PCA driven watershed prioritization based on runoff modeling and drought severity assessment in parts of Koel river basin, Jharkhand (India) Open Access

Water is a precious natural resource required for sustainability of human life, a healthy ecosystem and for socioeconomic development (World Health Organization 2005). Water scarcity has become the single greatest threat of climate change, affecting food security, human health and natural ecosystems (Wilhite & Vanyarkho 2000). Increasing population and rapid urbanizations exert pressure on fresh water resources. Water scarcity regions are also increasing rapidly across the world including India (University of Oxford 2019).

Drought is one of the major devastating problems and a recurring event in the Indian subcontinent because of large geographical coverage of arid and semi-arid regions and limited water resources rendering various parts of the country prone to recurrent drought (Gupta et al. 2011; Mohammed-Aslam et al. 2015). The essential factors for the recurrent drought in India are limited water resources, inadequate variation in climatic parameters and reduction in rainfall, high runoff, soil erosion and flash flood that mainly lead to reduction in crop growth and reduce fertility and productivity of agricultural land (Das et al. 2007).

Remote sensing based drought indices, viz., the Vegetation Condition Index (VCI), Temperature Condition Index (TCI) and Vegetation Health Index (VHI) have been used widely by numbers of researchers in parts of India (Heim Jr 2002; Bhuiyan 2004; Murthy et al. 2007; Chaudhary & Pandey 2019) to analyze climate change and its impact on vegetation pattern.

Modeling technique providing reliable runoff estimation plays a crucial role in mitigation of disasters like flood and drought for sustainable development and management of water resources in mountainous regions. There are many methods that have been used for rainfall runoff modeling. SCS–CN (Soil Conservation Services and Curve Number) technique is one of the primogenital and simplest method for rainfall runoff modeling. Several models based on SCS–CN are being referred by different researchers worldwide such as Mishra-Singh (MS) model (2003) (Mishra et al. 2003), Michel model (2005) (Michel et al. 2005) and Sahu model (2002) (Pandey & Sahu 2002).

Watershed management for conservation and development of natural resources require watershed prioritization primarily based on runoff information. Spatial data made it possible to accurately predict the runoff using the Curve number method (SCS–CN, 1972) as widely used for many hydrological applications. This method incorporates important properties of the watershed, specifically soil permeability, land use and antecedent soil water conditions (Rawat & Singh 2017) and used by several researchers (NRCS 1986; Pandey & Sahu 2002; Mishra et al. 2003; Michel et al. 2005; Rajbanshi 2016; Satheeshkumar et al. 2017).

There are so many methods such as quantitative analysis, statistical methods, fuzzy logic, Analytical Hierarchical Process (AHP) and Principal Componemt Analysis (PCA) which have been used for watershed prioritization study. Rahaman et al. (2015) did morphometric characteristics based sub-watershed prioritization using fuzzy AHP and geographical information system in Kallar watersheds in Tamil Nadu. Arefin et al. (2020) shows watershed prioritization for soil and water conservation using a PCA-based approach in the northern elevated tract of Bangladesh. Pandžić & Trninić (1992) explained precipitation anomaly and of a river basin discharge in the Kupa river basin. Javed et al. (2011) have carried out watershed prioritization for soil and water conservation using morphometric analysis through PCA study to demarcate sub-watersheds falling under very high priority of soil and water conservation. Bouvier et al. (2003) did Principal Components (PC) study based on decomposition of the covariance matrix to generate rainfall fields in Mexico City.

Studies based on PCA give good results in different types of prediction modeling. (Samani et al. 2007) applied a simple neural network model for the determination of a non-leaky confined aquifer by normalizing and applying the PCA, which gives good results. (Gurmessa & Bárdossy 2009) Their work presents an efficient data-driven modeling approach to simulate the spatio-temporal dynamics of bed-evolution reservoirs using principal components regression. The work is a step forward to advance the assimilation of numerical and data-driven approaches in modeling long-term sedimentation of reservoirs (Al-Alawi et al. 2008) The combined method, which is based on using both multiple regression combined with principal component analysis (PCA) and artificial neural network (ANN) modeling, was used to predict ozone concentration levels in the lower atmosphere. This combined approach was used to improve the prediction accuracy for ozone.

