Introduction

The Indian Himalayan topography is prone to flash flood-induced soil erosion processes that are triggered due to extreme rainfall incidents. These incidents result in landslides and mudflows, which further cause infrastructure failures (Pandey and Mishra 2021). Moreover, due to anthropogenic activities and climate change, the river basins in the Himalayas are becoming increasingly susceptible to soil erosions (Budakoti et al. 2019; Thakur et al. 2020). Thus, collective development and management of land and water resources are needed to address extreme watershed conditions in this region, viz. excessive runoff, enhanced soil erosion, inadequate crop yield, and poor infiltration (Bhattacharyya et al. 2015; Choudhari et al. 2018; Sofi et al. 2021). Hence, it is crucial for decision-makers to comprehend the properties of the watershed and the associated hydrological processes, which can be investigated adequately by morphometric analysis.

Morphometric analysis implies a quantitative assessment and evaluation of the earth’s surface’s size, shape, and features (Agarwal 1998; Strahler 1964). The relief, areal, and linear morphometric characteristics of a drainage basin provide significant criteria for interpreting numerical analyses of drainage networks (Meshram et al. 2019; Meshram and Sharma 2017; Patel and Srivastava 2013). Several parameters, including lithology structure, geomorphic setup, soil characteristics, land use and land cover (LULC), and slope, contribute to the acceleration of soil loss in various climatic regions (Bhat et al. 2022; Costache et al. 2020; Singh and Pandey 2021b). A broad interpretation of the hydrologic response, including surface runoff, groundwater potential, and infiltration capacity, can be derived from morphometric analysis (Bhat et al. 2022). In data-scarce ungauged watersheds with limited geomorphological, soil, hydrological, and geological information, morphometric analysis offers improved insight and accuracy for predicting basin characteristics like travel-time, intensity, and peak time of erosional processes (Meraj et al. 2015).

Using morphometric and statistical characteristics of basins, researchers have estimated flood peaks by incorporating the concept of GIUH, particularly for the ungauged basin. This approach posits that excess rainfall follows various probabilistic flow paths through the channel and overland regions before reaching the catchment outlet (Gupta et al. 1980; Rigon et al. 2016; Rodríguez-Iturbe et al. 1979). In the study by Kumar et al. (2007), the Nash and Clark IUH method was applied in the ungauged Ajay catchment, India, to simulate the direct runoff hydrograph for the ten rainfall-runoff events. They found that both the Nash IUH and the Clark IUH model options from the HEC-1 package effectively estimated the hydrographs. Cudennec et al. (2004) explored the geomorphological dimensions of the unit hydrograph concept and found that integrating geomorphological parameters contributed to a better understanding of both unit hydrograph and geomorphologic unit hydrograph theories. Bamufleh et al. (2020) utilized the self-similarity method to compute equivalent Horton-Strahler ratios in a semi-arid area for GIUH hydrograph modeling. Their findings indicated that the Nash model was superior to the Fréchet model in accuracy. In most cases, the GIUH proved a reliable method for estimating flood response, especially in regions with undulating topography (Sahoo and Jain 2018). It effectively predicts and mitigates extreme precipitation events’ impact on the hilly watersheds’ hydrological response (Dimri et al. 2016; Pandey and Mishra 2021; Singh and Pandey 2021a).

The hydrological response of the basin is further affected by anthropogenic activities like deforestation, mining, and built-up construction, as they alter the natural landscape and its associated processes. In this regard, prioritizing watersheds is crucial in managing watersheds in terms of development programs, project cost, and type. To prioritize sub-basin, morphometric factors and land use-land cover information of watersheds are useful in the absence of extensive hydrological data, especially in an ungauged basin. Utilizing satellite-derived terrain information, like a digital elevation model (DEM), enables the computation of morphometric characteristics for a given watershed. Unlike the contours on topographic maps, which are discrete, digital elevation models can be seamlessly incorporated into a geographic information system (GIS) (Das et al. 2018; Horton et al. 2011; Moore and Burch 1986). Moreover, it may aid in evaluating distinct drainage basins from diverse climatic and geological regimes (Meshram and Sharma 2017). In addition, the morphometric analysis may be useful in a multitude of scenarios, including natural resource management, flood frequency studies, landslide susceptibility mapping, groundwater potential estimation, erosion mitigation, and watershed prioritization (Aher et al. 2014; Farhan et al. 2016; Sujatha and Sridhar 2017; Tesema 2022b; Tiwari and Kushwaha 2021).

So far, numerous methodologies have been employed to prioritize the sub-basins for soil erosion and landslide prevention. Abdeta et al. (2020), Chandniha and Kansal (2017), and Tesema (2022b) used the compounding method; Dubey and Jha (2022), Pathare and Pathare (2021), Shekar and Mathew (2022) used principle component analysis (PCA); Duulatov et al. (2021), Gharibreza et al. (2021), Maury et al. (2019), Singh and Kansal (2023), and Srinivasan et al. (2021) used the Revised Universal Soil Loss Equation (RUSLE); Kumar and Sarkar (2022), Shelar et al. (2022), and Yalcin et al. (2011) used the analytical hierarchy process (AHP), and Joshi et al. (2021), Meshram et al. (2019), Rahaman et al. (2015, 2015), Sridhar and Ganapuram (2021), and Yalcin et al. (2011) used fuzzy-AHP aimed at sub-watersheds prioritization for erosion control.

The present study also strives to systematically prioritize watersheds susceptible to soil erosion by employing a multifaceted methodology. This approach amalgamates morphometric analysis with the compound average method (CM) and Principal Component Analysis (PCA), hydrological assessment utilizing the Geomorphological Instantaneous Unit Hydrograph (GIUH) technique, and semi-empirical modeling with the Revised Universal Soil Loss Equation (RUSLE). The GIUH method’s rationale lies in its efficacy in identifying regions prone to flash flood impacts. At the same time, the morphometric approach, integrating CM and PCA, effectively identifies zones with maximum landslide potential. Moreover, the RUSLE model prioritizes the sub-watersheds based on annual soil erosion loss. In the past three decades, the region has experienced substantial soil erosion instigated by natural disasters such as cloudbursts, flash floods, and ensuing landslides (Kansal and Singh 2022; Singh and Kansal 2022a). This research distinguishes itself in flash flood-induced erosion studies through several novel contributions. Firstly, it adopts a comprehensive approach, amalgamating morphometric, hydrological, and semi-empirical methods—a holistic blend seldom observed in prior studies, offering a more nuanced understanding of landslide complexities. Such an integrated approach yields a more robust evaluation of sub-watershed vulnerability than standalone methods. This research further sets itself apart by emphasizing a targeted analysis at the sub-watershed level, promoting localized mitigation strategies, contrasting with many studies focusing on broader regions or basins. The Alaknanda River basin, with its unique geographic and climatic intricacies, stands as the focal point of this study, addressing its distinct vulnerabilities. Additionally, by incorporating field investigations pertaining to soil erosion events from the past three decades, the study presents an updated and contemporary perspective on challenges in the basin of the Indian Himalayan Region.

