Remote sensing and GIS-based watershed prioritization for land and water conservation planning and management Open Access

An area that drains all the incoming rainwater as surface or subsurface runoff into a water body/reservoir is termed as a watershed (Chopra et al. 2005; Edwards et al. 2015). A watershed encompasses a complex of soils, landforms, land uses, and vegetation within the topographic boundary or water divide (Pareta & Pareta 2012). The runoff potential depends on the surface characteristics (geometry, network, texture, and relief aspects) of a catchment (Tripathi et al. 2005). Thus, hydrologic characterization becomes important through studying the morphometry of a watershed. Hydrologic characterization is the study of the features which affect the various components of a hydrologic cycle within the watershed boundary. Moreover, the hydrologic characterization is a way forward for prioritizing watersheds even in the absence of a soil map (Biswas et al. 1999), which is possible through morphometric analysis by finding out the several watershed characteristics (Tripathi et al. 2005; Rai et al. 2017c; Pande et al. 2018; Singh et al. 2021a, 2021b). Implementation of conservation measures is mainly dependent on the morphometry of a particular watershed (Rai et al. 2017c; Prakash et al. 2019). The quantification and scientific study of configuration of the Earth's surface, its shape, and landform dimensions is termed as morphometric analysis (Rai et al. 2017c, 2018). It is one of the key techniques to monitor and assess watershed response to variations in climate, drainage physiognomies (Rais & Javed 2014), flash flooding (Perucca & Angilieri 2010), and hydrological processes (Eze & Efiong 2010). It helps in identifying suitable locations for trenching, farm ponds, check dams, pits, and spillways (Gupta & Srivastava 2010; Srivastava et al. 2010). The features of watershed morphometry may be applied to designate geomorphic processes, namely, flood peak, erosion rate, and sediment yield (Patton & Baker 1976; Patton 1988; Singh et al. 2023). Moreover, the drainage physiognomies obtained from watershed morphometric analysis may provide evidence on the history of denudation, subsurface materials, soil type, vegetation status, and geological structure of a region, which in turn, encourages the optimum utilization of soil and water in a watershed for optimum crop production (Jawaharraj et al. 1998; Biswas et al. 1999).

The quantitative investigation of catchment morphometry helps to understand the drainage physiognomies, runoff occurrence, and infiltration rate of the land and aquifer recharge potential (Raj & Azeez 2012). The quantification of areal, linear, and relief aspects is required for an improved and systematic depiction of watershed geometry and its channels (Withanage et al. 2014). Morphometric analysis (Horton 1945) includes the measurement or computation of several parameters (linear, areal, and relief aspects) such as stream length and order, bifurcation ratio, length of overland flow, drainage density, form factor, stream frequency, circularity ratio, elongation ratio, texture ratio, compactness coefficient, and relief ratio (Nag & Chakraborty 2003). For studying the drainage pattern of a catchment, information related to geomorphology, geology, hydrology, and land use patterns can be highly reliable and informative (Binjolkar & Keshari 2007).

Geospatial technology is highly efficient for drainage pattern analysis, investigating spatial variability in the drainage features, prioritization of sub-watersheds, modeling water resources and managing floods (Vittala et al. 2004; Das & Mukherjee 2005; Miller & Kochel 2010; Bali et al. 2012). Geographical information system (GIS) is time efficient and capable to manage complex issues, as well as big database and retrieval (Vinothkumar et al. 2016). Remote sensing technology is used to provide LULC information by using digital image processing techniques (Yuksel et al. 2008). Thus, the increasing use of remote sensing and GIS technologies in geomorphologic mapping has enhanced the efficacy of landform dissection, quantification, and organization.

The geospatial technology-based watershed prioritization (based on morphometric analysis or principal component analysis (PCA) or LULC analysis or soil loss/sediment yield) is highly important for identifying the most critical watershed(s) and the major problems linked to catchment water spread, patterns of erosion, and aquifer recharge within a watershed (the highest degree of erosion is indicated by the highest priority). The geospatial tools help in preparing different maps (watershed (DEM), stream order, stream network, contour, watershed boundary, basin length/longest flow path, drainage, flow direction, flow accumulation, aspect, slope, hillshade, LULC, etc.) and identifying the potential zones for groundwater recharge in a catchment (Sreedevi et al. 2005; Patil & Mali 2013; Patil & Mohite 2014).

The formulae related to the morphometric parameters (basin geometry, stream network, drainage texture, and relief aspect) reported in this review were taken from several research papers published in the past and the original authors are cited.

The study of a stream network involves evaluation of the linear parameters of a catchment. It includes computation of stream order, length of a stream, mean stream length, stream length ratio, bifurcation ratio, basin length, length of overland flow, watershed perimeter, wandering ratio, fitness ratio, and sinuosity indices. Significant linear parameters have been discussed and their respective formulae are presented in Table 2.

Table 2

Linear morphometric parameters for drainage network analysis

Sr. no. .	Parameter .	Formulae/software .	Reference .	Unit of measurement .
1	Stream order ⁠)	Hierarchical rank (GIS analysis)	Strahler (1952, 1964)	–
2	Stream number (⁠⁠)	Total number of stream segments of order u (⁠⁠)/GIS analysis	Horton (1945)	–
3	Stream length	Length of the stream (km) of order u (⁠⁠)/GIS analysis	Horton (1945) and Strahler (1964)	km
4	Mainstream/channel length	GIS analysis	–	km
5	Mean stream length ⁠)		Horton (1945) and Strahler (1964)	km
6	Stream length ratio ⁠)		Horton (1945) and Strahler (1964)	–
7	Mean stream length ratio ⁠)	–	–	–
8	Weighted mean stream length ratio ⁠)	–	–	–
9	Bifurcation ratio ⁠)		Schumm (1956) and Strahler (1964)	–
10	Direct bifurcation ratio (⁠⁠)		–	–
11	Bifurcation index (Ib)		–	–
12	Mean bifurcation ratio ⁠)	Average of bifurcation ratios of all order	Strahler (1957, 1964)	–
13	Weighted mean bifurcation ratio ⁠)	–	Strahler (1953)	–
14	Rho coefficient ⁠)		Horton (1945)	km

Sr. no. .	Parameter .	Formulae/software .	Reference .	Unit of measurement .
1	Stream order ⁠)	Hierarchical rank (GIS analysis)	Strahler (1952, 1964)	–
2	Stream number (⁠⁠)	Total number of stream segments of order u (⁠⁠)/GIS analysis	Horton (1945)	–
3	Stream length	Length of the stream (km) of order u (⁠⁠)/GIS analysis	Horton (1945) and Strahler (1964)	km
4	Mainstream/channel length	GIS analysis	–	km
5	Mean stream length ⁠)		Horton (1945) and Strahler (1964)	km
6	Stream length ratio ⁠)		Horton (1945) and Strahler (1964)	–
7	Mean stream length ratio ⁠)	–	–	–
8	Weighted mean stream length ratio ⁠)	–	–	–
9	Bifurcation ratio ⁠)		Schumm (1956) and Strahler (1964)	–
10	Direct bifurcation ratio (⁠⁠)		–	–
11	Bifurcation index (Ib)		–	–
12	Mean bifurcation ratio ⁠)	Average of bifurcation ratios of all order	Strahler (1957, 1964)	–
13	Weighted mean bifurcation ratio ⁠)	–	Strahler (1953)	–
14	Rho coefficient ⁠)		Horton (1945)	km

