1 Introduction

The construction of dams stands as a prominent solution for securing a sustainable water supply [1,2,3], efficiently meeting the escalating water requirements stemming from a burgeoning global population [4]. A dam serves as a tangible infrastructure that plays a pivotal role in water conservation and holds substantial economic importance at the national level [5,6,7]. Dams serve a multifaceted purpose encompassing electricity generation, flood mitigation, and irrigation management among their diverse applications [8, 9]. While dams offer substantial societal benefits, it is imperative to acknowledge that instances of dam failure have, at times, led to catastrophic flooding disasters [10,11,12,13], it brought about economic losses, loss of life, ecological changes, and left vulnerable people with dire consequences [14,15,16,17].

Earthen dams are widely recognized as one of the most prominent dam types due to their versatility in accommodating various foundation conditions, straightforward construction processes, and relatively economical cost [18, 19]. Nevertheless, clay dams, owing to their lower structural rigidity, exhibit an increased susceptibility to failure [20]. A catastrophic flood wave resulting from dam failure has the potential to adversely impact downstream activities, disrupt sediment transport processes, and induce significant morphological alterations [21,22,23,24].

A dam breach is characterized as either a partial or catastrophic failure, resulting in uncontrolled downstream flooding [25, 26]. When a dam breaches, it generates a flood wave of increased height and velocity, thereby exerting a significant impact on both downstream life and property [27, 28]. Despite the incorporation of sufficient safety margins in dam engineering, their failure can still occur due to various factors, such as leakage, overtopping, seepage, seismic events, and other natural disasters [29,30,31]. The recent dam failure in Libya serves as a poignant reminder of the catastrophic consequences that can result from inadequate dam maintenance and the impact of extreme weather events. In this tragic incident, two smaller dams, left unmaintained since 2002, succumbed to the forces of nature. The first dam's collapse unleashed a powerful torrent, further compounded by the elevation difference between the two dams, ultimately leading to the downfall of the second dam. Experts estimate that a staggering 30 million cubic meters of water were released, causing widespread devastation as the deluge met parched ground unable to absorb the sudden influx. This sobering event underscores the critical importance of dam infrastructure management and preparedness in the face of increasingly unpredictable climate patterns [32,33,34]. The necessity for dam break analysis arises from the inherent risk of dam failure and its far-reaching consequences. Such analyses play a crucial role in assisting decision-makers with land use planning and the formulation of emergency response strategies, ultimately aimed at mitigating the potentially catastrophic loss of life and property damage [35,36,37]. The seismic events of 1971, which led to the collapse of the Lower Van Norman Dam, and the 1964 incident involving the Baldwin Hills Dam in California served as pivotal moments prompting dam authorities to initiate the creation of flood maps for potential dam failures. Consequently, the development of investigation methods for assessing breach hydrographs became imperative. These events, alongside the numerous dam disasters that unfolded across the United States during the 1970s, served as compelling catalysts for a heightened focus on dam safety within the research community [37,38,39].

Based on an extensive analysis of more than 1,065 earthen dam failures, it has been determined that overtopping and piping are significant contributing factors to the failure of earthen dams. In cases of overtopping failure, potential vulnerabilities can manifest in spillways, downstream slopes, and foundations. Conversely, in instances of piping failure, the risk extends across the entire dam section [40,41,42,43].

Many dam failures begin with an initial structural breach, leading to the release of previously impounded water upstream, which subsequently generates flood waves downstream of the dam [10]. Residents residing downstream should be apprised of the potential for dam breaches or any associated expenditures required to enhance the area's safety or mitigate potential issues [44].

The fundamental objectives of dam break analysis encompass the prediction and effective routing of the outflow hydrograph resulting from a breach [37, 45,46,47]. In the study of dam breaks, the critical factors for examination are the extent of impact along the river course and the characteristics of the flood hydrograph. The utilization of flood predictions facilitates the development of emergency plans aimed at safeguarding both individuals and property from flood-related hazards [48,49,50,51]. In practice, hydraulic models are commonly employed for precise estimation of dam breach flood wave characteristics and their subsequent propagation within the downstream valley [30, 52, 53].