Material and methods

Datasets and tools

In this research, the main datasets employed consist of the digital elevation model (DEM) acquired from the Shuttle-Radar Topography Mission (SRTM), which can be accessed through the Open Topography portal (https://portal.opentopography.org/). The DEM hydro-processing and watershed delineation are performed using the “Hydrologic Engineering Center Geospatial Hydrologic Modeling Extension” (HEC-GeoHMS) extension (Hydrologic Engineering Center) of ArcGIS 10.5. The resulting small sub-watersheds are merged into 12 major sub-watersheds using the basin merge tool. Then, the morphometric analysis is performed using the ArcGIS Morphometric tool extension (Ayad Ali Faris Beg 2015). Rainfall data for individual sub-basins from 1990 to 2022 are extracted using the Climate Hazards Group InfraRed Precipitation with Station (CHIRPS) dataset available at the Climate Engine portal (http://climateengine.org). Furthermore, a connection is established between the sub-watershed prioritization and past landslides and flash-flood inventories in the region. These locations are essential for identifying the region’s susceptibility, hazard, and risk of disasters. The landslide inventories for the Alaknanda River basin are downloaded from the Bhukosh website (https://bhukosh.gsi.gov.in/) of the Geological Survey of India (GSI). The GSI has carried out landslide inventories in different parts of India using various techniques, including remote sensing, field surveys, and geological and geophysical investigations. The landslide inventories by GSI in these areas provide valuable information on the types of landslides, their causes, and their distribution. Additionally, a comprehensive catalogue of prior flash flood incidents is assembled by collating information from published sources (Mishra et al. 2022; Singh and Pandey 2021a; Singh and Kansal 2022b), as well as from reports and newspaper articles available on the official website of the South Asia Network on Dams, Rivers, and People (SANDRP).

Methodology

The methodology is divided into four parts, as shown in Fig. 1. Starting from the watershed delineation, which estimates each sub-watershed boundary shapefile and drainage network, morphometrical analysis is carried out in each sub-watershed to estimate its linear, areal, and relief aspects. Followed by prioritizing the sub-watersheds using the PCA, CM, GIUH, and RUSLE. Finally, an overall prioritization of the sub-watershed is estimated based on the weighted average score, and the observations and validations are made based on the landslide’s inventories, flash-flood incidents, and field observations.

Fig. 1
figure 1

Flowchart of the methodology

Full size image

Watershed Delineation

For both hydrologic and environmental studies, watershed delineation is a common practice. Watersheds can be automatically derived from DEM using geographic information systems (GIS) technology because DEMs provide a good representation of the terrain. Automated watershed delineation methods have existed since the 1980s when they were first implemented in GIS (Fairfield and Leymarie 1991). The watershed delineation using the terrain pre-processing function of HEC-GeoHMS extension of ArcGIS incorporates “Fill Sink, Flow Direction, Flow accumulation, Stream Definition and Segmentation, Catchment Grid and polygon formation, and Drainage Line processing.”

Sub-watershed morphometric analysis

The morphometric analysis uses the input parameters, viz., sub-watershed-wise river order, boundary area and length, and main-channel length. By calculating the geometry of the sub-basin polygons, their areas and perimeters were determined. Stream orders were determined from Strahler’s method (Strahler 1952). From the analysis, about thirteen important parameters influencing landslides and soil erosion are taken, namely, “Mean Bifurcation Ratio (Rb), Drainage density (Dd), Stream frequency (Fs), Circularity ratio (Rc), Form factor (Rf), Elongation ratio (Re), Drainage Texture (T), Length of overland flow (Lo), Compactness Coefficient (Cc), Relief ratio (Rh), Relative relief (Rr), Ruggedness number (Rn) and average Slope (S).” The formula and the relationship used for the morphometric analysis are shown in Table 1.

Table 1 Relationships used in Morphometric analysis
Full size table

Sub-watershed prioritization using Compounding average Method (CM)

In order to prioritize sub-watersheds, linear parameters are ranked based on their values, with the highest value assigned a rank of 1, the next highest a rank of 2, and so forth. The erodibility of a region is directly related to linear drainage parameters like T, Rb, Dd, Fs, and Lo (Nookaratnam et al. 2005). Shape parameters such as Re, Cc, Rc, and Rf vary inversely with erodibility (Meshram and Sharma 2017; Rai et al. 2009). The shape parameters are ranked based on their absolute values in ascending order. The parameter with the lowest value is given a rank of 1, the next lowest is given a rank of 2, and so on. The stream gradient, which is determined by the relief parameters, such as Rh, Rr, Rn, and S, directly impacts channel erosion, runoff, and the lag time of water during high-flow events (Panwar et al. 2017). Therefore, relief parameters were ranked in descending order, with the highest value receiving a rank of 1, the next highest receiving a rank of 2, and so on. Equal weights are assigned to all the morphometric parameters. Now, the rank of each sub-watershed is combined using the compounding average method. The CM approach adds the ranked values of all thirteen parameters across the chosen twelve sub-watersheds as composite parameters. Ultimately, the average values of these composite parameters are employed to establish the final priority. The highest priority is given to the lowest average value and vice versa (Bhattacharya et al. 2021; Chandniha and Kansal 2017; Pathare and Pathare 2021; Shekar and Mathew 2022; Tesema 2022a).

Sub-watershed prioritization using Principal Component Analysis (PCA)

PCA is a technique used for reducing the dimensionality of a large set of correlated variables to a smaller set of uncorrelated variables. This is achieved by transforming the original variables to a new set of orthogonal axes, which generates new principal components (PCs) that are uncorrelated with each other (Dubey 2022; Meshram and Sharma 2017; Shekar and Mathew 2022). Maximum data variation is found in the first principal component (PC1), and variation gradually decreases in subsequent components. The new components are key in explaining the greatest possible variation in the original variables (Bhat et al. 2022). The original factor loading matrix and the rotational factor loading matrix are employed for the estimation. Summation and normalization of significant loadings from the four PCs yield the weights for each component. Finally, the sub-watersheds are ranked using the weighted sum method.

Sub-watershed prioritization using Geomorphological Instantaneous Unit Hydrograph (GIUH)

The GIUH is a hydrological model introduced by Rodríguez-Iturbe et al. (1979), which aimed to represent the relationship between rainfall and runoff in watersheds. The GIUH model considers that the hydrological response of a catchment is a result of the interactions between the rainfall and the physical characteristics of the catchment, such as its topography, soil, and vegetation cover. (Gupta et al. (1980) and Rigon et al. (2016) further enhanced the GIUH model by incorporating the concept of travel time distribution. This enhancement involves dividing the catchment into sub-catchments, considering each has a different travel time distribution function.