Jharkhand is a plateau terrain area and a major portion stands on hard rocks with poor primary porosity (Parveen et al. 2012). The water bearing capacity of these rocks depends on their secondary porosity, represented by interconnected fractures. Various districts of Jharkhand state are prone to drought and parts of Ranchi, Lohardaga, Gumla and Khunti districts were affected by climate change. These areas shows variation in maximum temperature, minimum temperature, rainfall and solar radiation in rabi and kharif season, which result poor vegetation health of crops in these areas (Pandey et al. 2012; Tirkey et al. 2018; Chaudhary & Pandey 2019).The plateau type hard rocky terrain in these regions experiences high runoff, inducing soil moisture stress and concomitant drought. Palamu district, located in the northwestern part of Jharkhand state and being a part of the undulating plateau region, has a long history of droughts of various geographic extent, severity, and duration (Pandey et al. 2012). Tigga & Malini (2003) studied drought based on aridity index and found that Palamu district experienced 49 droughts during the 1901–2000 period.

The Koel river basin lies in plateau terrain where rainfall-induced runoff is a major problem, which has led to a water scarcity in the region. Long-term climatic variability showed rising temperature and reducing rainfall over the Koel river basin and, as a consequence, there is a reduction in crop productivity owing to increasing drought events (Tirkey et al. 2018).

The area taken up for the present study forms a part of South Koel river basin of Jharkhand, which is a drought-prone area facing a high runoff problem (Tirkey et al. 2013; Pandey & Stuti 2017). In India, the majority of watersheds do not have rainfall-runoff past records and therefore the SCS-CN method becomes a more appropriate method for estimation of surface runoff due to its simplistic approach.

The main objective of the study was to demarcate high runoff zones and drought-prone zones to create a correlation matrix using PCA for sub-watersheds prioritization. To counter runoff-induced problems in the study area, SCS CN curve number method was used to estimate runoff of 82 sub-watersheds.

The study area is a part of Koel river basin of Jharkhand state covering major five districts namely, Lohardaga, Ranchi, Gumla, Khunti, and Chakradharpur. It is located between 22°48′N and 24°41′N latitude and 83°91′E and 86°12′E longitude. It covers an area of 7,261 km² covering 82 sub-watersheds and exhibits an undulating topography (Figure 1). The area exhibits gently sloping terrain with dominance of agricultural areas over coarse-loamy and fine soils. The mean annual precipitation in this area is 1,200 mm and most of the rainfall water goes as runoff and much less infiltrates to recharge ground water (Pandey & Stuti 2017). The total rainfall runoff volume drains into the Koel river and Karo river.

In the present study, various satellite data and ancillary data have been used for runoff estimation, drought prone area mapping and watershed prioritization, details of which are given in Table 1.

High resolution optical data of Sentinel-2 having thirteen spectral bands with blue, green, red and NIR bands having 10 meter spatial resolution was downloaded for the month of October from the USGS website and used for land use land cover mapping.

The Moderate resolution imaging spectro-radiometer (MODIS) is a multispectral imaging system on board National Aeronautics and Space Administration's (NASA's) Terra and Aqua satellites. It operates using 36 spectral bands and provides twice daily global coverage at 250 m spatial resolution. These data were available in Hierarchical Data Format (HDF) and distributed by the NASA Land Processes Distributed Active Archive Centre (LP DAAC). NDVI, LST MODIS data as MOD13Q1 and MOD13Q3 of 16 days interval was downloaded for the period from 2000 to 2017 with cell size 100*100 m for the month of October to analyze the changes in vegetation coverage.

Topographical maps of 1:50,000 scale were acquired from Survey of India. The study area was covered in 35 toposheets, which were used for the delineation of drainage network and to demarcate sub- watershed boundaries.

To estimate runoff, daily rainfall data of sixteen points within the study area for the period from 2000 to 2014 was acquired from National Centers for Environmental Prediction (NCEP) Climate Forecast System Reanalysis (CFSR). The CFSR was designed and executed as a global, high resolution, coupled atmosphere-ocean-land surface-sea ice system to provide the best estimate of the state of these coupled domains over the period from 1974 to 2014. Rainfall data of CFSR was used for runoff estimation but due to non-availability of CFSR rainfall data after 2014, NASA POWER, IMD rainfall data was considered for further analysis for the same points for the time period of 2015 to 2017. Rainfall data obtained from both the sources showed similar rainfall values hence well complement each other.

Soil data of the study area was acquired from the National Bureau of Soil Science and Land Use Planning (NBSS & LUP).Further, the soil map was validated with the district level soil maps acquired from Jharkhand Space Application center (JSAC) under Department of Information Technology, Govt. of Jharkhand.