View Large

Stream number is the total number of streams per stream order and can be calculated in GIS (Horton 1945). It has a direct relationship with the size of the watershed and channel measurements (Hajam et al. 2013). It is dependent upon geology, slope, soil type, vegetation, and climate (mainly rainfall) in a catchment (Sujatha et al. 2015). According to Horton (1945), N_u of each order creates an inverse geometric sequence with the order number. In general, N_u decreases in geometric progression with increasing stream order (Rao et al. 2010; Sreedevi et al. 2013; Pande & Moharir 2017). In other words, the stream order increases with a decrease in N_u at each successive stream order (Horton 1945). Generally, a logarithm plot of the stream number of a given order versus stream order forms a linear relationship (Horton 1945). The departure from its normal behavior specifies that the terrain has a higher relief, slope steepness, lithological variability, and possible uplift through the catchment (Singh & Singh 1997). The higher number of streams indicates impermeability nature/low infiltration, and ultimately, large surface runoff conditions. The higher number of streams of first order designates the probability of flash flooding after heavy rain (Chitra et al. 2011; Pande & Moharir 2017).

Stream length ratio is defined as a ratio between the mean stream length segment of order u (L_u) to the mean stream length segment of order u − 1 (L_u−1) (Horton 1945; Strahler 1964). It has an important link with the surface flow, discharge, and erosion stage of the basin (Horton 1945; Kanth & Hassan 2012). It is likely to be the same among the successive orders in a basin. However, it is variable for different orders in the sub-watersheds in a studied area. The variation in R_Lu between successive stream orders of a watershed arises because of slope and topographical differences (Sudheer 1986; Sreedevi 1999; Sreedevi et al. 2005).

Horton (1932) introduced this term or ratio (R_b), which is associated with the branching pattern of a drainage network (Schumm 1956). It is the ratio of the number of streams of order U and the number of streams of the next order (U + 1) (Schumm 1956; Strahler 1964). It depends on the physiography, slope, and climatic conditions of the watershed. It is a dimensionless property (ranging from 3.0 to 5.0). It illustrates the tectonic and geological features of a catchment (Gajbhiye et al. 2014). Strahler (1957) stated that R_b indicates a small variation for environmentally different areas, except wherever the geological control dominates. Later, it has been reported that R_b is not the same from one order to the next order and these anomalies depend on the geologic and lithologic development of a basin (Chow 1964; Strahler 1964; Singh et al. 2014). According to Horton (1945) and Strahler (1968), R_b ranges from 2 to 5 for a well-established drainage network. The value of R_b < 2.0 indicates a flat or rolling drainage basin, whereas the R_b value of 3–4 represents highly stable and dissected drainage basins (Horton 1945). According to Strahler (1964) and Nautiyal (1994), higher R_b values indicate more structural disturbances, which can be effectively controlled by the watershed shape, showing smaller variations between 3 and 5 in homogeneous bedrocks. According to Strahler (1964) and Schumm (1956), the R_b value in the range of 3–5 does not have a dominant effect on the drainage configuration. The smaller R_b values indicate watersheds with less or partial structural disturbances (Strahler 1964; Nag 1998) with no distortion in drainage pattern because of geological control (Schumm 1956; Chow 1964; Nautiyal 1994; Chopra et al. 2005). A lower value of R_b specifies the higher infiltration rate and lesser number of stream segments in the watershed. With the increase in R_b, the possibility of flood damage increases (McCullagh 1978). The higher R_b value indicates more soil erosion in relation to the severe overland flow and lower recharge potential in the watersheds. An elongated basin has a high R_b value, whereas a circular basin has a low R_b value. A relatively higher R_b value shows an early hydrograph peak with a possibility of flash flood during rain events (Hajam et al. 2013).

Direct bifurcation ratio is the ratio of the number of fluvial segments of a given order (N_fu) to the number of segments of the next order (N_fu + 1) (Guarnieri & Pirrotta 2008). It is the measure of the degree of branching within the hydrographical network without taking the hierarchical anomaly into account (Horton 1945; Strahler 1952).

Mean bifurcation ratio can be computed by averaging the bifurcation ratio of all orders (Strahler 1957, 1964). The lower R_mb value indicates the influence of geomorphology on watershed developments. Normally, for a basin, R_mb ranges from 3.0 to 5.0 under the negligible effect of geological structures on the drainage networks (Verstappen 1983a). The relatively lower value of R_mb offers heterogeneous geology, higher permeability, and decreased structural stability in a catchment (Hajam et al. 2013).

Weighted mean bifurcation ratio is the product of the bifurcation ratio for consecutive pairs of stream orders with the total stream segments involved, and taking the mean of the values so obtained (Strahler 1953, 1957, 1964; Magesh et al. 2012; Rai et al. 2017a, 2017b, 2018). Strahler (1953) proposed the concept of R_wmb to determine a single value as in most of the watershed areas, R_b varies from one order to another. The lower value of R_wmb indicates a normal pattern of the basin with no structural disturbances (Singh et al. 2013).

Rho coefficient is defined as the ratio between the mean stream length and the bifurcation ratio (Horton 1945). It helps to determine the connection between D_d and the physiographical advancement of the catchment. It is used to measure the storage capability of drainage networks (Horton 1945). Hence, it is an indicator of the degree of drainage development in a watershed.

The basic parameters of basin geometry, namely, watershed area (A_w), basin perimeter (P), basin length (L_b), longest flow path (L_fp), basin centroid longest flow path (L_cfp), mainstream/channel length (L_ms/L_mc), valley length (L_v), and minimum areal distance are computed using GIS software (ArcGIS or QGIS). The behavior of the parameters of basin geometry on watershed hydrology is explained below and their respective mathematical formulae are listed in Table 3.