The instantaneous hydrograph peak of the GIUH is given by,

$${{\varvec{q}}}_{\mathbf{p}}=(\frac{1.31}{{{\varvec{L}}}_{{\varvec{\Omega}}}}){{\varvec{R}}}_{\mathbf{L}}^{0.43}{\varvec{v}}$$
(1)

The time to peak (tp) of the GIUH is given by,

$${{\varvec{t}}}_{\mathbf{p}}=0.44 {(\frac{{{\varvec{R}}}_{\mathbf{B}}}{{{\varvec{R}}}_{\mathbf{A}}})}^{0.55}{{{\varvec{R}}}_{\mathbf{L}}}^{-0.38}\frac{{{\varvec{L}}}_{{\varvec{\Omega}}}}{{\varvec{v}}}$$
(2)
$${{\varvec{t}}}_{\mathbf{b}}=\frac{2}{{{\varvec{q}}}_{\mathbf{p}}}$$
(3)

where tb = base time (hour), tp = time to peak (hour); qp = instantaneous hydrograph peak (/hour); v = dynamic parameter velocity (m/s), Ln = length of the highest order stream (km), and are RA = stream-area ratio, RB = bifurcation ratio, RL = stream-length ratio.

To estimate the dynamic parameter velocity (V) for a watershed, a combination of velocity and the Kirpich formula can be employed, as suggested by Nongthombam et al. (2011)

$${{\varvec{t}}}_{\mathbf{c}}=0.01947 {{\varvec{L}}}^{0.77}{{\varvec{S}}}^{-0.385}$$
(4)
$${{\varvec{t}}}_{\mathbf{c}}=\frac{1}{60}\times \frac{{\varvec{L}}}{{\varvec{v}}}$$
(5)
$${\varvec{v}}=0.8562 {{\varvec{L}}}^{0.23}{{\varvec{S}}}_{{\varvec{B}}}^{0.385}$$
(6)

where tc = time of concentration (min), SB = mean slope of the basin (m/m), and L = mainstream length (m).

Derivation of GIUH based on the Nash model

The Nash Geomorphic Instantaneous Unit Hydrograph (GIUH), a distributed rainfall-runoff model, employs routing instantaneous inflow through linear reservoirs with equal storage coefficients. In a given watershed with “n” reservoirs, a unit pulse of rainfall is input over an infinitesimally short time Δt → 0, generating an outflow with ordinate u(t) representing the Instantaneous Unit Hydrograph (IUH).

Outflow resulting from the first reservoir is calculated using Eq. 7,

$${{\varvec{u}}}_{1}\left({\varvec{t}}\right)=-\frac{{{\varvec{e}}}^{- \frac{{\varvec{t}}}{{\varvec{k}}}}}{{\varvec{k}}}$$
(7)

The outflow u1(t) from the first reservoir flows into the second reservoir, given by,

$${{\varvec{u}}}_{2}\left({\varvec{t}}\right)=-\frac{{{\varvec{e}}}^{-\frac{{\varvec{t}}}{{\varvec{k}}}}}{{\varvec{k}}}(1-{{\varvec{e}}}^{-\frac{{\varvec{t}}}{{\varvec{k}}}})$$
(8)

Thus, continuing the same process for “n” no. of reservoirs, the resultant outflow, known as Nash model ordinate of IUH, is given by,

$${\varvec{u}}\left({\varvec{t}}\right)=({\frac{{\varvec{t}}}{{\varvec{k}}})}^{{\varvec{a}}-1}\frac{{{\varvec{e}}}^{-\frac{{\varvec{t}}}{{\varvec{k}}}}}{{\varvec{k}}\Gamma ({\varvec{a}})}$$
(9)

The Horton ratios and parameters of the gamma distribution are mathematically related, where u(t) represents the ordinate of IUH (hour−1), t represents time interval sampling (hour), Γ (a) is the gamma function, and n and k are parameters of the Nash model, representing number of linear reservoirs and storage coefficient (hour), respectively.

Estimation of geomorphological parameter of Nash model

Connecting the scale (k) and shape parameter (n) of the Nash model with the qp and tp of GIUH, one can fully determine the shape of GIUH. On substituting and simplifying Eq. 9 (Rai et al. 2009),

$$\frac{\partial \mathbf{l}\mathbf{n}[{\varvec{u}}({\varvec{t}})]}{\partial {\varvec{t}}}=[-\frac{1}{{\varvec{k}}}+\frac{\left({\varvec{n}}-1\right)}{{\varvec{t}}}]$$
(10)
$$\frac{({\varvec{n}}-1)}{\Gamma ({\varvec{n}})}\mathbf{exp}\left[-\left({\varvec{n}}-1\right)\right]{\left({\varvec{n}}-1\right)}^{{\varvec{n}}-1}=0.5764{(\frac{{{\varvec{R}}}_{\mathbf{B}}}{{{\varvec{R}}}_{\mathbf{A}}})}^{0.55}{{\varvec{R}}}_{\mathbf{L}}^{0.05}$$
(11)

The “n” parameter is determined by solving Eq. 11 through the Newton–Raphson method, while the estimation of the “k” value for a specific “v” value is done using Eq. 12, given by,

$${\varvec{k}}= 0.44{(\frac{{{\varvec{R}}}_{\mathbf{B}}}{{{\varvec{R}}}_{\mathbf{A}}})}^{0.55}{{\varvec{R}}}_{\mathbf{L}}^{0.05}\frac{1}{({\varvec{n}}-1)}\frac{{{\varvec{L}}}_{\boldsymbol{\Omega }}}{{\varvec{v}}}$$
(12)

The estimated instantaneous hydrograph peak, velocity, and time to the peak are then used to identify the critical sub-watershed vulnerable to flash-flood impacts. The sub-watershed with a high value of velocity, instantaneous hydrograph peak, and low value of time to peak is given high priority. This information helps develop flood warning systems and mitigate future flash flooding risks. Identifying critical sub-watersheds is crucial in minimizing the impact of flash floods.