In their comprehensive analysis of the Hawaii Dam Break, a consortium comprising federal, state, and consulting engineers found that the outcomes yielded by both FLO-2D 6.0 and HEC-RAS 5.0.7 exhibited a noteworthy degree of comparability in relation to submerged area and water surface elevation [54, 55]. To ascertain critical parameters such as dam breach flood wave height, velocity, and length, Nistoran et al. [56] applied HEC-RAS modeling technology in their analysis of Romania's Bicaz Dam. Xiong [10] conducted an exhaustive dam break analysis for the Foster Joseph Sayers Dam in the United States, focusing on unstable flow scenarios and accounting for the potential impact of a maximum flood event. Kulkarni and Jagtap [57] harnessed the capabilities of HEC-RAS to conduct an extensive investigation into India's Pawana Dam, where they estimated key characteristics, including maximum flow rate, velocity, and stage as part of a study aimed at evaluating the potential consequences of dam failure scenarios. Abhijith et al. [5], in their examination of India's Idukki Dam, emphasizing the overtopping failure mode. It predicts dam break parameters, flood hydrograph, and peak flow, utilizing HEC-RAS 1D model. The results indicate a peak flow of 28,345.8 m3/s downstream and highlight the significant influence of Manning's roughness coefficient (N) on the analysis, providing valuable insights into flood extent and risk assessment. Furthermore, Chandrabose et al. [58] leveraged HEC-GeoRAS to extract river geometry data from the SRTM digital elevation model, thus facilitating the acquisition of data pertaining to stage, flow hydrograph, and inundation areas for the Malankara Dam in India. Numerous researchers have investigated both one-dimensional and two-dimensional hydraulic models, with HEC-RAS emerging as the predominant choice for dam break analysis due to its renowned accuracy in predictions and its accessibility as an open-source tool [59]. The accuracy of predictions in HEC-RAS is attributed to its incorporation of sub -rid technology, allowing for precise visualization of 2D flooding events[60]. Furthermore, researchers have opted for the HEC-RAS model in flood simulation due to its capacity to yield dependable results, even when confronted with intricate channel geometries and bed discontinuities, a situation that poses substantial challenges for more intricate and less stable hydraulic models [61,62,63,64,65].

The US Army Corps of Engineers (USACE) developed HEC-RAS to oversee and manage rivers, channels, harbors, and other public works. HEC-RAS software offers a wide range of simulation capabilities, including one-dimensional steady flow, one- and two-dimensional unsteady flow, sediment transport, bed calculations, and modeling for water temperature and water quality. This versatile software plays a crucial role in the efficient management of water resources and infrastructure [66,67,68]. This model is commonly employed for evaluating floodway encroachments in the context of floodplain management and investigations related to flood insurance [69, 70]. The user is required to make selections regarding the ultimate breach parameters and the time required for the breach formation. This is because most models do not instantaneously replicate the breach; rather, they initiate the process by inputting data into the hydraulic model and gradually simulate the development of the breach. Ultimately, the determination of breach parameters relies on several empirical equations that take into account specific dam and reservoir characteristics, such as dam height and storage volume [44]. In dam break modeling, there are two primary tasks: forecasting the dam breach outflow hydrograph and generating flood maps [54]. The prediction of the outflow hydrograph involves two main components: firstly, the estimation of breach parameters such as breach width, depth, and formation time; and secondly, the modeling of how the reservoir storage and inflow through the breach are routed downstream [10, 71]. Effective evacuation planning and the safe operation of reservoirs rely on precise forecasts of the flash flood hydrograph's dimensions, shape, and timing resulting from a dam failure [1, 72, 73]. The choice of selecting Mangla Dam stems from its global significance as the 7th largest earthen dam and its proximity to Jhelum city with a population of 1.2 million. This selection is further underscored by the critical need for flood hazard maps in case of dam failure, highlighting the importance of this study.

The primary objective of this study is to utilize the dam-break model and the hydrodynamic capabilities inherent in HEC-RAS, a widely recognized software tool for hydraulic analysis. This application aims to investigate the downstream flooding potential resulting from a hypothetical dam breach scenario involving the Mangla dam. Within this framework, the computed outflow hydrograph originating from the dam breach event serves as the foundational boundary condition for the flood routing model. This enables the accurate prediction of water level fluctuations and discharge rates at key locations along the river, including the Jhelum Bridge near Jhelum City, as well as downstream points of interest. This hydraulic analysis extends across a defined reach spanning 64 km, encompassing the area between the Mangla Dam and the Rasul Barrage. Additionally, an essential component of this research involves the execution of a sensitivity analysis, exploring a range of parameters within the HEC-RAS module. This analysis considers diverse modes and configurations associated with dam failure and breaching scenarios.

2 Materials and methodology

2.1 Geographic location of investigation

Mangla Dam, also referred to as the “Mangla Reservoir,” is located in the northeastern region of Pakistan, amidst the imposing Western Himalayan Mountains. Its vast watershed encompasses a latitude range from approximately 32°59′59.7″ N to 35°10′51.4″ N and a longitude span from 73°7′38.2″ E to 75°35′29.4″ E. Constructed in 1967 along the banks of the Jhelum River, Mangla Dam stands as a cornerstone of Pakistan's water infrastructure, serving critical roles in electricity generation and irrigation. Impressively, the dam possesses an electricity generation capacity of approximately 1000 MW [74], facilitated by its four power generation units. Notably, ongoing construction efforts are dedicated to expanding the reservoir's capacity, with plans to increase the current water storage volume of 5846.71 million cubic meters by an additional 3577.10 million cubic meters through elevation adjustments. The primary water sources for Mangla Dam originate from the Jhelum River and its tributaries, which include the Neelum, Kunhar, and Poonch rivers. This stored water plays a vital role in sustaining agricultural activities, providing essential irrigation support to rice and wheat cultivation areas during their respective seasons. Moreover, Mangla Dam holds the distinction of being the second-largest water reservoir in Pakistan, boasting an impressive capacity to retain 7.39 million acre-feet (MAF) of water. Its physical dimensions include a dam length of 3140 m (approximately 10,303 feet) and an imposing height of 147 m (about 482 feet) [75]. Climatically, the Mangla Watershed experiences diverse temperature variations, with average maximum temperature ranging from 10.6 to 31.6 °C and average minimum temperatures fluctuating between 4 and 25 °C. The Mangla Watershed's vast expanse spans about 33,490 km2 and has witnessed significant historical floods, notably in 1928, 1929, 1958, and 1959. The flood of 1929 stands out with a peak discharge of 1,040,000 cusecs. Drawing from this historical data and related meteorological information, a probable maximum flood (PMF) of staggering proportions, totaling 2,600,000 cusecs, was computed [76]. Consequently, the authors of this study endeavor to analyze the situation during a dam breach through comprehensive dam break analysis, offering valuable insights and guidance for authorities responsible for flood protection in the region, considering the critical role played by Mangla Dam in water storage and flood mitigation. The geographic location of the study area is shown below in Fig. 1 with details.