Table 3

Morphometric parameters for the analysis of basin geometry

Sr. no. .	Parameter .	Formulae/software .	Reference .	Unit of measurement .
1	Watershed/basin area	GIS analysis	Schumm (1956)	km2
2	Basin length	GIS analysis	Schumm (1956) and Horton (1932)	km
3	Longest dimension parallel to principle drainage line (Llp)	GIS analysis	Schumm (1956)	km
4	Longest flow path	GIS analysis	Schumm (1956)	km
5	Basin centroid longest flow path	GIS analysis	Schumm (1956)	km
6	Valley length	GIS analysis	Schumm (1956)	km
7	Perimeter	GIS analysis	Schumm (1956)	km
8	Relative perimeter (⁠		Schumm (1956)	km
9	Mean basin width (⁠⁠)		Schumm (1956)	km
10	Minimum areal distance	GIS analysis	Mueller (1968)	km
11	Down valley distance	GIS analysis	Mueller (1968)	km
12	Texture ratio (⁠		Horton (1945) and Schumm (1956)	km−1
13	Fitness ratio	or	Melton (1957)	–
14	Elongation ratio	or	Horton (1945), Miller (1953), and Schumm (1956)	–
15	Ellipticity index (⁠⁠)		–	–
16	Wandering ratio (⁠⁠)		Smart & Surkan (1967)	–
17	Watershed eccentricity (⁠⁠)		Black (1972)	–
18	Circularity ratio	or	Miller (1953) and Strahler (1964)	–
19	Form factor		Horton (1932, 1945)	–
20	Compactness coefficient (⁠	or	Horton (1945) and Gravelius (1914)	–
21	Shape factor (⁠		Horton (1932, 1945)	–
22	Lemniscate ratio (K)		Chorley et al. (1957)	–
23	Channel sinuosity (⁠		Le Roux (1992)	–
24	Standard sinuosity index (⁠⁠)		Mueller (1968)	–
25	Hydraulic sinuosity index (⁠⁠)		Mueller (1968)	–
26	Topographic sinuosity index (⁠⁠)		Mueller (1968)	–
27	Channel index ⁠)		Miller (1953) and Mueller (1968)	–
28	Valley index ⁠)		Miller (1953) and Mueller (1968)	–
29	Hypsometric integrals (⁠⁠)		Pike & Wilson (1971)	–
30	Time of concentration		Kirpich (1940)	h
31	Time of recession		Mustafa & Yusuf (2012)	h

Sr. no. .	Parameter .	Formulae/software .	Reference .	Unit of measurement .
1	Watershed/basin area	GIS analysis	Schumm (1956)	km2
2	Basin length	GIS analysis	Schumm (1956) and Horton (1932)	km
3	Longest dimension parallel to principle drainage line (Llp)	GIS analysis	Schumm (1956)	km
4	Longest flow path	GIS analysis	Schumm (1956)	km
5	Basin centroid longest flow path	GIS analysis	Schumm (1956)	km
6	Valley length	GIS analysis	Schumm (1956)	km
7	Perimeter	GIS analysis	Schumm (1956)	km
8	Relative perimeter (⁠		Schumm (1956)	km
9	Mean basin width (⁠⁠)		Schumm (1956)	km
10	Minimum areal distance	GIS analysis	Mueller (1968)	km
11	Down valley distance	GIS analysis	Mueller (1968)	km
12	Texture ratio (⁠		Horton (1945) and Schumm (1956)	km−1
13	Fitness ratio	or	Melton (1957)	–
14	Elongation ratio	or	Horton (1945), Miller (1953), and Schumm (1956)	–
15	Ellipticity index (⁠⁠)		–	–
16	Wandering ratio (⁠⁠)		Smart & Surkan (1967)	–
17	Watershed eccentricity (⁠⁠)		Black (1972)	–
18	Circularity ratio	or	Miller (1953) and Strahler (1964)	–
19	Form factor		Horton (1932, 1945)	–
20	Compactness coefficient (⁠	or	Horton (1945) and Gravelius (1914)	–
21	Shape factor (⁠		Horton (1932, 1945)	–
22	Lemniscate ratio (K)		Chorley et al. (1957)	–
23	Channel sinuosity (⁠		Le Roux (1992)	–
24	Standard sinuosity index (⁠⁠)		Mueller (1968)	–
25	Hydraulic sinuosity index (⁠⁠)		Mueller (1968)	–
26	Topographic sinuosity index (⁠⁠)		Mueller (1968)	–
27	Channel index ⁠)		Miller (1953) and Mueller (1968)	–
28	Valley index ⁠)		Miller (1953) and Mueller (1968)	–
29	Hypsometric integrals (⁠⁠)		Pike & Wilson (1971)	–
30	Time of concentration		Kirpich (1940)	h
31	Time of recession		Mustafa & Yusuf (2012)	h

View Large

The basin area can be computed using GIS software (e.g. ArcGIS or QGIS). It is calculated as the total area projected on a flat plane contributing to gathering all orders of a basin. It is one of the important basic parameters to study watershed morphometry and has a strong association with the mean annual runoff. Hydrologically, it is significant as it affects the size of the rainstorm hydrograph, mean runoff, and the magnitudes of the peak (Rao et al. 2010).

Watershed/mean basin width is the ratio between the catchment area and the length of the basin. The effective or mean basin width (W_mb) has a strong relationship with the stream ordering of the catchment, as it is directly related to A_w and inversely to basin length.

Based on the data from the published literature, supplemented with satellite imagery measurement, Downing et al. (2012) reported a strong relationship between stream order and the mean width of rivers around the world.

Different authors have given different meanings for L_b (Schumm 1956; Gregory & Walling 1973; Cannon 1976; Gardiner 1978). Gregory & Walling (1973) have defined L_b as the length of the longest path from the head waters to the confluence point. Schumm (1956) has defined L_b as the lengthiest measurement parallel to the mainstream line in a watershed. Likewise, Gardiner (1978) has defined L_b as the distance between a basin mouth and a point on its periphery, which is halfway from the mouth. It determines the shape of the basin. The higher basin length indicates an elongated basin. Downing et al. (2012) had reported a relation of world stream area and length with stream order.

Watershed/basin perimeter is the length of the basin boundary measured beside the divides among the neighboring watersheds (Ahmed et al. 2010). It indicates the size and shape of a catchment. It can be computed using GIS software such as QGIS or ArcGIS.

Therefore, L_mc = 3 × π/2 and L_b = 1 + 1 + 1 = 3 units.

Now, L_mc: L_b = π/2 = 1.57 (approx.)

For determining the sinuosity parameter, a river network can be divided into several stream segments as suggested by Mueller (1968). For the computation of C_i and V_i, the measurement of the length of the channel, length of the valley, and the smallest distance between the mouth and the source of a river is done.

DEM dataset can be used for preparing a flow direction map of a basin or watershed. The flow of water takes place toward the steepest downhill slope. Every pixel in the DEM dataset is potentially enclosed by eight adjacent pixels. The elevation difference is divided by the center-to-center spacing between the water flow directions.