Sub-watershed prioritization using Revised Universal Soil Loss Equations (RUSLE) (obtained from Singh and Kansal 2023)

RUSLE is a model that employs a comprehensive series of mathematical equations to estimate the average annual soil loss and sediment yield due to inter-rill and rill erosion processes (Renard et al. 1997) and is denoted as per Eq. 13:

$$\mathbf{A}=\mathbf{R} \times \mathbf{K} \times \mathbf{L}\mathbf{S} \times \mathbf{C}\times \mathbf{P}$$
(13)

In the given equation, the rainfall-runoff erosivity factor is represented by R (MJ mm ha−1 h−1 year−1), which considers various rainfall characteristics, including volume, duration, and intensity. The average annual soil loss is given by A (ton ha−1 year−1). In this study, the annual average precipitation (AAP) data from 1991 to 2020 is extracted at grid locations and interpolated for the basin to estimate the rainfall erosivity factor using Eq. 14 by (Das et al. 2018; Dutta et al. 2015; Sandeep et al. 2021; Singh et al. 1981):

$$\mathbf{R}=79+0.363\times \mathbf{A}\mathbf{A}\mathbf{P}$$
(14)

The soil erodibility factor K (t ha h ha−1 MJ−1 mm−1) represents the resistance of soil particles to erosion caused by storm events. It is typically determined by examining a specific location’s soil and surface characteristics (Wischmeier and Smith 1978). To estimate soil characteristics, the NBSS-LUP soil map is vectorized using ArcGIS software. K values are calculated through nomographs that consider soil structure, soil texture and permeability, and percentage of silt, clay, organic matter, and sand (Wischmeier and Smith 1978). Equations 15 and 16 are utilized to determine the K factor.

$$\mathbf{K}=\frac{2.1\times {10}^{-4}\left(12-\mathbf{O}\mathbf{M}\right){\mathbf{M}}^{1.14}+3.25\left(\mathbf{s}-2\right)+2.5(\mathbf{p}-3)}{759.4}$$
(15)

The variables used in this context are as follows: s represents soil structure code, p represents soil permeability code, OM represents %organic matter, and M denotes the primary soil particle size fraction, given by:

$$\mathbf{M}=(\mathrm{\%}\mathbf{s}\mathbf{i}\mathbf{l}\mathbf{t}+\mathrm{\%}\mathbf{s}\mathbf{a}\mathbf{n}\mathbf{d})\times (100-\mathrm{\%}\mathbf{c}\mathbf{l}\mathbf{a}\mathbf{y})$$
(16)

The topographic steepness factor, or LS, represents the combined effect of slope length and steepness on soil loss rates (Batar and Watanabe 2021; Biswas et al. 2021; Das et al. 2020). So, in this study, the LS factor values are estimated using the Eq. 17 given by (Moore and Burch 1986) as:

$$\mathbf{L}\mathbf{S}={1.4\times (\mathbf{F}\mathbf{l}\mathbf{o}\mathbf{w}\; \mathbf{a}\mathbf{c}\mathbf{c}\mathbf{u}\mathbf{m}\mathbf{u}\mathbf{l}\mathbf{a}\mathbf{t}\mathbf{i}\mathbf{o}\mathbf{n} \times \frac{\mathbf{C}\mathbf{e}\mathbf{l}\mathbf{l}\; \mathbf{S}\mathbf{i}\mathbf{z}\mathbf{e}}{22.13})}^{0.4} \times {(\frac{\mathbf{sin}\;\mathbf{s}\mathbf{l}\mathbf{o}\mathbf{p}\mathbf{e} }{0.0896})}^{1.3}$$
(17)

In this equation, cell size denotes the grid cell size or DEM resolution, and sin slope refers to the slope degree value.

For determining the C factor, this study employs the method introduced by Van der Knijff et al. (2000) as they found that the C factor does not exhibit a linear relationship with NDVI (Normalized Difference Vegetation Index) but rather decreases exponentially, given as per Eq. 18:

$$\mathbf{C}=\mathbf{e}\mathbf{x}\mathbf{p}[-{\varvec{\upalpha}}\frac{\mathbf{N}\mathbf{D}\mathbf{V}\mathbf{I}}{{\varvec{\upbeta}}-\mathbf{N}\mathbf{D}\mathbf{V}\mathbf{I}}]$$
(18)

In this case, α and β are parameters defining the NDVI-C curve shape, with α = 2 and β = 1. The C factor, also known as the cover and cropping management factor, describes the influence of vegetation cover on soil erosion rates, as vegetation prevents rainwater from directly contacting soil particles and causing erosion (Das et al. 2020).

The supporting practices factor, represented by P, signifies the ratio of soil erosion caused by a specific support practice to the soil loss produced by straight-line tillage in both uphill and downhill directions, which yield equivalent soil loss amounts (Ganasri and Ramesh 2016). The estimation of the P factor value, which varies from 0 to 1 (as shown in Table 2), considers the basin’s land use and land cover (LULC). A “P” factor value of 1 indicates that no effective conservation practices have been implemented. The RUSLE model integrated with GIS technology can comprehensively estimate soil erosion by predicting soil loss at a pixel level. Additionally, it is worth noting that the P, C, and LS values used in this model are dimensionless (Das et al. 2018; Dutta et al. 2015).

Table 2 Commonly used conservation practice (P) factors as per the LULC
Full size table

This study is crucial for understanding and addressing the flash flood-induced erosion contributing to significant soil loss and degradation of pedosphere strata in the Indian Himalayan Region. Thus, the Alaknanda River basin is chosen as a case study for the analysis due to the confluence of factors exacerbating this phenomenon, including extreme precipitation events, undulating topographical features, and suboptimal soil and water conservation methodologies.

Study area

The Alaknanda basin in Uttarakhand, India, is part of the upper Ganga basin and has an outflow at Rudraprayag (Fig. 2). It has a basin size of 10272 sq km. The Bhagirath Kharak and Satopanth Glaciers, situated within the western Garhwal Himalayas, form the Alaknanda River Basin. This basin lies in Uttarakhand, with geographic coordinates spanning 78°45′ E to 80°15′ E in longitude and 30°10′ N to 31°5′ N in latitude. Elevations within the region vary between 609 and 7804 m above sea level. The study area’s northern section experiences a severe winter climate. The region is dominated by a tropical monsoon climate, with roughly 75% of annual precipitation, averaging 1600 mm, between June and September (Panwar et al. 2016; Singh and Kansal 2020). The area’s vulnerability to natural disasters, such as earthquakes, landslides, flash floods, and cloudbursts, makes it susceptible to soil erosion and landslides.

Fig. 2
figure 2

Alaknanda River basin showing landslide occurrence and flash-flood sites

Full size image

Over the past three decades, the Alaknanda River basin in Uttarakhand has experienced numerous severe weather incidents, leading to flash floods, landslides, and erosions. The river basin’s districts of Chamoli and Rudraprayag have been particularly affected by such events, including flash floods, landslides, and cloudbursts. As a result, mudslides, stream erosion, and debris movement downstream have caused extensive damage to properties, bridges, and other human-made infrastructure (Azmeri et al. 2016; Singh and Pandey 2021b). In February 2021, an avalanche-induced flash flood occurred in the Chamoli district’s Rishiganga and Dhauliganga river watersheds. This disaster claimed the lives of approximately 200 people and inflicted severe damage on the Rishiganga and Tapovan Vishnugad hydropower projects. Furthermore, the catastrophe led to riverbank collapse, scouring, erosion, bridge congestion, debris flows, and sediment deposition. (Mishra et al. 2022; Singh et al. 2023; Singh and Kansal 2022a). In October 2022, a landslide hit three houses in the Tharali area of Uttarakhand’s Chamoli district due to landslides. Four people were killed and another injured in the incident. In May 2023, a landslide blocked the Badrinath highway in Uttarakhand’s Chamoli, leaving several tourists stranded. Escalating human activities, including hydropower projects, tourism, and construction, have placed tremendous stress on the region’s delicate ecosystem, resulting in a rise in natural disasters. Despite expert warnings, these activities persist in endangering the region’s ecological equilibrium and the livelihoods of those who rely on it.