Fig. 1
figure 1

Location of study area on the map

Full size image

The primary focus of this research encompasses two critical stages: firstly, the estimation of flood hydrographs resulting from dam break scenarios at the dam site, and secondly, the comprehensive analysis of these outcomes. Subsequently, the research proceeds to generate floodplain maps that account for various dam failure scenarios and facilitate the routing of dam breach flood hydrographs to downstream monitoring stations. To accomplish these objectives, the hydraulic modeling software HEC-RAS 6.4.1 is employed. This tool is instrumental in simulating dam break events and their hydraulic consequences. In parallel, HEC-GeoRAS is utilized to generate essential river geometric data and create detailed floodplain maps. The hydraulic modeling conducted in HEC-RAS adheres to the principles of St. Venant's equations, specifically tailored for analyzing unstable flow scenarios. This approach ensures the accuracy and reliability of the flood routing methods employed, thereby enhancing the credibility of the research findings from both a scientific and practical perspective [26, 54, 62, 77, 78].

Below are the fundamental one-dimensional (1-D) Saint–Venant equations, which encapsulate the principles of mass and momentum conservation in unsteady flow scenarios:

$$\frac{\partial \mathrm{A}}{\partial \mathrm{x}}+ \frac{\partial \mathrm{Q}}{\partial \mathrm{x}}=0$$
(1)
$$\frac{\partial \mathrm{Q}}{\partial \mathrm{t}}+ \frac{\partial }{\partial \mathrm{x}} \left(\frac{{\mathrm{Q}}^{2}}{\mathrm{A}}\right)+\mathrm{gA }\left(\frac{\partial \mathrm{z}}{\partial \mathrm{x}}+{\mathrm{S}}_{\mathrm{f}}- {\mathrm{S}}_{\mathrm{e}} \right)=0$$
(2)

Equation (1) represents mass conservation, expressing the continuity equation, where ∂A/∂x signifies the rate of change of flow area with respect to distance, and ∂Q/∂x denotes the rate of change of discharge with respect to distance. Equation (2) represents momentum conservation, incorporating parameters such as ∂Q/∂t, which signifies the rate of change of discharge with respect to time, Q2/A representing the specific kinetic energy of the flow, g representing the acceleration due to gravity, ∂z/∂x indicating the rate of change of elevation with respect to distance, and Sf and Se representing the friction and expansion-contraction slopes, respectively.

Figure 2 demonstrates the techniques employed in conducting dam break analysis, while the subsequent sections will outline the steps involved in hydraulic modeling, model simulation, and floodplain mapping.

Fig. 2
figure 2

Methodological framework for the current study

Full size image

2.2 Geometric data and spatial information

The characterization of river geometry serves as an essential prerequisite in hydraulic modeling. Gathering river geometry information necessitates either field surveys or the utilization of remote sensing data extraction techniques. Traditional survey methods, although labor-intensive and time-consuming especially for extensive areas, yield precise river geometry data [79]. Conversely, there is a growing trend in adopting remote sensing data for hydraulic modeling due to its efficiency in terms of time and cost savings [80,81,82]. To obtain the river geometries essential for the investigation, the Digital Elevation Model (DEM) featuring a 12.5-m cell size was downloaded from the Alaska Satellite Facility, specifically the ALOS PALSAR DEM [83]. Notably, the ALOS PALSAR DEM is highly regarded for its high-resolution terrain-corrected data, which significantly enhances the precision and reliability of the river geometry extraction process. The next step involved converting this acquired data into Triangular Irregular Network (TIN) format for obtaining geometric data from it in HEC-RAS and HEC-GeoRAS.

