BarbucM.BatemanA.et al. (2009). A compilation of data on European flash floods. J. Hydrol.367, 70–78. doi: 10.1016/j.jhydrol.2008.12.028

  • CrossRef
  • Google Scholar

• 25

  GonzalezF. C. G.CavacantiM. D. C. R.RibeiroW. N.MendonçaM. B.HaddadA. N. (2024). A systematic review on rainfall thresholds for landslides occurrence. Heliyon10:e23247. doi: 10.1016/j.heliyon.2023.e23247,

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 26

  HamiltonW. (1979). Tectonics of the Indonesian Region. Denver: U.S. Geological Survey.

  • Google Scholar

• 27

  HeckmannT.SchwanghartW. (2013). Geomorphic coupling and sediment connectivity in an alpine catchment - exploring sediment cascades using graph theory. Geomorphology182, 89–103. doi: 10.1016/j.geomorph.2012.10.033

  • CrossRef
  • Google Scholar

• 28

  HongY.AdlerR. F.HuffmanG. J.PierceH. (2010). “Applications of TRMM-based multi-satellite precipitation estimation for global runoff prediction: prototyping a global flood modeling system,” in Satellite Rainfall Applications for Surface Hydrology, eds. GebremichaelM.HossainF. (Dordrecht: Springer), 245–265.

  • Google Scholar

• 29

  IlhamniF.AzizahC.SatriawanH.NuraidaRoboS.Misnawatiet al. (2023). Mapping analysis for vulnerable areas and erosion rate of Laut Tawar Lake, Peusangan watershed. Pol. J. Environ. Stud.32, 3617–3626. doi: 10.15244/pjoes/163499

  • CrossRef
  • Google Scholar

• 30

  IndartoHidayahE. (2019). Preliminary assessment of morphometric and hydrological properties of six watersheds in the eastern part of East Java. Forum Geogr.33, 113–128. doi: 10.23917/forgeo.v33i1.7858

  • CrossRef
  • Google Scholar

• 31

  LabrièreN.LocatelliB.LaumonierY.FreyconV.BernouxM. (2015). Soil erosion in the humid tropics: a systematic quantitative review. Agric. Ecosyst. Environ.203, 127–139. doi: 10.1016/j.agee.2015.01.027

  • CrossRef
  • Google Scholar

• 32

  LiS.ChenL.YiB.YangB. (2026). Optimal timescale of antecedent precipitation index for flood peak prediction in data-scarce mountain river basins. J. Hydrol. Reg. Stud.66:103624. doi: 10.1016/j.ejrh.2026.103624

  • CrossRef
  • Google Scholar

• 33

  LiZ.LiangH.ChenS.LiX.LiY.WeiC. (2024). Performance assessment of satellite-based precipitation products in the 2023 summer extreme precipitation events over North China. Atmos.15:1315, 2–18. doi: 10.3390/atmos15111315

  • CrossRef
  • Google Scholar

• 34

  LiX.WeiY.LiF. (2021). Optimality of antecedent precipitation index and its application. J. Hydrol.595:126027. doi: 10.1016/j.jhydrol.2021.126027,

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 35

  LiuC.GuoL.YeL.ZhangS.ZhaoY.SongT. (2018). A review of advances in China’s flash flood early-warning system. Nat. Hazards92, 619–634. doi: 10.1007/s11069-018-3173-7

  • CrossRef
  • Google Scholar

• 36

  MaM.HeB.WanJ.JiaP.GuoX.GaoL.et al. (2018). Characterizing the flash flooding risks from 2011 to 2016 over China. Water10:704, 1–12. doi: 10.3390/w10060704

  • CrossRef
  • Google Scholar

• 37

  MacalaladR. V.BadillaR. A.CabreraO. C.BagtasaG. (2021). Hydrological response of the Pampanga River basin in the Philippines to intense tropical cyclone rainfall. J. Hydrometeorol.22, 781–794. doi: 10.1175/JHM-D-20-0184.1

  • CrossRef
  • Google Scholar

• 38

  MarchiL.BorgaM.PrecisoE.GaumeE. (2010). Characterisation of selected extreme flash floods in Europe and implications for flood risk management. J. Hydrol.394, 118–133. doi: 10.1016/j.jhydrol.2010.07.017

  • CrossRef
  • Google Scholar

• 39

  MashuriKarlinaSujonoJ. (2025). Assessment of satellite-based rainfall products for drought monitoring in the Siak watershed, Indonesia. Environ. Chall.19:101134. doi: 10.1016/j.envc.2025.101134

  • CrossRef
  • Google Scholar

• 40

  Muñoz-VillersL. E.McDonnellJ. J. (2013). Land use change effects on runoff generation in a humid tropical montane cloud forest region. Hydrol. Earth Syst. Sci.17, 3543–3560. doi: 10.5194/hess-17-3543-2013

  • CrossRef
  • Google Scholar

• 41

  NofitaC.AzizahC.SatriawanH.IriadiR.RoboS. (2025). Land-use changes and their impact on runoff coefficient and flood volume in the Keureuto watershed, Aceh Province. J. Watershed Manag. Res.9, 105–120. doi: 10.59465/jppdas.2025.9.1.105-120