Results

Morphometric parameters estimation

Linear parameters

The linear parameter of morphometry is a set of measurements used to describe a watershed’s size, shape, and drainage characteristics. Stream order, stream network length, watershed perimeter, and longest flow path are important linear parameters. Other morphometric parameters commonly include area, slope, elevation, and shape measurements. These parameters are useful for understanding and managing regional water and land use interactions.

Stream number (Nu) :

The number and size of streams in a region are largely influenced by their physiography, geomorphology, and geology (Rai et al. 2017). Nu is important in understanding the structure and function of river systems, as different orders of streams have distinct characteristics in terms of flow, sediment transport, and ecological processes. In this study, WS12 has the highest Nu, with 525 first-order, 112 s-order, 25 third-order, 5 fourth-order, and 1 fifth-order streams, as shown in Table 3 and Fig. 3.

Table 3 Sub-watershed wise stream order and numbers
Full size table
Fig. 3
figure 3

Stream order in various sub-watersheds of Alaknanada Basin

Full size image

Stream length (Lu) :

Lu is an important parameter in hydrology, as it can be used to estimate the volume and velocity of water flow within a watershed and predict the potential for flooding or erosion. It measures the total length of a river or stream channel within a given watershed or drainage basin. In all sub-watersheds, the Lu is the longest for the first order, decreasing with an increase in the stream order (Table 3). The basin has a total stream length of 6825.1 km, with 3546.8 km for first-order streams, 1578.7 km for second-order streams, 838.5 km for third-order streams, 551.2 km for fourth-order streams, 300.6 km for fifth-order streams, and 9.2 km for sixth-order streams. The WS12 has a maximum Lu of 1206.2 km, while the WS7 has a minimum Lu of 136.2 km.

Bifurcation ratio (Rb) :

Rb is an important parameter in understanding the structure and function of river systems, as it can provide insight into the connectivity and branching patterns of streams within a network. High bifurcation ratios indicate a more branching, dendritic pattern, while low ratios indicate a more linear, trellis-like pattern (Horton 1945; Rathore et al. 2022; Sridhar and Ganapuram 2021). Rb indicates relief and erosion. The Rb ranges between 3.6 (WS6) and 5.72 (WS11), as shown in Table 3.

Drainage density (Dd) :

Dd is a geomorphological parameter that describes the degree of stream branching and the efficiency of water flow through a watershed. It is defined as the total length of all stream channels in a drainage basin per unit area of the basin. The Dd of a watershed is influenced by several factors, including climate, topography, geology, and vegetation cover (Farhan et al. 2016). Regions with highly permeable materials and vegetation cover, as well as low relief, tend to exhibit a low drainage density (Dd) value, while Dd values tend to be high in regions with impermeable subsurface material, mountainous relief, and sparse vegetation (Gayen et al. 2019; Halder et al. 2021; Rathore et al. 2022). In the study area, Dd ranges between 0.62 (WS5) and 0.78 (WS8), as shown in Table 4.

Table 4 Morphometric analysis results of Alaknanda sub-watersheds
Full size table

Stream frequency (Fs) :

Fs measures the number of stream channels that intersect a given area within a drainage basin (Horton 1932). A high stream frequency value indicates a larger surface runoff, resulting in an early peak discharge. In contrast, a low stream frequency value indicates a landscape with high permeability and low relief. It has been noticed that high Fs are observed in WS8 (0.38) and low Fs in WS7 (0.33), as shown in Table 4.

Drainage Texture (T):

T refers to the degree of channel spacing in a topography that rivers have dissected, and various factors such as vegetation, climate, precipitation, infiltration ability, soil characteristics, rock type, and extent of landscape development influence T values (Smith 1950). WS1 has a maximum T = 2.13, while WS9 has a minimum T = 0.9 (Table 4).

Length of overland flow (Lo):

Lo is a parameter describing the distance water travels as overland flow before it reaches a stream channel. The Lo ranges between 0.31 (WS5) and 0.39 (WS8), as shown in Table 4.

Areal parameters

The areal parameters of morphometry are used to characterize the physical properties of a watershed and help to understand the hydrological processes that occur within it.

Compactness coefficient (Cc):

Cc measures the shape complexity of a basin. A higher value of Cc indicates a more compact and spherical shape, while a lower value indicates a more irregular and elongated shape (Horton 1945). The WS has the maximum Cc = 2.33, while WS10 has the minimum T = 1.71 (Table 4).

Circularity ratio (Rc):

The Rc is used to describe the shape of a basin (Miller 1953). Basins with higher circularity ratios tend to have more uniform drainage patterns. They may be more resilient to changes in water flow and sediment transport, while basins with lower circularity ratios may be more susceptible to erosion and sedimentation in certain areas. Rc ranges between 0.19 (WS12) and 0.35 (WS10), as shown in Table 4.

Form factor (Rf):

The Rf is often used to evaluate the shape of watersheds, which can affect the distribution of water and sediment within the basin. The smaller value indicates a more elongated basin. The Rf ranges between 0.26 (WS12) and 0.70 (WS5), as shown in Table 4.

Elongation ratio (Re):

Re is a geomorphometric parameter that describes the shape of a watershed and helps understand its hydrological nature. Re ranges between 0.58 (WS12) and 0.94 (WS5), as shown in Table 4.

Relief parameters

Relief parameters in morphometry refer to the measurement of variations in elevation or vertical differences across the surface of a landscape. These parameters are essential in understanding a watershed’s topography and its role in controlling the movement of water and other substances through the landscape. Understanding the relief parameters of a landscape is crucial in determining the spatial distribution of water and sediment and identifying areas of potential erosion or flooding.

Relief ratio (Rh):

Rh is a morphometric parameter used to measure the overall steepness of the terrain within a defined area. High relief ratios indicate steeper slopes and more rugged terrain, which can increase the likelihood of erosion and landslides during heavy precipitation events. Basins with steeper slopes typically exhibit faster runoff, resulting in more pronounced peak discharges and increased erosive potential (Bhattacharya et al. 2021; Chandniha and Kansal 2017). The Rh ranges between 0.07 (WS12) and 0.24 (WS7), as shown in Table 4.

Relative relief (Rr):

The elevation difference between the maximum and minimum points in a landscape relative to its perimeter is described by Rr. Watersheds with high Rr values tend to have steep slopes and rugged terrain and are more susceptible to erosion and sediment transport. The Rr value ranges between 1.70 and 4.90, as shown in Table 4.