Form factor is a ratio between the catchment area and the square of the catchment length (Horton 1932, 1945). F_f designates the flow intensity in a catchment and its shape. For a basin having an exact circular shape, F_f is greater than 0.79 (Rekha et al. 2011; Gajbhiye et al. 2014). According to Vinutha & Janardhana (2014) and Rai et al. (2017c), F_f < 0.78 and F_f > 0.78 designate an elongated and circular shape of the watershed, respectively. A watershed with a high F_f value achieves a greater peak (hydrograph) for a smaller period, whereas a watershed with a smaller value of F_f can have water flow for an extended period with a flatter or lower peak (Nautiyal 1994; Waikar & Nilawar 2014), and can be managed with no trouble as compared to circular-shaped basins (Reddy et al. 2002). F_f is related to many other parameters of catchment geometry, drainage texture, and relief aspects, as reported below.

Wandering ratio can be expressed as the ratio of the length of the mainstream to the basin length or length of the valley (Melton 1957; Pareta & Pareta 2011; Hajam et al. 2013). The mathematical expression for this parameter is given in Table 3.

The distance between the outlet of a basin and the remotest point on the ridge is termed as the valley length (Pareta & Pareta 2011, 2012).

The ratio between the channel length and the longest flow path is termed as channel sinuosity (Le Roux 1992; Pareta & Pareta 2011, 2012). It describes the stream/channel pattern in a basin. Usually, C_s varies from 1 to 4 or more. The standard sinuosity index (I_ss) is an important quantifiable index for understanding the importance of stream segments in landscape evolution. It is quite important for hydrologists, geologists, and geomorphologists. For computing this parameter, a channel is divided into several stream segments (Mueller 1968). Rivers having C_s or I_s < 1.5 are termed as straight or sinuous, whereas, rivers having C_s or I_s ≥ 1.5 are termed as meandering (Leopold et al. 1964).

For the measurement of S_i, Mueller (1968) proposed a sinuosity index in terms of topographic and hydraulic sinuosity indices (I_tsi and I_hsi), measured as the ratio between the channel lengths (L_c), valley lengths (L_v), and their distance between mouth and source of the river (L_a). Both I_tsi and I_hsi are important parameters to define the stages of river basin development and control the sinuosity aspects. During the youth stage, I_tsi (>60%) is greater than I_hsi. While in mature, late mature, and old stages, I_hsi (>60%) dominates I_tsi due to loss in relief by erosion factors.

Elongation ratio can be expressed as a ratio of the diameter of a circle having a similar area as the basin to the basin length (Horton 1945; Miller 1953; Schumm 1956). It is a significant parameter to investigate the basin shape (Strahler 1968). In general, the R_e value varies from 0.6 to 1.0 in relation to the widespread disparity in climatological and geological properties (Strahler 1964). The lower value of R_e indicates an elongated basin with a steep slope and higher vulnerability to soil erosion and sedimentation (Reddy et al. 2002). R_e value closer to 1.0 indicates a quite lower relief. Whereas, the R_e value in the range of 0.6–0.8 is associated with higher relief and moderate-to-steep land slope (Strahler 1964). According to Pareta & Pareta (2011), the shape of a watershed is highly elongated (R_e< 0.5), elongated (R_e = 0.5–0.7), less elongated (R_e = 0.7–0.8), oval (R_e = 0.8–0.9), and circular (R_e = 0.9–0.10).

Circulatory ratio can be expressed as a ratio between the watershed area and the area of a circle having the same perimeter as that of the watershed (Miller 1953; Strahler 1964). It is mainly affected by stream length, drainage frequency, land condition, climatological variation, LULC, slope steepness, and relief in a basin (Das et al. 2012; Patel et al. 2013). According to Miller (1953), R_c ranges from 0.40 to 0.50, indicating highly elongated, extremely permeable, and homogeneous geologic materials (Pareta & Pareta 2011, 2012; Singh et al. 2014). The R_c values of 0.0 and 1.0 designate highly elongated and circular shapes, respectively (Sreedevi et al. 2013). R_c is related to many other morphometric parameters (areal, linear, and relief aspects).

Texture ratio can be expressed as a ratio of the sum of first-order streams to the basin perimeter (Horton 1945; Schumm 1956), which is one of the important parameters for drainage morphometric analysis (Altaf et al. 2013) used to indicate the existence of higher numbers of first order streams in the catchment (i.e. a variation in topology). It depends on lithological characteristics (e.g. soil type), infiltration rate, and relief parameters of the basin (Vijith & Satheesh 2006). Generally, the smaller R_t values indicate a plain basin with fewer slope variations. Whereas the higher value of R_t indicates a low infiltration rate and higher runoff. Gupta et al. (2019) categorized R_t as very high (>6.0), high (5.0–6.0), moderately high (4.0–5.0), moderate (3.0–4.0), and low (<3.0).

Compactness coefficient can be expressed as a ratio between the catchment/basin perimeter and the perimeter of a circular/round area, which equals the watershed area (Gravelius 1914; Horton 1945; Pareta & Pareta 2011). It is reciprocal of the circularity ratio which expresses runoff characteristics in the basin. It is independent of the watershed size and highly dependent on watershed relief (Rai et al. 2017a, 2017b). Its smaller value specifies higher runoff and erodibility (i.e. the lower value of C_c is more prone to erosion). The C_c value equal to 1.0 designates a fully circular-shaped catchment, while the C_c value greater than 1.0 indicate a significant deviation from the circular shape of the basin. In case of a circular basin, the drainage yields the shortest T_c before the occurrence of the peak in the basin (Ratnam et al. 2005; Javed et al. 2009).

The parameters of drainage texture analysis are discussed under this head and their respective mathematical formulae are presented in Table 5.

Table 5

Morphometric parameters for drainage texture analysis

Sr. no. .	Parameter .	Formulae/software .	Reference .	Unit of measurement .
1	Drainage density		Horton (1932, 1945)	km/km2
2	Drainage intensity (⁠		Faniran (1968)	km−1
3	Drainage texture		Horton (1945) and Smith (1950)	km−1
4	Drainage pattern (⁠	–	Horton (1932)	–
5	Infiltration number (⁠		Faniran (1968) and Umrikar (2016)	km−3
6	Stream frequency		Horton (1932, 1945)	km−2
7	Length of overland flow (⁠	or	Horton (1945) and Langbein & Leopold (1964)	km
8	Constant of channel maintenance (⁠		Horton (1945), Strahler (1952), and Schumm (1956)	km2/km

Sr. no. .	Parameter .	Formulae/software .	Reference .	Unit of measurement .
1	Drainage density		Horton (1932, 1945)	km/km2
2	Drainage intensity (⁠		Faniran (1968)	km−1
3	Drainage texture		Horton (1945) and Smith (1950)	km−1
4	Drainage pattern (⁠	–	Horton (1932)	–
5	Infiltration number (⁠		Faniran (1968) and Umrikar (2016)	km−3
6	Stream frequency		Horton (1932, 1945)	km−2
7	Length of overland flow (⁠	or	Horton (1945) and Langbein & Leopold (1964)	km
8	Constant of channel maintenance (⁠		Horton (1945), Strahler (1952), and Schumm (1956)	km2/km