Land use land cover classification was carried out for the area of 7,241 km² using image interpretation techniques (shape, size, color, texture, pattern and association) with the help of high-resolution satellite data of Sentinel-2. The prepared LULC map was further clipped into 84 sub-watersheds to compute runoff at sub-watersheds. Drainage network and stream order map in the study area was prepared by using the SOI toposheet of scale 1: 50,000.

A total of sixteen points were taken for rainfall stations within the study area from NCEP/NASA power and IMD. Day-wise rainfall data was arranged based on initial abstraction value for individual sub-watersheds. The location data in .csv format was projected to real world coordinates (as shapefile) in ArcGIS softwere to perform spatial analysis. Nearest neighbor rainfall values have been considered for different individual sub-watersheds. All the rainfall data have been arranged in MS-Excel and further in the Arc-GIS environment to prepare spatial rainfall distribution maps for individual sub-watersheds for the time period of 2000 to 2017. Values of different locations of rainfall value were added as attributes by using Arc GIS software 10.2. Four different classes of soil were mapped in the area based on the soil map obtained from NBSS/LUP and georeferenced using Erdas Imagine software. The prepared soil map was validated with soil data obtained from JSAC. Runoff of 82 sub-watersheds was estimated using the SCS-CN method. Satellite-based drought indices (VCI, TCI, VHI) were used to demarcate drought prone zones. All the raw NDVI values were corrected to make it at the same scale factor by multiplying by 0.0001. Changes in vegetation condition index over the last eighteen years (2000–2018) for the month of October were calculated; long term fluctuation in the condition of the agricultural pattern in the study was calculated by using Raster calculator in the ARC-GIS environment. Rainfall anomaly index was calculated to demarcate drought and non-drought years based on seasonal rainfall values. The comprehensive methodology adopted in this study is shown in Figure 2.

To simplify Equation (3), (Ia) initial abstraction is related to potential maximum retention and the value of P is only considered when it is greater than or equal to Ia. It was considered that if the rainfall is less than the initial abstraction value then there is no run-off available for that rainfall event, and only the rainfall higher than the initial abstraction value is considered for the run-off estimation (Anbazhagan et al. 2005). Hence, only those rainfall data have been considered which are greater than or equal to the value of initial abstraction.

The curve number value is different for different antecedent moisture conditions. AMC –II condition has been considered for this study with an initial abstraction (Ia) value of 0.2S (AMC-II). Initial abstractions (Ia) value was calculated for 17 years (2000–2017) annually using Equation (5) for different sub-watersheds.

By using Equation (6), direct runoff was estimated for 82 sub-watersheds annually for 17 years (2000 to 2017). The spatial distribution maps of runoff were prepared by using Arc-GIS software. Total runoff volume was calculated by multiplying the watershed area with computed direct runoff for individual sub-watersheds.

Further results obtained from TCI, VCI, and VHI indicate they were classified into severity classes. A drought zone map was prepared by compiling resultant output data from VCI, TCI and VHI. A drought zone map that shows values less than 20 is considered as a drought severity zone and a value greater than 50 is a no-drought zone (Amalo & Hidayat 2017). VHI was computed for the entire area to identify drought severity conditions among 82 sub-watersheds lying in the study area.

PCA analysis is used to determine the more sensitive parameter for watershed prioritization (Javed et al. 2011; Arefin et al. 2020). Principal component analysis was applied for this study to deduce the inter-correlation matrix of four parameters (rainfall, drought area, direct runoff and total runoff) to get the first-factor loading matrix using SPSS (ver.25) statistical software. For better results, the first factor loading matrix is then rotated to the loading matrix by using orthogonal transformation. The rotated factor loading matrix from the first-factor loading matrix was obtained by post-multiplying the transformation matrix with the selected component. In the present study, the correlation matrix among the four parameters was created by the first factor loading matrix and rotated factor loading matrix. Watershed prioritization was then done based on the values of highly correlated parameters.

Dendritic drainage pattern is dominant in the Koel river basin (Parveen et al. 2012), drained by two major rivers, South Koel and North Karo, which confluence at the southern tip of the study area (Figure 3). A total of 26,000 streams were observed with the highest 8th order stream within the watershed of area 7,261 sq·km comprising 87 sub- watersheds. Based on drainage network and drainage flow direction (Figure 4), only 3rd order or greater than 3rd order sub-watersheds were considered for runoff estimation. Sub-watersheds as 65,67,81,83 and 85 are of lower order; that is, 1st and 2nd order and hence not considered for analysis. The total number of sub-watersheds in the study area is 87, therefore in the present study 82 sub-watersheds were considered for further analysis.