HEC-RAS provides two distinct reservoir modeling options. Firstly, the reservoir can be treated as a storage entity with an elevation-volume relationship, facilitating level pool routing. The second approach involves utilizing 1D cross-sections within the reservoir pool, enabling dynamic flood routing. Notably, it is observed that reservoirs characterized by long and narrow profiles tend to exhibit more significant water surface slope upstream of the dam, as documented by the US Army Corps of Engineers (USACE) in 2016 [84]. To accurately predict pool elevations and outflow for such long, narrow reservoirs, the most suitable modeling technique involves full dynamic wave (unsteady flow) routing. In this research, the comprehensive dynamic wave routing methodology within HEC-RAS was employed. For the current study, the reservoir was represented using 2D storage area, and an elevation-volume relationship was established for the storage area. The bathymetric data for the storage reservoir was obtained from the relevant department, the Water and Power Development Authority (WAPDA) in Pakistan [85]. On the downstream side of the storage region, a 2D flood routing area was designated. The connection between these two areas was established using the SA/2D area connector option available in the geometry editor. Figure 3 provides a visual representation of the spatial extent of the 2D geometry within this study which covers an extent of 65 km downstream up to the point of Raul barrage.

Fig. 3
figure 3

2D view of study area in HEC-RAS geometry editor

Full size image

2.3 Integration of land use land cover (LULC) data with geometric parameters

In the context of this comprehensive study, the LULC map was obtained from ESRI's Land Cover website [86], featuring a spatial resolution of 10 m per cell. This dataset played a pivotal role as it provided crucial insights into the land cover types within the research area. Each land cover category was linked to a Manning's n value, representing the surface roughness of the terrain. To ensure accurate flood routing downstream of the dam, it became imperative to integrate the LULC data with the established geometric data. HEC-RAS 6.4.1 facilitated this process by offering the “Manage Geometry Association” option within the RAS Mapper tool. This feature allowed users to seamlessly associate the geometry with LULC data for the specific study area. Within the spatial extent of the 2D geometry, encompassing both the upstream storage area and downstream 2D mesh area, eight distinct land cover types were identified.

Table.1below provides a comprehensive overview of each land cover category and its corresponding Manning's n value. The integration of land cover data with the geometric parameters significantly enhances the precision and reliability of flood routing simulations, ensuring a comprehensive representation of hydraulic processes within the study area [87]. This integration of land cover data with the geometry can be observed in Fig. 4 below.

Table 1 Land cover types with corresponding roughness coefficients [88]
Full size table
Fig. 4
figure 4

Land cover integration with geometry

Full size image

2.4 Dam hydraulic modeling

In the context of dam modeling, HEC-RAS is employed to represent a dam structure as an inline construction featuring a minimum of two predefined upstream cross-sections. This entails the comprehensive specification of the dam's geometrical attributes, material characteristics, crest dimensions, top and bottom evaporation rates, and various other relevant parameters. The ultimate objective is to enable the simulation of the dam's behavior within the HEC-RAS framework. The provided Fig. 5 illustrates the cross-sectional profile of the Mangla Dam along the length of its crest, which serves as a pivotal reference in the present study. To facilitate a robust representation of reservoir storage, an elevation-volume curve mechanism is employed for the storage reservoir. This approach, in turn, facilitates the implementation of full dynamic wave routing, as it accommodates the creation of a sloping water surface upstream of the dam in the event of a failure.

Fig. 5
figure 5

Mangla dam cross-section along its crest length

Full size image

2.5 Dam breach data

The analysis of dam breaches necessitates consideration of various critical factors, which collectively contribute to a comprehensive understanding of the potential risks involved. These variables include the ultimate breadth of the breach at its base, the final elevation of the breach bottom, the breach's side slope characteristics, the exact timing of the creation of the breach, and the starting reservoir elevation that corresponds to the start of the breach event. These elements serve as the basic building blocks of a dam break analysis. HEC-RAS provides a useful set of tools and approaches for dam breach assessments. Five different regression equations, each drawn from the vast body of hydraulic engineering research, are among these resources. These equations, which have been put forth by MacDonald et al. (1984), Froehlich (1995), Froehlich (2008), Von Thun and Gillete (1990), and Xu and Zang (2009), are extremely helpful computational tools for determining crucial breach parameters [89]. These parameters include the breach bottom width, side slope dimensions, and the duration of breach development. To harness the predictive power of these equations, HEC-RAS breach parameter calculator facilitates engineers and researchers by requiring specific data entry, including the top of dam elevation, breach bottom elevation, pool volume at failure, pool elevation at failure, and the failure mode of the dam. These parameters are uniform across all the empirical methods employed by HEC-RAS software, streamlining the input process for accurate predictions. This process results in the calculation of critical breach parameters, which are essential for modeling and assessing the potential consequences of dam breaches. It is important to emphasize that the choice of regression equation is a pivotal step in the analysis process. While any of the mentioned equations can be utilized, careful consideration must be given to the suitability of the selected equation for the dam breach scenario under investigation. In practice, a thoughtful approach involves selecting data from the regression equations that align most closely with the conditions and characteristics of the dam in question [90]. In the current research work, the breach parameters devised by Froehlich (2008) were utilized based on their suitability and precision in estimating breach parameters for both pipe and overtopping failure modes [89, 91] Dam breach parameters for both piping and overtopping failure are shown in Table 2 below.