View Large

Drainage density is expressed as the ratio of the total stream length to the watershed area (Horton 1932, 1945; Strahler 1964). It is correlated to the peak discharge time in a basin (Wilford et al. 2004) and indicates the closeness of channel spacing (Ahmed et al. 2010). It is dependent on climate, lithology, structural characteristics, surface runoff, slope, permeability, and vegetation cover of the basin. It provides the link between lengthwise operating processes in the stream course and basin attributes (Gregory & Walling 1973). It helps in the quantitative analysis of a watershed (Strahler 1964). It is mainly affected by soil infiltration capacity, resistance to erosion of rocks, and climatic conditions (Verstappen 1983a, 1983b). For humid regions, it varies from 0.55 to 2.09 km/km² (Langbein 1947). It is one of the significant indicators of various landform elements and offers a numerical assessment of landscape segmentation and overflow or runoff potential (Chorley 1969). Several studies (Smith 1950; Vittala et al. 2004; Chandrashekar et al. 2015; Tavassol & Gopalakrishna 2016) have categorized D_d as very coarse/highly permeable (<2.0), coarse/permeable (2.0–4.0), moderately coarse/moderately permeable (4.0–6.0), fine/impermeable (6.0–8.0), and very fine/highly impermeable (>8.0). The low D_d value (<6.0) indicates coarse texture (Ahmed et al. 2010; Ramaiah et al. 2012) with a poor drainage system and slower response to hydrologic conditions (Hajam et al. 2013). Typically, the low D_d value arises in regions that contain gneiss, granite, and schist (Rao et al. 2010). On the other hand, the high D_d value indicates fine texture of the drainage basin (Ahmed et al. 2010; Ramaiah et al. 2012; Singh et al. 2013), poor land permeability, steeper slope, and limited vegetative cover that contribute to large runoff in a watershed. Such drainage basins are vulnerable to flood risk, as a result of rapid runoff in the channels. Moreover, the high D_d leads to a highly dissected drainage basin with a relatively faster hydrological response to rainfall events (Hajam et al. 2013).

Drainage texture can be expressed as a ratio between the total number of stream segments and the basin perimeter (Horton 1945; Smith 1950). It is an important concept of geomorphology related to the relative spacing of stream/drainage lines. It is dependent on the climate, rainfall, vegetation, subsurface lithology, infiltration rate, and relief aspects of the basin (Sreedevi et al. 2013). Smith (1950) and Tavassol & Gopalakrishna (2016) classified the drainage texture into five categories, namely, highly coarse (D_t < 2), coarse (D_t = 2–4), moderately coarse (D_t = 4–6), fine (D_t = 6–8), and very fine (D_t > 8). The higher number of stream segments and D_t value indicate impermeability of the basin. The coarse drainage texture indicates a flatter peak flow for an extended duration.

Constant of channel maintenance is expressed as the reciprocal of drainage density (Horton 1945; Strahler 1952; Schumm 1956). According to Schumm (1956), it helps to find out the minimum area needed for developing a drainage channel having a length of 1 km. It designates the relative size of the landform units in a catchment (Strahler 1957). The lower C_cm values indicate structural instability in the catchment having lower permeability and higher runoff conditions.

Drainage intensity is expressed as a ratio between stream frequency and drainage density (Faniran 1968). The low values of D_i in relation to lower D_d and F_s indicate that the runoff cannot be removed easily from a watershed or sub-watershed, which makes it highly vulnerable to flood risk, gully erosion, and landslides (Rai et al. 2018).

Infiltration number can be expressed as a product of stream frequency and drainage density (Faniran 1968; Umrikar 2016). It forms an inverse relation with the infiltration rate. The I_n value provides information about the infiltration characteristics and reveals impermeable bedrock and high relief areas in the watershed (Umrikar 2016). A higher I_n value indicates a low infiltration rate, and vice versa.

The study of drainage patterns can help to identify the stages in an erosion cycle in a basin. It reflects the effect of gradient, lithology, and structure. It presents the characteristics of a drainage basin in terms of D_t (which reflects the effects of climate, permeability, vegetative cover, relief ratio, etc.). Through the study of D_p, basin geology, the strike and dip of rocks depositing, and information related to a geological structure can be easily deduced. Howard (1967) has related stream/drainage patterns to geological information. According to Horton (1932), the drainage pattern in a terrain is termed as dendritic, where the bedrock is uniform in nature.

Length of overland flow can be expressed as half the inverse of drainage density (Horton 1945; Langbein & Leopold 1964). It is inversely related to the mean channel slope and is fairly identical to the sheet flow length. It is the surface flow length of water before it confines into well-defined stream channels (Horton 1945) because of the incapability of water to penetrate into the soil, either due to poor infiltration rate or high rainfall intensity. It is one of the significant self-regulating factors which affect both physiographic and hydrologic development of the basin (Horton 1932). It is considerably influenced by temporal and spatial variations in infiltration rates (Kanth & Hassan 2012). Chandrashekar et al. (2015) have categorized L_o into three groups, namely, high (>0.3), moderate (0.2–0.3), and low (<0.2). The smaller the L_o value, the greater the surface runoff entering the streams. A significant quantity of surface runoff can be contributed to stream discharge, even by a small rainfall in a relatively even topography (Rao 1978; Muthukrishnan et al. 2013).

The assessment of relief parameters is extremely significant to monitor the hydrologic response of a watershed. The relief parameters are discussed in this section and their respective mathematical expressions are given in Table 6.

Table 6

Relief morphometric parameters

Sr. no. .	Parameter .	Formulae/software .	Reference .	Unit of measurement .
1	Minimum height (elevation) of basin (⁠	GIS analysis/DEM	–	m
2	Maximum height (elevation) of basin (⁠	GIS analysis/DEM	–	m
3	Contour interval (⁠⁠)	GIS analysis	Magesh et al. (2012)	m
4	Total contour length (⁠⁠)	GIS analysis	Magesh et al. (2012)	km
5	Basin or watershed slope	or	–	–
6	Slope analysis (⁠⁠)	GIS analysis/DEM	–	–
7	Average slope (⁠⁠)		Wentworth (1930)	–
8	Mean slope ratio (⁠⁠)		Wentworth (1930)	–
9	Mean slope of the overall basin (⁠⁠)			–
10	Basin relief or maximum watershed relief		Strahler (1952, 1957) and Schumm (1956)	m
11	Absolute relief (⁠⁠)			m
12	Relative relief	⁠) × 100 or (%)	Melton (1957)	–
13	Relief ratio	or	Schumm (1956, 1963)	–
14	Gradient ratio		Sreedevi et al. (2005)	–
15	Ruggedness number	or	Strahler (1964) and Patton & Baker (1976)	–
16	Melton ruggedness number	or	Melton (1957, 1965)	–
17	Dissection index		Singh & Dubey (1994)
18	Channel gradient (Cg)	or	Prasad et al. (2008)	–