Ruggedness number (Rn):

Rn measures the topographic variability of the landscape and provides a quantitative measure of the ruggedness or roughness of the terrain. The Rn value ranges between 1.67 (WS10) and 4.58 (WS8), as shown in Table 4.

Average slope (S):

S is a measure of the steepness of the land surface within the watershed and thus influences the flow of water and affects erosion and sediment transport. A high average slope generally indicates a steep and rugged terrain, while a low average slope indicates a flatter landscape. The S value varies from 0.57(WS9) to 0.77(WS4), as shown in Table 4.

Morphometry-based compounding average method (CM)

Using the compounding method, Table 5 displays the final scores and rankings for all twelve sub-watersheds. The highest and lowest prioritized scores among the sub-watersheds are 8.2 (WS3) and 4.1 (WS8), respectively (Fig. 4). A higher score indicates a higher degree of erosion in a particular sub-watershed and, therefore, a higher priority for conservation or restoration efforts.

Table 5 The sub-watershed prioritization map using the compounding approach
Full size table
Fig. 4
figure 4

The sub-watershed prioritization map using the compounding approach

Full size image

Morphometry based-PCA approach

Intercorrelation among the morphometric parameters

The PCA is exclusively utilized for morphometric analysis to decrease the number of dimensions (Madiona et al. 2019). Utilizing SPSS 25.0 software, the inter-correlation analysis of the morphometric parameters is conducted for the Alaknanda sub-watershed. The correlation matrix, consisting of 13 geomorphic parameters, revealed strong associations between Dd and Lo, Rc and Cc, Rf and Re, and Rh and Rr. Furthermore, high correlations, with correlation coefficients ranging from 0.7 to 0.9, are observed between Rb and Rf, between Rb and Re, between Re and Cc, and between T and ruggedness number (Rn). Some moderate correlations (correlation coefficient between 0.5 and 0.7) exist between Dd and Fs, between Dd and Rc, Dd and Cc, Fs and Lo, Fs and Rn, Rc and Rf, Rc and Re, Rc and Lo, Rc and Rh, Rc and Rr, Rf and Cc, Rf and Rh, Re and Rh, Lo and Cc, Cc and Rh, Cc and Rr, and Rh and S. At this stage, it is challenging to group the parameters into components, so to simplify the process, the next step involves applying principal component analysis to the correlation matrix (Table 6).

Table 6 Intercorrelation matrix of the morphometric parameters
Full size table

First-factor loading matrix

The correlation matrix of 13 geomorphic parameters generates the first factor loading matrix. As shown in Table 7, the first four components account for approximately 89.71% of the total variance in the Alaknanda watershed. Table 8 indicates that the first component exhibits a very high correlation with Cc and Rc, a high correlation with Re, Rf, and Rh, and a moderate correlation with Rb, Dd, Rr, and Lo. The second component is highly correlated with Fs, Rn, Dd, and Lo and moderately correlated with Dt. The third component is moderately correlated with Rb and Dt, while the fourth is moderately correlated with Rh, Rn, and S in the Alaknanda watershed. According to the analysis, certain parameters exhibit strong correlations, while others show moderate correlations with some components, and some parameters do not correlate with any. However, it is currently impossible to determine which components are significantly correlated. Thus, it is essential to rotate the initial factor loading matrix to enhance the correlation.

Table 7 Total variance based on PCA
Full size table
Table 8 Unrotated matrix
Full size table

Rotation of the first‑factor loading matrix

The transformation matrix is multiplied with the selected components to create a rotated factor loading matrix (Table 9) from the first-factor loading matrix. It is observed that the first component is highly correlated with Re and Rf, highly correlated with Rb, and moderately correlated with Cc and Rc, all as determined by the rotated factor loading matrix. The second factor has a moderate correlation with Cc, Rc, and Fs but a very high correlation with Dd and Lo. The third component displays a very high correlation with Rr and Rh and a moderate correlation with S. The fourth component has a very high correlation with Rn and a high correlation with Dt. A total of 13 parameters were utilized in the morphometric analysis for prioritization; the PCA-based approach reduces the number of components from 13 to 4, as shown in Fig. 5, thereby saving computational time and assisting fluvial geomorphologists and hydrologists in selecting significant component weights (Meshram et al. 2019; Meshram and Sharma 2017). Normalization of each component indicates that PC2 carries the most weight at 0.3, followed by PC4 at 0.26, PC3 at 0.25, and PC1 at 0.2.

Table 9 Rotated matrix
Full size table
Fig. 5
figure 5

The first four principal components map for each sub-watershed

Full size image

The matrix multiplication of the rotated matrix with their corresponding weights and the rank matrix of each parameter results in the prioritization of each sub-watershed, as shown in Table 10. The first rank is given to the sub-watershed with the lowest weighted average value, and vice versa. The average values range from 6.9 (WS8) to 26.9 (WS10), resulting in a priority rank of 1 for WS1 and 12 for WS10. Figure 6 illustrates a comprehensive map of prioritization using PCA.

Table 10 The sub-watershed prioritization map based on PCA approach
Full size table
Fig. 6
figure 6

The sub-watershed prioritization map based on PCA approach

Full size image

GIUH based on the Nash model

The flow parameters of the GIUH model are evaluated using Eq. 1 to Eq. 12 as follows:

  1. 1.

    For sub-watershed 1

    Bifurcation ratio (RB) = 4.65, length ratio (RL) = 0.56,

    Main channel length (L) = 69615.35 m, mean slope (SB) = 0.66

    Highest order stream length (LΩ) = 46.67, area ratio (RA) = 0.70

    Thus, velocity (v) \(=0.8562 {L}^{0.23}{S}_{\mathrm{B}}^{0.385}=0.8562\times 69615.{35}^{0.23} \times {0.66}^{0.385} = 9.478\mathrm{ m}/\mathrm{s}\)

    Peak discharge (qp) \(=\left(\frac{1.31}{{L}_{\Omega }}\right){R}_{\mathrm{L}}^{0.43}v=\left(\frac{1.31}{46.67}\right)\times {0.56}^{0.43}\times 9.48=0.207\mathrm{ inch}/\mathrm{h}\)

    Time to peak (tp) \(=0.44 {(\frac{{R}_{B}}{{R}_{A}})}^{0.55}{{R}_{L}}^{-0.38}\frac{{L}_{\Omega }}{v}=0.44 {\times (\frac{4.65}{0.7})}^{0.55}{\times 0.56}^{-0.38}\times \frac{46.67}{9.478}=7.691\mathrm{ h}\)

    Base time (tb) \(= 2/{q}_{\mathrm{p}} = 2/0.207 = 9.67\mathrm{ h}\)