Table 2 Parameters for Dam breach piping and overtopping failure
Full size table

2.6 Boundary conditions and flow data

Boundary conditions and flow data are pivotal elements in the context of dam breach analysis within HEC-RAS software. These factors exert a significant influence on the dynamics of downstream floodwaters and play a critical role in accurately simulating the consequences of a dam breach event. When conducting unsteady flow analysis, it is imperative to consider various boundary conditions that mirror the actual characteristics of the site. These boundary conditions encompass a range of factors, each of which contributes uniquely to the accuracy of the simulation [92]. In the context of unsteady flow analysis within HEC-RAS, the provision of both upstream and downstream boundary conditions is imperative for the model’s successful execution [93]. Within the framework of this research, the establishment of upstream boundary conditions centered on the definition of an inflow hydrograph for the storage area. This particular hydrograph serves as a representation of the most historically significant inflow event contributing to the Mangla reservoir, specifically identified as the flood wave of 1992. It was realized as biggest flood event that the Mangla dam has ever experienced [76]. Flooding can have a critical impact on dams. Inadequate spillway management during a flooding event can lead to dam failure through piping or overtopping, underscoring the paramount importance of effective flood mitigation and emergency response strategies for dam safety [94]. The dataset corresponding to the 1992 Probable Maximum Flood, sourced from WAPDA, Pakistan, was meticulously incorporated into the model. This integration involved the insertion of data at hourly intervals, commencing on September 9, 1992, at 00:00 h, and concluding on September 13, 1992, at 24:00 h. The resultant inflow hydrograph for the 1992 flood event is visually presented in the accompanying Fig. 6. The peak inflow discharge is observed during the initial 48 h, as depicted in Fig. 6. Consequently, our simulation concentrates on this time interval to generate the output flood and stage hydrographs, which will be detailed in the results section. Downstream boundary conditions encompass the upstream recorded daily water level immediately preceding the Rasul Barrage, located 65 km downstream from the Mangla Dam. This recorded water level effectively serves as the stage-discharge rating curve for the Rasul Barrage obtained from WAPDA, Pakistan [85]. The rating curve of Rasul barrage is served as downstream boundary condition for this study which is a standard approach in hydraulic modeling. While it simplifies the simulation, this method is widely accepted and accurate for practical purposes [95]. The use of a steady rating curve in our simulations is a common and suitable representation for modeling downstream conditions in HEC-RAS, and it is consistent with standard practices in hydraulic modeling.

Fig. 6
figure 6

1992 Flood inflow hydrograph Mangla Dam

Full size image

2.7 Mapping flood plains in the context of dam breach analysis

For generating inundation maps and delineating areas vulnerable to potential flooding, a pivotal step involves the integration of simulation data from the HEC-RAS software into HEC-GeoRAS, a geospatial extension specifically designed to enhance the geospatial capabilities of HEC-RAS [96]. This integration process is instrumental in producing a comprehensive representation of the flood dynamics. The crux of this integration lies in the amalgamation of georeferenced cross sections with associated water surface elevations, forming a continuous water surface representation [97]. This continuous water surface becomes a critical element in the analysis as it accurately depicts the spatial extent of water levels throughout the area of interest. Subsequently, a meticulous comparison is undertaken between the water surface, as represented by the simulation data, and the existing landscape features. This comparative analysis yields invaluable insights into the potential interactions between floodwaters and the surrounding terrain, thereby aiding in the identification of flood-prone regions [98]. The final step in this process involves the creation of floodplain maps that are directly linked to the extent of the flood event under consideration. These maps offer both a quantitative and visual depiction of the regions susceptible to inundation across a range of flood event scenarios. It's worth emphasizing that this approach holds the capability to assess the consequences of diverse dam failure scenarios comprehensively, thereby facilitating a comprehensive evaluation of the affected areas.

3 Results and discussions

3.1 River flow pattern and water level variation

In the event of a full reservoir, unsteady flow simulations were conducted to assess potential dam failures, focusing on both piping and overtopping failure scenarios. The analysis in this study spanned a 65-km stretch downstream of the dam. The simulations, using the HEC-RAS model with Probable Maximum Flood (PMF) data, commenced on September 9, 1992, at 0:00 and concluded precisely on September 13, 1992, at 24:00. The output results are consistent with the chosen simulation period. Figure 7a, b provide a comprehensive visual representation for both pipping and overtopping failures respectively. Both figures detail the stage hydrographs, headwater and tailwater stages. These insightful data points were collected at a critical location, which is the immediate downstream side of the dam. The analysis also involved an in-depth examination of essential parameters, the summary of which can be found in Table 3. These parameters include the maximum headwater stage, tailwater stage, peak discharge, and arrival time for both piping and overtopping failure scenarios. Notably, it can be observed from that the overtopping failure scenario exhibited the highest values for headwater stage, tailwater stage, and peak discharge within the breach outflow hydrograph. Additionally, the occurrence time for the overtopping failure was notably earlier compared to the piping failure scenario.