Sr. no. .	Parameter .	Formulae/software .	Reference .	Unit of measurement .
1	Minimum height (elevation) of basin (⁠	GIS analysis/DEM	–	m
2	Maximum height (elevation) of basin (⁠	GIS analysis/DEM	–	m
3	Contour interval (⁠⁠)	GIS analysis	Magesh et al. (2012)	m
4	Total contour length (⁠⁠)	GIS analysis	Magesh et al. (2012)	km
5	Basin or watershed slope	or	–	–
6	Slope analysis (⁠⁠)	GIS analysis/DEM	–	–
7	Average slope (⁠⁠)		Wentworth (1930)	–
8	Mean slope ratio (⁠⁠)		Wentworth (1930)	–
9	Mean slope of the overall basin (⁠⁠)			–
10	Basin relief or maximum watershed relief		Strahler (1952, 1957) and Schumm (1956)	m
11	Absolute relief (⁠⁠)			m
12	Relative relief	⁠) × 100 or (%)	Melton (1957)	–
13	Relief ratio	or	Schumm (1956, 1963)	–
14	Gradient ratio		Sreedevi et al. (2005)	–
15	Ruggedness number	or	Strahler (1964) and Patton & Baker (1976)	–
16	Melton ruggedness number	or	Melton (1957, 1965)	–
17	Dissection index		Singh & Dubey (1994)
18	Channel gradient (Cg)	or	Prasad et al. (2008)	–

View Large

Absolute relief is expressed as an elevation difference between a known location in a watershed and the sea level (Pareta & Pareta 2011, 2012; Selvan et al. 2011). It is the maximum elevation of a known location in the river basin. B_ar can be computed using GIS software.

Relative relief is expressed as the ratio of basin relief to the basin perimeter (Melton 1957). It is also known as amplitude of relief or local relief, indicating the variation in altitude per unit area with respect to its local base level. Mathematically, it has been expressed in Table 6.

Relief ratio is expressed as a ratio between basin relief and basin length (Schumm 1956, 1963). In other words, it is a ratio between basin relief and the lengthiest dimension of the basin parallel to the principal drainage line. In general, it increases with the decrease in drainage/stream area and the size of the watershed. It indicates the overall slope of the watershed area and helps to relate the steepness and erosion in the basin. Several studies have reported a close correlation of sediment loss per unit area with R_r (Schumm 1954, 1956; Ahmed et al. 2010; Rai et al. 2018). The higher values of R_r indicate the steepness (high slope) (Vittala et al. 2004) and high relief of the basin. In the steeper sloped basins, the runoff enhances the possibility of erosion. On the other hand, the lower R_r value indicates a low degree of slope (steepness) and low relief.

Gradient ratio is one of the significant indicators of channel gradient, which allows an evaluation of the runoff generated on a volumetric basis (Sreedevi 1999; Sreedevi et al. 2005; Rai et al. 2017a, 2017b, 2018). The expression for this parameter is given in Table 6.

The total drop or fall in elevation from the source to the mouth of a basin is termed as channel gradient. The greatest drop can be recorded in hilly regions occupied by lower stream orders and the lowest in the plains dominated by higher stream orders, which depicts hilly regions as more efficient in down cutting in higher altitudes, whereas stream widening with braided channels in lower altitudes. It indicates that the mean slope of the channel decreases with the increase in stream order. This validates Horton's Law of stream slopes, which states a definite association among the stream slopes and their respective orders, which can be described through inverse geometric series law.

From Equation (18), the inverse of m gives the angular value of the slope.

Ruggedness number can be computed as the product of basin relief and drainage density (Strahler 1964, 1968; Patton & Baker 1976). It connects the steepness of the slope with its length (Rai et al. 2017a, 2017b). An extremely high value of R_n occurs when both of these variables are large under steep slopes (Umrikar 2016) and high D_d values. Such basins are found to be susceptible to the risk of flooding.

Melton ruggedness number is expressed as the ratio between basin relief and square root of the basin area (Melton 1965). It is one of the slope indices that give a specific representation of relief-ruggedness within the basin (Melton 1965; Rai et al. 2018). Wilford et al. (2004) have classified a watershed as debris flow, debris flood, and flood hazard watershed. According to this grouping, sub-watersheds are termed as debris flood basins, where a significant amount of sediments can be deposited by stream segments beyond the channel on the fan.

Lemniscate ratio can be expressed as a ratio between the square of basin length and basin area (Chorley et al. 1957). It is important to define the slope of a basin. The high K value indicates a large number of higher order streams in the basin, indicating higher soil erosion. According to Chorley et al. (1957), a basin is circular in shape for K < 0.6, oval for K value between 0.6 and 0.9, and elongated in shape for K > 0.9.

Dissection index is expressed as a ratio between relative relief and watershed absolute relief. It indicates the degree of vertical erosion or dissection and illustrates the stages of landscape development in a watershed (Singh & Dubey 1994; Thornbury 1969; Sarma et al. 2013). It always varies between 0.0 (complete absence of dissection) and 1.0 (extreme case, vertical cliff at the sea shore).

Multiple relationships have been formed between different morphometric parameters of a watershed and are expressed in Tables 7 and 8. The values of some important morphometric parameters along with their categorization are given in Table 9.

Table 7

Circulatory ratio ()	Constant of channel maintenance ()	Fitness ratio ()
or		–
	Length of overland flow ()	Texture ratio ()
		–
Compactness coefficient ()	Form factor ()	Shape factor ()
or
	–	–

Circulatory ratio ()	Constant of channel maintenance ()	Fitness ratio ()
or		–
	Length of overland flow ()	Texture ratio ()
		–
Compactness coefficient ()	Form factor ()	Shape factor ()
or
	–	–

View Large

Table 8

Leminiscate ratio ()
Basin relief ()	Gradient ratio ()	Melton ruggedness ratio (
	Ruggedness number ()
Relative relief ()

Leminiscate ratio ()
Basin relief ()	Gradient ratio ()	Melton ruggedness ratio (
	Ruggedness number ()
Relative relief ()