    Now, \(\frac{(n-1)}{\Gamma (\mathrm{n})}\mathrm{exp}\left[-\left(n-1\right)\right]{\left(n-1\right)}^{n-1}=0.5764{(\frac{{R}_{B}}{{R}_{A}})}^{0.55}{R}_{L}^{0.05}=0.5764{\times (\frac{4.65}{0.7})}^{0.55}\times {0.56}^{0.05}=1.591\)

    Solving for n gives n = 17

    Hence, \(k=0.44{(\frac{{R}_{B}}{{R}_{A}})}^{0.55}{R}_{L}^{0.05}\frac{1}{(n-1)}\frac{{L}_{\Omega }}{v}=0.44{\times (\frac{4.65}{0.7})}^{0.55}{\times 0.56}^{0.05}\times \frac{1}{(17-1)}\times \frac{46.67}{9.478}=0.481\)

    A similar approach is utilized to estimate these parameters for all the sub-watersheds, as shown in Table 11. The dynamic velocity varies from 7.335 m/s (WS7) to 10.761 m/s (WS12); qp is the peak rainfall intensity, which varies from 0.120 inch/h (WS12) to 1.034 inch/h (WS9); tp varies from 1.453 h (WS9) to 12.315 h (WS12); tb varies from 1.934 h (WS9) to 16.654 h (WS12); k varies from 0.102 (WS9) to 0.880 (WS12), and n varies from 11.3 (WS4) to 17.0 (WS1). These parameters are displayed in Fig. 7. These values are important in the hydrological modeling of watersheds that help predict the erosion potential of the sub-watersheds. The values are applied to prioritize sub-watersheds based on the assumption that high dynamic velocity and instantaneous peak magnify flash flood susceptibility, whereas a larger lag time and time to peak attenuate its impact. The resultant sub-watershed prioritization value obtained using GIUH is displayed in Table 12, and the map is shown in Fig. 8.

Table 11 GIUH parameters as per the Nash model
Full size table
Fig. 7
figure 7

The GIUH parameters map for each sub-watershed

Full size image
Table 12 The sub-watershed prioritization map based on GIUH approach
Full size table
Fig. 8
figure 8

The sub-watershed prioritization map based on GIUH approach

Full size image

RUSLE approach

Table 13 offers information on the soil erosion rate and priority ranking of 12 sub-watersheds within the Alaknanda watershed, as determined by the RUSLE model. The RUSLE model calculates soil erosion rates based on the product of five factors: the R factor, the K factor, the LS factor, the C factor, and the P factor, as well as the soil erosion rate. The R factor, which reflects rainfall erosivity, is the erosive force of rainfall, and it ranges between 294 (WS3) and 526 MJ mm/ha/h/year (WS12). A bigger R factor suggests a greater likelihood of soil erosion caused by intense rainfall. The K factor, which reflects soil erodibility, measures the soil’s susceptibility to erosion, ranging from 0.042 t ha h/MJ/ha/mm (WS8) to 0.088 t ha h/MJ/ha/mm (WS9). Soil with a higher K factor is more susceptible to erosion. The LS factor, which measures the topographic effect on soil erosion, spans from a minimum of 1.3 (WS9 and WS10) to a maximum of 1.5 (WS1, WS4, WS7, and WS8). Due to steeper slopes and longer slope lengths, a higher LS factor indicates a greater soil erosion risk. The C factor goes from a minimum of 0.2 (WS9 and WS10) to a maximum of 1.1 (WS2, WS3, and WS8) and depicts the land cover and management approach. A higher C factor suggests a lesser vegetation cover and a greater possibility for soil erosion. The P factor, which reflects the support practice, quantifies the influence of soil conservation practices on soil erosion and ranges from 0.22 (WS1) to 0.95(WS10). A lower P factor suggests more effective soil conservation methods and lower erosion risk. The projected soil erosion rate ranges from a minimum of 6 tons per hectare per year (WS2) to a high of 27.9 tons per year (WS7). Lastly, the sub-watersheds are sorted according to their soil erosion rate, with the sub-watershed with the greatest soil erosion rate (WS7) receiving the highest priority and the sub-watershed with the lowest soil erosion rate receiving the lowest priority (WS5). All the parameters are displayed in Fig. 9, and the sub-watershed prioritization map is shown in Fig. 10.

Table 13 The sub-watershed prioritization map based on RUSLE approach (Singh and Kansal 2023)
Full size table
Fig. 9
figure 9

The RUSLE parameters map for each sub-watershed

Full size image
Fig. 10
figure 10

The sub-watershed prioritization map based on RUSLE approach

Full size image

Comparative assessment of the individual and overall sub-watershed prioritization with the landslide and flash-flood inventories

A comparative analysis is conducted to evaluate the efficacy of the four different approaches to prioritization. The priority rankings are shown in ascending order, with the highest-priority sub-watershed receiving the lowest number and the lowest-priority sub-watershed receiving the highest number. According to the RUSLE approach, WS7 is ranked top priority, with WS5 and WS11 closely behind. On the other hand, the GIUH approach prioritizes WS6 as the most critical, with WS9 and WS4 being of secondary importance. Lastly, the CM approach designates WS8 as the highest priority, with WS1 and WS11 being the next in line. The PCA method identifies WS1 as having the highest priority, followed by WS8 and WS2.

The overall ranking of each sub-watershed is assessed through a weighted average methodology. The weighted average approach provides a more nuanced and customizable prioritization that considers the relative importance of different factors in determining sub-watershed priority (Kouli et al. 2014). By assigning different weights to each assessment method, stakeholders can focus on specific aspects of watershed management that are most relevant to their goals and objectives. Considering the average ranking of morphometric PCA and CM approaches, equal weightage was given to the morphometric, hydrological GIUH, and semi-empirical RUSLE techniques, as shown in Table 14. The overall priority indicates that sub-watersheds WS7, WS4, and WS6 require more urgent attention and resources to address their issues. These sub-watersheds have more significant erosion, flooding, or other environmental problems that must be mitigated to protect the surrounding ecosystem and communities. The observations are also validated from the landslides and flash-flood inventories depicting 264 landslides and 8 flash-flood incidents. The sub-watersheds WS9, WS8, WS11, WS5, and WS12 still require attention and resources, but the urgency is lower than the higher priority sub-watersheds. These sub-watersheds exhibit moderate environmental issues that can be effectively addressed through targeted interventions and regular monitoring activities. Furthermore, landslides are observed in these sub-watersheds at approximately 483 incidents, with an additional 19 flash-flood events. This information suggests that these sub-watersheds may benefit from focused interventions to mitigate the frequency and severity of environmental risks and hazards to safeguard the surrounding ecosystem and communities. The sub-watersheds WS1, WS3, WS2, and WS10 exhibit a relatively lower level of environmental issues and are considered less critical regarding watershed management. Nevertheless, these sub-watersheds should still undergo regular monitoring and management practices to prevent the occurrence of environmental hazards in the future. Additionally, the frequency of observed landslides in these sub-watersheds is estimated to be approximately 176 incidents, with a further 2 flash-flood events recorded. The combined sub-watershed prioritization map is shown in Fig. 11. The results indicate that the effectiveness of each method varied for each sub-watershed. Thus, it is crucial to have a well-informed decision-making process considering each sub-watershed’s specific conditions. Depending on the prevailing conditions, this insight aids in selecting the most efficient strategy for mitigating risks, safeguarding communities, and preserving infrastructure from the detrimental impacts of flash floods and associated landslides and erosions.