Fig. 7
figure 7

Hydrograph depicting water level and flow near dam for: a piping failure; b overtopping failure

Full size image
Table 3 Hydraulic Information for Jhelum Bridge: Water Level and Flow
Full size table

3.2 Maps depicting flood inundation

In the context of comprehensive floodplain mapping and hydraulic analysis, the research employed HEC-GeoRAS and HEC-RAS tools to meticulously investigate the downstream implications of potential piping and overtopping failures of Mangla dam. The inundation maps, graphically depicted in Fig. 8a, b for piping and overtopping failure respectively, facilitate a direct comparison with the pre-inundation conditions of the downstream area, as illustrated in Fig. 9. This comparison of pre-inundation and post-inundation maps provides a clear depiction of the submerged region downstream of the dam. This study discerned that, in the event of a piping failure, a substantial downstream area encompassing 379 km2 is susceptible to inundation. Conversely, in the case of overtopping failure, the inundation area expands to 394 km2, extending up to 65 km downstream from the dam's location. Moreover, the analysis revealed noteworthy variations in the maximum depth of flow. For the piping failure scenario, this depth ranged from 40.95 m immediately downstream of the dam to 14.5 m at the farthest point along the river reach which is located 65 km downstream at Rasul Barrage. In contrast, the overtopping failure scenario exhibited a depth variation ranging from 41.94 to 18.62 m. The inundated area due to overtopping failure is approximately 15 km2 greater than that of piping failure as shown in Table 4. This difference in flooded area can be attributed to the nature of the failure scenarios. During overtopping failure, the dam erodes relatively at a faster rate, leading to a rapid release of a large volume of water with an early peak arrival time. This results in a broader inundation area This research underscores that the downstream zone faces a significantly higher level of risk in the overtopping failure scenario, attributable to both the extensive inundation area and the substantial maximum depth of flow. Nevertheless, it is imperative to note that through the implementation of correct and effective reservoir operation strategies, the potential for overtopping failure can be mitigated and ultimately prevented.

Fig. 8
figure 8

Inundation extent downstream for; a piping failure; b overtopping failure

Full size image
Fig. 9
figure 9

Downstream of dam before inundation

Full size image
Table 4 Extent of inundation for different land cover types in piping and overtopping failure scenarios
Full size table

3.3 Integration of inundation maps with land use land cover

In the context of this research, the integration of inundation maps with land use and land cover (LULC) data has been instrumental in unveiling critical insights into the potential consequences of dam breaches [99]. The process involved the spatial clipping of the LULC raster layer, guided by the extent of the inundation maps in ArcGIS 10.8 [100]. This integration has illuminated the intricate relationship between land cover patterns and flood inundation scenarios. Through the spatial analysis, the attributes of the clipped LULC layer revealed seven distinct land cover types: water bodies, trees, flood plains, crops, built areas, bare lands, and rangelands that can be observed in Fig 10 for the inundated area up to 65 km reach of river downstream. Remarkably, each of these land cover types is susceptible to inundation in the event of a Mangla dam breach. Table 4presents the quantified extent of inundation for each land cover type, differentiating between piping and overtopping failure scenarios. For the piping failure scenario, the findings indicate that 56 km2 of built areas and 217 km2 of cropland are subject to complete inundation. Similarly, in the case of overtopping failure with relatively high tolls, 59 km2 of built areas and 220 km2s of cropland face complete inundation. Notably, the substantial inundation of built areas underscores the potential for significant human losses, particularly in the absence of an effective warning system. It is worth highlighting that within the inundated built area lies Jhelum city, with a population of 1.2 million, as per the 2017 census records [101]. These figures, when juxtaposed with the inundation data, signify the magnitude of potential economic losses, and underscore the critical need for robust preparedness and mitigation strategies in the face of such hydrological risks.

Fig. 10
figure 10

Clipped raster representation of LULC data within the extent of inundation maps

Full size image

3.4 River station flow dynamics downstream

Within this analysis, Fig 11a, b present the flood hydrographs captured at distinct river locations downstream of the dam for piping and overtopping failures, respectively. These river stations are situated at intervals of 20 km, 35 km, 50 km, and 65 km from the dam. The observed data vividly illustrates the notable dynamics of peak flow magnitudes and their temporal evolution within the context of piping and overtopping failures. The salient findings underscore that, immediately downstream of the dam, the peak flow magnitude is significantly pronounced, gradually diminishing as one progresses downstream. This attenuation in the flood wave is attributed to natural hydrological processes. The detailed specifics of peak flow rates and the corresponding timing of their arrival at various river locations are meticulously documented in Table 5. For instances of piping failure, the recorded peak flow rates are 174,850 m3/s immediately downstream of the dam and 149,845 m3/s at the terminal point of the river reach. These peak flow events manifest with respective arrival times of 8.50 h and 14.16 h from the initiation of breach formation. In contrast, for overtopping failure scenarios, the peak flow rates are notably higher, registering at 177,850 m3/s near the dam and 154,171 m3/s at the river's farthest extent. Correspondingly, the arrival times for these peak flows are 7.98 h and 12.97 h from the onset of breach formation. A discernible trend emerges from these observations, indicating that as the distance from the dam increases, the peak flow within the downstream zone undergoes gradual attenuation. Moreover, it is noteworthy that peak flow values consistently exhibit higher magnitudes in overtopping failure scenarios when compared to piping failures at all designated river stations. These insights provide valuable data for understanding and managing the dynamics of flood events and their potential impact on downstream areas.