View Large

Remote sensing is a process of identifying and observing the features of a region (usually from satellite or aircraft) through measurement of its reflected and emitted radiations at a distance. GIS is used for integrating and analyzing geospatial data from several sources, including GPS data and satellite imageries. GIS includes data visualizations by the development of interactive maps, representation of various location-based variables or features, identification of spatial and temporal trends and patterns in datasets, and applications of tools and services in user-friendly interfaces. Remote sensing and GIS techniques can be effectively used for studying the drainage pattern of a watershed, sub-watershed prioritization, water resources modeling, and flood management (Miller & Kochel 2010; Bali et al. 2012). GIS applications are capable, time-efficient, and appropriate for 3D planning in relation to their capabilities to handle complex problems, as well as large datasets and retrieval (Kumar et al. 2016). To develop an environmental model, GIS can be efficiently used in relation to its higher data storage capacity, display, management, and analysis (Burrough & McDonnell 1998). Remote sensing provides the LULC information of a watershed by a digital image processing method (Yuksel et al. 2008). In the past, several authors have used remote sensing and GIS for watershed prioritization (Wolka et al. 2015; Tavassol & Gopalakrishna 2016; Abebe et al. 2018; Malik et al. 2019; Singh et al. 2021b, 2023). Wolka et al. (2015) identified the erosion risk areas of Cheleka wetland situated in the Central Rift Valley of Ethiopia with Revised Universal Soil Loss Equation (RUSLE). Tavassol & Gopalakrishna (2016) investigated the morphometry of four sub-watersheds in Tehran–Karaj plain using geospatial techniques. Abebe et al. (2018) studied the morphometry of Hawassa Lake in the Ethiopian Rift Valley using advanced global positioning system (GPS), geostatistics, portable sonar sounder technology, and geospatial techniques. Malik et al. (2019) prioritized 14 sub-watersheds of Naula watershed located in Uttarakhand, India. Singh et al. (2021b) prioritized the six reservoir catchments located in the Kandi region of Indian Punjab. Most recently, Singh et al. (2023) prioritized the 25 sub-watersheds located in the Banas basin of Rajasthan, India through integration of the RUSLE model in the GIS platform.

It is a programming language and environment for statistical computation and graphics. It provides a wide range of statistical (linear and non-linear modeling, statistical testing, cataloging, clustering, time-series analysis, etc.) and graphical methods (Anon 2023a). It is an integrated set of packages for data management, computation, and graphical display. The syntax of R mostly consists of three items, namely, variables (for data storage), comments (for improving the code readability), and keywords (words having a special meaning for the compiler).

IBM SPSS Statistics is a powerful statistical platform offering data analysis and machine learning solutions (Anon 2023b). It is quite easy to use, flexible, and scalable having a robust set of features. It provides new opportunities, minimizes risk, and improves efficiency. It can be applied to projects of all dimensions and levels of complexities. The highly developed statistical procedures of this software help to ensure high precision and quality decision making. Different agencies apply the SPSS platform for data preparation and discovery, forecasting/prediction, model management, and operations.

XLSTAT is a complete set of solutions for data analysis. It helps to perform correlation and regression analysis, as well as parametric and non-parametric tests (Anon 2023d; Singh et al. 2022). In this, data can be visualized by scatter plots, histograms, probability plots, 2D plots, error bars, motion charts, and univariate plots. Analyses of variance and covariance can be performed in several ways. A multivariate analysis of variance may also be carried out to model a combination of dependent variables. Using this software, binary, ordinary, or ordinal data can be modeled for logistic regression. It is also used for data mining using PCA, correspondence analysis, factor analysis, and discriminant analysis. This software package allows the relationship between two categorical variables to be summarized. While dealing with missing quantitative and qualitative data, it allows the use of techniques such as mean imputation, Markov Chain Monte Carlo multiple imputation algorithms, the nearest neighbor approach, non-linear iterative partial least squares algorithm, and mode imputation.

PCA is a technique of reducing the dimensionality of big datasets by transforming a big set of variables/parameters into a smaller one that still contains the majority of information in a large dataset (Sharma et al. 2014; Gajbhiye et al. 2015; Farhan et al. 2017; Meshram & Sharma 2017; Singh et al. 2021b). It is an unsupervised learning algorithm. The PCA algorithm is based on some mathematical concepts such as variance and covariance, as well as eigenvalues and eigen factors. The manual reduction of the dimensionality of large datasets is a cumbersome process, and PCA makes this process easier and time efficient (Meshram & Sharma 2017; Singh et al. 2021b). PCA involves standardization of continuous variables in the initial dataset, computation of covariance matrix for identifying correlations, computation of eigen vectors and eigen values of the covariance matrix for identifying the principal components, creation of feature vector for deciding principal components to be kept, and recasting of data along the principal component axes. Globally, several authors have applied PCA for watershed prioritization (Brown 1992; Pandzic & Trninic 1992; Mishra & Satyanarayana 1988; Bouvier et al. 2003; Gurmessa & Bárdossy 2009; Farhan et al. 2017; Arefin et al. 2020).

For performing PCA on a dataset in statistical software such as SPSS, the data should pass some assumptions, namely, (1) the variables/parameters should be measured at a continuous interval; (2) a linear relationship should exist between all variables. However, this assumption may be relaxed to some extent; (3) sampling adequacy (enough sample size) should be there. The sampling adequacy can be detected using the Kaiser–Meyer–Olkin (KMO) measure for both the complete dataset and individual variables in SPSS software; (4) the data should be appropriate for data reduction. The data reduction can be done using Bartlett's test of sphericity in SPSS software; and (5) no significant outliers should be there. Outliers can be determined using SPSS statistics. KMO test is carried out to investigate the strength of the partial correlation among the different variables. The KMO value varies between 0.0 and 1.0. KMO values closer to 0.0 indicate huge partial correlations as compared to the sum of the correlations (Anon 2023c), those between 0.8 and 1.0 indicate adequate samplings (Anon 2023c), and below 0.5 indicate inadequacy in samplings.

The morphometric parameters play a significant role in watershed prioritization, as these parameters form direct or inverse relation with soil erodibility. The higher values of linear morphometric parameters indicate the higher soil erosion potential of an area due to their direct relation with erodibility (Ratnam et al. 2005; Singh et al. 2021b). Moreover, the smaller values of shape parameters indicate a higher soil erosion potential of the area due to their inverse relation with erodibility i.e. the lower the values of shape parameters, the higher the soil erosion potential (Singh et al. 2021b). Thus, watershed prioritization is highly needed for implementing land and water conservation practices in an area suffering from soil erosion. Watersheds can be prioritized on the basis of morphometric analysis, PCA, LULC, and soil loss/sediment yield (Figure 1). While prioritizing the sub-watersheds, ranks are assigned to all parameters based on their priority, and then, the compound parameter is computed to give the final priority ranking. The compound parameter is computed by adding the ranks of all selected parameters for each sub-watershed and then dividing by the total number of parameters.

On the basis of morphometric analysis, D_d, D_t, F_s, R_c, C_c, R_e, F_f, and R_b are termed as erosion risk assessment parameters (Biswas et al. 1999). The higher the values of linear parameters such as F_s, D_d, D_t, and R_b, the higher will be the erodibility, as these parameters have a direct relationship with erodibility. Thus, the highest value of linear parameters is assigned with rank 1 and the lowest value with the last rank. On the other hand, the shape parameters, namely, R_e, C_c, R_c, and F_f have an inverse relationship with erodibility (Ratnam et al. 2005), the lower the value (R_e, C_c, R_c, and F_f), the more the erodibility. Thus, rank 1 is assigned to the shape parameter with the lowest value and the last rank to the highest value. After ranking all the parameters, the compound parameter is computed. The compound parameter with the least value is assigned with the highest rank. The sub-watershed with the highest priority rank is considered as most affected and treated with land and water conservation measures (agronomic or engineering). However, morphometric analysis-based watershed prioritization is a time-consuming process (Singh et al. 2021b). Thus, a time-efficient technique such as PCA is required for watershed prioritization.