  • CrossRef
  • Google Scholar

• 42

  NorbiatoD.BorgaM.EspostiS. D.GaumeE.AnquetinS. (2008). Flash flood warning based on rainfall thresholds and soil moisture conditions: an assessment for gauged and ungauged basins. J. Hydrol.362, 274–290. doi: 10.1016/j.jhydrol.2008.08.023

  • CrossRef
  • Google Scholar

• 43

  ObeidatM.AwawdehM.Al-HantouliF. (2021). Morphometric analysis and prioritisation of watersheds for flood risk management in Wadi Easal Basin (WEB), Jordan, using geospatial technologies. J. Flood Risk Manag.14:e12711. doi: 10.1111/jfr3.12711

  • CrossRef
  • Google Scholar

• 44

  OmayP. O.MuthamaN. J.OludheC.KinamaJ. M.ArtanG.AtheruZ. (2025). Evaluation of satellite-based rainfall estimates over the IGAD region of eastern Africa. Meteorog. Atmos. Phys.137:22. doi: 10.1007/s00703-025-01068-w

  • CrossRef
  • Google Scholar

• 45

  PennaD.MeerveldH. J. T.GobbiA.BorgaM.FontanaG. D. (2011). The influence of soil moisture on threshold runoff generation processes in an alpine headwater catchment. Hydrol. Earth Syst. Sci.15, 689–702. doi: 10.5194/hess-15-689-2011

  • CrossRef
  • Google Scholar

• 46

  Quichimbo-MiguitamaF.PesántezJ.CrespoP.Sánchez-MurilloR.LimH.MunksgaardN.et al. (2024). Key controls on water transit times along a tropical precipitation gradient. J. Hydrol.645:132134. doi: 10.1016/j.jhydrol.2024.132134,

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 47

  SalgadoM. J. H.NájeraJ. Z. (2022). Assessing flood early warning systems for flash floods. Front. Clim.4:787042. doi: 10.3389/fclim.2022.787042

  • CrossRef
  • Google Scholar

• 48

  SoulisK. X.ValiantzasJ. D. (2012). SCS-CN parameter determination using rainfall-runoff data in heterogeneous watersheds: the two-CN system approach. Hydrol. Earth Syst. Sci.16, 1001–1015. doi: 10.5194/hess-16-1001-2012

  • CrossRef
  • Google Scholar

• 49

  SpitalarM.GourleyJ. J.LutoffC.KirstetterP. E.BrillyM.CarrN. (2014). Analysis of flash flood parameters and human impacts in the US from 2006 to 2012. J. Hydrol.519, 863–870. doi: 10.1016/j.jhydrol.2014.07.004

  • CrossRef
  • Google Scholar

• 50

  StoffelM.WyżgaB.MarstonR. A. (2016). Floods in mountain environments: a synthesis. Geomorphology272, 1–9. doi: 10.1016/j.geomorph.2016.07.008

  • CrossRef
  • Google Scholar

• 51

  SyofyanE. R.HerdiantoR.ArlisA. P.YuwonoW.ThaariqA. (2025). Hydrological and hydraulic simulation of flooding in the Anai Valley: a case study of the 11 may 2024 event. Jurnal Ilmiah Rekayasa Sipil.22, 94–103.

  • Google Scholar

• 52

  ThohaA. S.SlametB.HarahapM. M.SariT. Y.HuluD. L. N. (2023). Spatial distribution of flood vulnerability in Langkat regency, north Sumatera Province, Indonesia. IOP Conf. Ser.: Earth Environ. Sci1188:12009. doi: 10.1088/1755-1315/1188/1/012009

  • CrossRef
  • Google Scholar

• 53

  WaskoC.NathanR. (2019). Influence of changes in rainfall and soil moisture on trends in flooding. J. Hydrol.575, 432–441. doi: 10.1016/j.jhydrol.2019.05.054

  • CrossRef
  • Google Scholar

• 54

  WohlE.BarrosA.BrunsellN.ChappellN. A.CoeM.GiambellucaT.et al. (2012). The hydrology of the humid tropics. Nat. Clim. Chang.2, 655–662. doi: 10.1038/nclimate1556

  • CrossRef
  • Google Scholar

• 55

  WuL.RenY.HuangP.YuanS.ZhouC.GuZ.et al. (2026). Evaluation of GSMaP and MSWEP precipitation products for runoff simulation in the Lhasa River basin. PLoS One21:e0342995. doi: 10.1371/journal.pone.0342995,

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 56

  YuniawanR. A.Rifa’iA.FarisF.SubiyantoroA.SatyaningsihR.HidayahA. N.et al. (2022). Revised rainfall threshold in the Indonesian landslide early warning system. Geosciences12:129, 2–17. doi: 10.3390/geosciences12030129

  • CrossRef
  • Google Scholar

• 57

  ZhangP.SunW.XiaoP.YaoW.LiuG. (2022). Driving factors of heavy rainfall causing flash floods in the middle reaches of the Yellow River: a case study in the Wuding River basin, China. Sustainability14:8004, 1–11. doi: 10.3390/su14138004

  • CrossRef
  • Google Scholar

• 58

  ZhaoG.XuZ.PangB.TuT.XuL.DuL. (2019). An enhanced inundation method for urban flood hazard mapping at the large catchment scale. J. Hydrol.571, 873–882. doi: 10.1016/j.jhydrol.2019.02.008

  • CrossRef
  • Google Scholar