Fig. 11
figure 11

Flood hydrographs at downstream river stations for: b piping failure; b overtopping failure

Full size image
Table 5 Peak flow rates and arrival times at various river stations for piping and overtopping failures
Full size table

3.5 Water levels at different river locations

In the evaluation of both piping and overtopping failure scenarios, a comprehensive analysis of stage hydrographs has been conducted at two critical locations: the dam site and downstream river locations situated at distances of 20 km, 35 km, 50 km, and 60 km. These hydrographs are thoughtfully illustrated in Fig 12a, b, respectively. Upon meticulous examination of the data, a discernible trend emerges: the highest stage levels exhibit a notable elevation immediately downstream of the dam, steadily receding as one progresses further downstream. This hydraulic phenomenon underscores the dynamic nature of flood propagation. To provide a quantitative perspective, Table 6 documents the highest flood depths and their corresponding arrival times at various river locations. For the piping failure scenario, the maximum flood depth attains 14.50 m at the terminal point of the river reach under scrutiny, while it peaks at 40.95 m directly downstream of the dam. Importantly, the maximum stage is anticipated to reach the dam downstream and the terminus of the river reach, arriving at 8.45 h and 13.98 h, respectively, following the inception of breach formation. In the case of overtopping failure, a maximum flood depth of 18.62 m is noted at the selected end of the river reach, with an immediate downstream peak of 41.98 m. Remarkably, the highest stage is predicted to reach the dam downstream and the end of the river reach, manifesting at 7.89 h and 12.87 h, respectively, post-commencement of breach formation. These empirical findings robustly emphasize that as distance from the dam increases, the pinnacle stage within the downstream region undergoes a consistent attenuation. Furthermore, it is noteworthy that, at each station along the river, peak stage values for overtopping failure consistently surpass those observed in the context of piping failure. These insights provide critical data for understanding the evolving dynamics of flood scenarios and the inherent implications for downstream regions.

Fig. 12
figure 12

Stage hydrographs at downstream river locations for: a piping failure; b overtopping failure

Full size image
Table 6 Highest flood depths, and arrival times at various river locations for piping and overtopping failures
Full size table

3.6 Model’s calibration and validation

The calibration of a hydraulic model hinges primarily on a single critical variable: channel resistance, often referred to as Manning's roughness coefficient or simply “n”. In the context of flood forecasting, the challenge arises from the scarcity of data from past floods of comparable magnitude, which is typically required for calibration. In such circumstances, a conventional and prudent approach involves conducting a model sensitivity analysis by varying the Manning's roughness coefficient across a range of values. The objective of this sensitivity analysis is to gauge the model's responsiveness to changes in this parameter. If the model exhibits minimal sensitivity or only minor variations in response to adjustments in Manning’s ‘n’, it engenders greater confidence in the accuracy and reliability of the model outputs. The breach outflow hydrograph, representing both piping and overtopping failure scenarios, are presented respectively in Fig. 13a, b for various Manning’s ‘n’ values. This figure unequivocally demonstrates that when Manning's roughness coefficient is subject to alteration, the resulting variations within the flow hydrograph are exceedingly limited. These subtle fluctuations, or lack thereof, underscore the model's robustness and its ability to yield consistent and dependable results across varying Manning’s ‘n’ values [63, 64, 102]. In essence, this analysis provides valuable insights into the model's calibration process. It not only validates the model's efficacy in the absence of historical flood data but also underscores its capacity to deliver accurate forecasts, even in scenarios where Manning’s roughness coefficient may be subject to uncertainty. Through this calibration approach, the model's reliability is bolstered, enhancing its utility in flood forecasting and management applications.

Fig. 13
figure 13

Breach outflow hydrograph for varying manning's 'n' values: a piping failure; b overtopping failure

Full size image

3.7 Evaluation of empirical methods

For evaluating the potential risks associated with the piping failure mode, a simulation of the HEC-RAS model has been conducted. This simulation involves the utilization of breach parameters that have been determined from five distinct empirical methods built in HEC-RAS. The dam's spillway effectively discharges the Probable Maximum Flood (PMF). From 2004 to 2009, Mangla Dam’s elevation increased from 1202 ft above sea level (msl) to 1242 ft msl, primarily to enhance storage capacity and flood wave management [103]. The spillway of the dam is now capable of efficiently accommodating the Probable Maximum Flood (PMF) [104]. Consequently, breach parameters are simulated specifically for piping failure, given the relatively reduced likelihood of overtopping. The breach parameters found in Table 7 were derived employing empirical techniques integrated into the HEC-RAS software. The results of this analysis, particularly the maximum stage and peak flow values, were determined by applying empirical formulas at various points downstream of the dam and are visually represented in Fig 14 and Fig 15 respectively. Among the diverse equations employed within the scope of this investigation, it is evident that the empirical equation put forth by MacDonald et al. (1984) produces the highest stage and peak flow values. This phenomenon can be attributed to the equation’s estimation of a broader breach width and a relatively shorter duration required for breach formation. In contrast, the equation developed by Xu and Zhang (2009) yields the lowest values for both stage and peak flow. This is primarily attributable to their equation's prediction of a narrower breach width and a more protracted period of time for the formation of the breach. From these findings, it can be inferred that breach width directly impacts peak flow and stage values. Specifically, a greater breach width corresponds to higher simulated values, while a lesser breach width corresponds to lower values. Conversely, the duration necessary for breach formation indirectly influences these values, with longer breach formation times leading to diminished simulated peak flow and stage values.