PCA helps to express the behavior and interactions among the key morphometric variables (most significant parameters) for the watersheds (Farhan et al. 2017). PCA is applied to the inter-correlation matrix for getting the first factor loading matrix. Afterward, the first factor loading matrix is rotated via orthogonal transformation for identifying the most important parameters to be used for further priority measurement of the watersheds. Most recently, Arefin et al. (2020) have efficiently applied PCA in watershed prioritization. In the PCA technique, the inter-correlation among several morphometric variables can be studied using different tools such as MATLAB, XLSTAT, SPSS, R, Minitab, and Q research software. The correlation coefficient values of >0.9, >0.75 and ≤0.90, and >0.6 and ≤0.75 can be taken as standard limits for representing strong, good, and moderate correlation, respectively (Meshram & Sharma 2017). However, the correlation values ≤0.6 can be considered for indicating a poor correlation. Numerous parameters indicate no association with the other parameters, which creates difficulty in finding the relationship between the selected variables (Meshram & Sharma 2017). Thus, PCA is applied to the inter-correlation matrix for the grouping of components. The most important or highly correlated parameters are identified and ranked (Sharma et al. 2014; Gajbhiye et al. 2015; Meshram & Sharma 2017; Singh et al. 2021b). Once the ranking is done, the ranking values of all parameters are added and divided by the number of parameters involved to compute the compound parameter. The compound parameter with the least value is assigned the highest rank. This technique simplifies the prioritization process with a significant saving in time taken to prioritize the watersheds (Jolliffe & Cadima 2016; Meshram & Sharma 2017; Singh et al. 2021b).

Based on LULC analysis, common land use categories, namely, water bodies (e.g. lake, dam, river, etc.), forest area, agricultural land, wasteland, built-up, pasture, plantation, marshy land, and scrubland, are considered for watershed prioritization. The change (negative for decrease and positive for increase) in area, in every land use category, is assigned a rank. Ultimately, the ranking under each land use category is done by computing the compound parameter/relative weight. The lowest value of the compound parameter/relative weight is assigned with the highest priority. The results of the morphometric study, PCA, and LULC change analysis can be compared to identify the watersheds with common priority ranking for implementing appropriate conservation measures or remedial treatments.

The geospatial technology-based study of watershed morphometry followed by prioritization would be extremely helpful for soil and water conservation perspectives through recognizing the crucial problems correlated to resource conservation. From such studies related to watershed prioritization, the recognized crucial catchment/watershed can be considered for building conservation structures such as contour trench, terracing, farm bund, water arresting trenches, gully plugs, stone bunding, and farm ponds. At higher altitudes/elevations where the building of these structures is not feasible, plantation may be done. The planning of appropriate best management practices like ridge and valley area treatment is possible if in-depth information of a catchment is available. Several river basin or watershed scale models such as RUSLE, SWAT, and WEPP can also be utilized for setting up and implementing the aforementioned structures for soil and water conservation in a catchment.

The present study presents the watershed prioritization techniques on the basis of morphometric analysis, PCA technique, and soil loss models. These are the ranking techniques for designating the priority of watersheds to be considered for the treatment and implementation of conservation measures. However, incomplete knowledge of the process of classifying hydrology and the selection of invalid models for soil loss estimation may be the limitations of these methods. In the future, robust methods such as weighted index, correlation analysis, and influence analysis should be used for watershed/catchment prioritization for the planning and implementation of conservation measures. Multi-criteria decision-making (MCDM) techniques such as analytical hierarchy process, weights of evidence, analytical network process, and evidential belief function may be used. But, these techniques are biased, lengthy, and have time-consuming computational procedures. To overcome these issues, fuzzy logic based machine learning algorithms may be coupled with MCDM techniques. Thus, fuzzy logic and AHP-based land and water conservation (SWPC) models should be developed in the future.

All morphometric parameters are important for the hydrologic characterization of a watershed and sub-watershed prioritization. However, some of them are highly important in relation to the inter-correlation (strong either positive or negative, good and moderately good) among them for a better understanding of the watershed morphometry as well as their response to the rainfall-runoff conditions. This study reveals that the lower shape parameter values (i.e. F_f < 0.78, R_e < 0.8, and R_c < 0.50) indicate a basin having an elongated shape and flatter peak for an extended period and higher erodibility due to their inverse relationship with erodibility. Whereas F_f > 0.78, R_e > 0.8, R_c > 0.50, and C_c ≥ 1 indicate a basin having a circular shape and a higher peak for a smaller period, and the drainage yields the shortest T_c before the occurrence of peak flow. A relatively higher R_b value designates an early-peak hydrograph with a huge possibility of flash flood during rainstorm events. The runoff conditions and erosion potential of a watershed can be better expressed in terms of R_c, R_t, C_c, R_b, D_d, D_t, F_s, C_cm, D_i, I_n, L_o, and B_r. The higher values of N_u, R_t, D_d, D_t (>0.6), F_s (>10.0), I_n, B_r, R_r, and R_n indicate low infiltration rate, high runoff conditions with the least potential for groundwater recharge in a basin mainly due to fine texture, sparse vegetation, high relief and a large number of first-order streams. Thus, the greater magnitudes of D_d, F_s, D_t, and R_b designate more erodibility due to their direct connection with soil erodibility. Furthermore, the lower values of C_c, C_cm, D_i, and L_o indicate the higher surface runoff conditions susceptible to flooding, gully erosion, and landslide incidences mainly due to structural disturbances, low infiltration rate, and higher slope/relief. In contrast, the smaller values of L_u, R_b, R_mb, R_t, D_d (<6.0), D_t (<6.0), F_s (<10.0), and I_n indicate low runoff conditions, mainly due to a relatively plain basin with a lesser number of stream segments, coarse texture, high infiltration rate, dense vegetation, and low relief. Such basins are less susceptible to erosion. This review reveals that watersheds can be prioritized using different techniques, including the study of watershed morphometry, LULC analysis, and soil loss estimation. However, morphometric analysis-based watershed prioritization is laborious and time-consuming, and can be replaced by a suitable data dimension reduction technique (e.g. PCA). Thus, geospatial techniques-based watershed prioritization using an appropriate data reduction technique (e.g. PCA) would be useful to identify the key problems associated with water spread, pattern of soil erosion, and aquifer recharge in a catchment.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.