Table 14 Overall sub-watershed prioritization
Full size table
Fig. 11
figure 11

The sub-watershed prioritization map based on weighted average approach

Full size image

Discussions

This paper attempts to compare the morphometric PCA and CM, GIUH, and RUSLE methods for sub-watershed prioritization and estimate the overall priority of the sub-watershed. The morphometric approach focuses on the topography and shape of the catchment area (Arnous and Omar 2018; Das et al. 2021). It prioritizes sub-watersheds using slope, drainage density, and relief ratio metrics. This approach helps identify areas with a higher potential for landslides (Carrara et al. 1995; Jiménez-Perálvarez et al. 2009; Martha et al. 2010). The GIUH method, on the other hand, considers the catchment response based on the characteristics of the hydrograph, including velocity, peak discharge, time to peak, and base time (Rai et al. 2009; Sorman 1995). This approach provides a more accurate representation of the catchment response and is useful for identifying areas likely to be impacted by flash floods (Khaleghi et al. 2014; Kumar et al. 2007; Sahoo and Jain 2018). The RUSLE model calculates soil erosion by evaluating factors such as rainfall erosivity, soil erodibility, slope dynamics, cover management, and conservation practices, providing a comprehensive estimation of erosion under varying conditions (Bhattacharya et al. 2021; Fayas et al. 2019; George et al. 2021). By integrating the four techniques, a holistic watershed perspective emerges, enhancing the understanding of hydrological processes, soil erosion, landform traits, and land cover management. This insight guides informed decisions for watershed conservation and sustainable development.

The results of this research are evident in the classification of sub-watersheds pertaining to flash flood-induced landslides. The overall priority rankings, which categorized the sub-watersheds into three main categories, are drawn from an intricate analysis of landslide frequency, flash-flood incidents, and historical erosional patterns. These could be summarized as follows:

Category I (high priority): Sub-watersheds WS7, WS4, WS9, and WS6 fall under this category. These areas displayed significant signs of erosion and flooding. For instance, WS9, influenced heavily by its unique topographical characteristics, witnessed enhanced runoff, while WS6’s susceptibility is amplified by the region’s consistent heavy rainfall events, pushing its erosivity index higher. Specifically, WS6 recorded 244 landslides and 7 flash-flood incidents, depicting a watershed under consistent environmental stress. W7, W6, and WS4 are also characterized by very high soil erosion rates.

Category II (moderate attention): Sub-watersheds like WS11, WS8, WS5, and WS12, although not as severely impacted as the first category, still showcased vulnerabilities. These regions recorded moderate erosional traits but also hinted at upcoming potential risks if not addressed. WS12, for instance, had a higher rainfall erosivity and a history of rapid hydrological flow characteristics and landslide incidents. WS5, WS11, and WS12 demonstrated higher soil erosion rates.

Category III (regular monitoring): This group, including WS1, WS3, WS2, and WS10, presented lesser erosional characteristics but suggested underlying vulnerabilities that could escalate if left unchecked. While WS10 is presently stable, certain regions within it possess soil types susceptible to accelerated erosion because of 160 landslide incidents, warranting consistent monitoring.

To validate our findings, field surveys were undertaken (Fig. 12). Detailed photographs, damaged area coordinates, and local narratives confirmed our computational assessments, emphasizing the need for strategic interventions tailored for each category. The results showed that the areas with the highest impact were also located within the high-priority sub-basins, indicating the validity of the computed findings. The current situation also indicates the need for suitable mitigation strategies against flash-flood disasters in the Himalayan watersheds. The resulting landslides and induced erosion negatively impact agricultural productivity and cause downstream sedimentation, decreasing reservoir storage, and increasing flood risks.

Fig. 12
figure 12

Field survey photographs (October 2020 and 2021): a A damaged site close to the bridge in Raini village (WS7). b A flash flood eroded area in WS5. c An eroded upstream section of the Alaknanda River (WS8). d Flash flood induced landslides in Kedarnath (WS6). e Landslides and erosion near Gaurikund (WS6)

Full size image

While this study offers a comprehensive and integrative analysis of flash flood-induced landslides in the Alaknanda River basin, several limitations should be noted. The inherent assumptions of the utilized methods, PCA, CM, GIUH, and RUSLE, may not always mirror the intricate real-world scenarios, especially as RUSLE assumes steady-state conditions and the accuracy of results depends on the fidelity of input datasets. The findings are specific to a part of the Indian Himalayan Region, but there is an underlying assumption that they might apply to other regions. However, unaccounted anthropogenic and climatic factors can influence vulnerability assessments.

Conclusions

This study intricately juxtaposed four methodologies—morphometric PCA, morphometric combined (CM), Geomorphic Instantaneous Unit Hydrograph (GIUH), and Revised Universal Soil Loss Equation (RUSLE)—to discern the most suitable approach to sub-watershed prioritization concerning flash flood-induced landslides. Leveraging the insights from morphometric analysis, the profound influence of topographical nuances on potential landslide zones became evident. The GIUH method provided a granular perspective on hydrological aspects, especially highlighting vulnerabilities to flash floods. Additionally, the RUSLE model illuminated the intricacies of erosion dynamics. With the confluence of these methods, distinct vulnerability categories emerged: Category I (high priority) encapsulates WS7, WS4, WS9, and WS6. Most notably, WS6 endured a staggering 244 landslides and 7 flash-flood incidents. Category II (moderate attention) includes WS11, WS8, WS5, and WS12. To illustrate, WS12’s hydrological traits indicate potential vulnerabilities due to landslides. Category III (regular monitoring) comprises WS1, WS3, WS2, and WS10. For instance, WS10, although presently modeled as stable, exhibits 160 landslide incidents. Validation exercises through field surveys reinforced these computational conclusions, with the most impacted regions aligning with high-priority basins. However, it is essential to underscore that while this research presents a comprehensive view of the Alaknanda River basin, the methods’ inherent assumptions might not capture all real-world complexities. Even though the insights are tailored to a segment of the Indian Himalayan Region, the possibility of extrapolating them to other areas exists. Nevertheless, unforeseen anthropogenic and climatic deviations could alter the vulnerability landscape. These findings necessitate stakeholders to craft mitigation strategies that holistically cater to each watershed’s distinct attributes, championing ecosystem preservation and community well-being.