Table 7 Breach parameters for empirical methods
Full size table
Fig. 14
figure 14

Empirical methods performance across maximum stages

Full size image
Fig. 15
figure 15

Empirical methods performance across maximum discharges

Full size image

3.8 Sensitivity analysis of dam breach parameters

Sensitivity analysis is a valuable technique employed to assess how changes in independent variables impact a specific dependent variable while adhering to established assumptions. It aims to quantify the influence of various factors on a given model or outcomes. In the context of this study, sensitivity analysis was conducted to gain insights into how breach parameters affect both peak flow and maximum stage. The process involved systematic adjustments of each breach parameter, sourced from Froehlich’s (2008) equations, using a consistent 25% step difference, with all other parameters held constant [10, 11, 105]. The ensuing figures vividly depict the relationship between changes in breach parameters and the resulting fluctuations in maximum stage and peak flow. From Fig. 16a, it is apparent that breach width directly impacts both peak flow and maximum stage values. As the breach width increases, so do the values of both peak flow and maximum stage. The steep incline of these curves underscores the rapid rise in magnitudes. Figure 16b reveals a more modest change in peak flow and maximum stage values as the side slope of the breach is altered. Conversely, Fig. 16c demonstrates an inverse relationship for maximum stage and peak flow values against breach formation time. An Increase in the breach formation time leads to a decrease in both maximum stage and peak flow values. While the highest stage experiences a slight reduction, the decline in peak flow is more pronounced. Figure 16d highlights the precise connection between peak flow and maximum stage values and changes in breach depth. The steepness of these lines emphasizes the significant influence of breach depth alterations on the magnitudes of peak flow and maximum stage. Upon careful observations, it is noted that peak flow is highly sensitive to variations in breach depth, width, and formation time, while displaying relatively lower sensitivity to changes in side-slopes. Conversely, the maximum stage exhibits pronounced sensitivity to variations in breach depth and width, with lower sensitivity to changes in formation time and side-slopes.

Fig. 16
figure 16

Peak flow rate and maximum stage variation for: a breach width; b breach side-slopes; c breach formation time; d breach depth

Full size image

4 Conclusions

Dams are inherently complex structures, and their failure constitutes a catastrophic event, the intricacies of which remain not entirely elucidated. The application of the HEC-RAS 2D model forms the backbone of the ongoing investigation centered around the Mangla dam in Azad Kashmir, Pakistan, with the primary objective of comprehending dam breaches. The dataset of flow data chosen for this endeavor involves the largest Probable Maximum Flood (PMF) ever recorded at the Mangla reservoir, dating back to 1992, serving as the foundation for simulating unsteady flow conditions essential for gaining insights into the dynamics of dam breaches. HEC-RAS, as a hydraulic modeling tool, serves a dual purpose: it facilitates the modeling of dam break scenarios and the subsequent routing of breach-induced floodwaters downstream. Moreover, it allows for the assessment of hydraulic conditions at precise locations along the river’s course. By exporting the results from HEC-RAS to HEC-GeoRAS, inundation maps were generated. These maps are instrumental in delineating submerged areas and discerning the various land cover categories affected by dam breaches, including both piping and overtopping failures. The immediate downstream segment experiences peak flows of 174,850 m3/s for piping failure and 177,850 m3/s for overtopping failure, with inundation areas spanning 379 km2 and 394 km2, respectively. These numbers suggest that overtopping failure, characterized by the larger peak flow, assumes a position of critical importance. It's noteworthy, however, that the Mangla Dam Raising Project, executed between 2004 and 2009, has substantially reduced the likelihood of overtopping failure. In the context of assessing the sensitivity of breach parameters to peak flow and maximum stage, a dedicated sensitivity analysis has been executed, utilizing hydraulic outputs stemming from piping failure as a basis. The outcomes of this analysis underscore the high sensitivity of maximum stage to breach depth and width, with relatively lower sensitivity to breach formation time and side slopes. In contrast, peak flow demonstrates high sensitivity to breach depth and width but comparatively lower sensitivity to breach side slopes. In terms of inundation, the analysis reveals that in the event of piping failure, 56 km2 of urban areas and 217 km2 of agricultural land are subject to complete submersion. Similarly, overtopping failure results in 59 km2 of urban areas and 220 km2 of farmland being inundated. These substantial figures shed light on the severity of the potential consequences. Dam break analysis is crucial for policymakers as it provides insights into flood dynamics, including flood wave arrival times at downstream locations, aiding effective evacuation planning. Additionally, flood hazard maps identify vulnerable areas, allowing prioritized response efforts. In cases of multiple failure scenarios, this analysis informs informed policy decisions, enhancing preparedness and resilience.