The study area has been classified into eight LULC classes among which the maximum area was occupied by agricultural land and less area was covered with built-up, reservoirs and water bodies as shown below in Table 2 and Figure 5. Due to water scarcity, most of the agricultural land is transformed into barren land or exists as permanent fallow land (Pandey & Stuti 2017). Different land use land cover lying within 82 sub-watersheds was clipped (Supplementary Figure SF-1). Area of sub-watersheds varies from 25 km² to 310 km². Most of the sub watersheds in the southern part or lower part of the watersheds exhibit lower area and are covered with dense deciduous forest. Sub-watersheds lying in the upper or central part of the watershed constitute a larger area and are dominant with agricultural landuse and fallow land (SF-2).

Spatial rainfall distribution computed for 17 years was classified into six classes which show excellent (>1,400 mm), very good (1,400 mm), good (1,200 mm), moderate (1,000 mm), low (800 mm) or very low (<800 mm) rainfall in different sub-watersheds. The majority of sub-watersheds in the upper and central part of the catchment received less rainfall; that is, from very low to moderate, reflecting their greater proneness to drought conditions with severe water scarcity. Water-scarce regions fall under Ranchi, Lohardaga, Gumla and part of Khunti, where minimum rainfall was recorded in the years 2001, 2005, 2010, 2012, 2015 and 2017 as shown in Figure 6.

The rainfall anomaly index for 17 years was calculated, which results in separate drought and non-drought years. Negative values of rainfall anomaly index represent drought and positive values represent non-drought years. The average rainfall value calculated in the study area is 1,143.86 mm per year. Rainfall anomaly index values vary between −0.4 to 0.3 (Figure 7). Negative RAI was observed in five years (2001, 2005, 2010, 2015 and 2017) in comparison to other years. Drought and non-drought years were categorized based on RAI values (Dutta et al. 2013). Major variations in rainfall was recorded in the sub-watersheds lying within Ranchi, Khunti, Gumla and Lohardaga districts and these areas are facing severe water scarcity and are prone to drought.

Based on drainage condition, infiltration rate, depth, texture and water transmission capacity, the soil was classified into four different hydrological soil groups as A, B, C and D. The criteria which were adopted for selection of different classes are illustrated below in Table 3. The soil taxonomy classes are mainly represented by the major four classes; that is, fine, loamy, fine loamy and coarse loamy, which were categorized into sub-classes according to the published NBSS & LUP soil map. The watershed is mainly dominated by fine soil (Typic Haplaustalfs) with areal coverage of 2175.05 km² and loamy soil (Lithic Ustorthents) with 1740.81 km² (Table 4). Soils covering the maximum part (S.F-3) of watersheds are vulnerable to soil erosion during high rainfall because of less presence of sand and greater percentage of clay in the soil hindering infiltration (Pandey & Stuti 2017). Based on the proportion of sand, silt and clay, different hydrological soil groups have been identified (S.F-4) for the development of hydrologic soil groups (HSG) by following with the help of the standard guideline (Section 2C 5—NRCS TR-55 Methodology). Dominantly, the HSG represented by types (B and C) cover the maximum portion of the study area, reflecting proneness to high runoff potential in major parts.

The SCS CN method involves the relationship between land cover, hydrological soil group and curve number. The method is based on the assumption of proportionality between retention and runoff (Boughton 1989). The areal characteristics of different land use land cover and their hydrological soil group and curve number are shown in Table 5. Runoff was computed for 82 sub-watersheds, weighted curve number and initial abstraction for each individual sub-watershed is shown in Supplementary Figure S.F-5.

Direct Runoff and Total Runoff volume for 82 sub-watersheds was estimated for 17 years from 2000 to 2017 (S.F-6). Direct Runoff has been categorized into the major six classes; that is, very low (<200 mm), low (300 mm), moderate (400 mm), normal (500 mm), high (600 mm) and very high (700–800 mm) (Figure 8). The average runoff (2000–2017) varies from 439.53 mm to 849.33 mm. Sub-watersheds lying in the lower catchments (14 to 43) were not showing much variation in observed runoff during the last 18 years, despite high rainfall received by these watersheds compared to sub-watersheds in upper catchments. Total runoff volume is low because of their small aerial coverage coupled with dominance of dense deciduous forest over fine loamy soils together rendering good water infiltration and less runoff.

Maximum total runoff volumes have been observed in sub-watersheds lying in parts of Ranchi, Lohardaga, Gumla and Khunti. These areas receive less rainfall compared to lower catchment areas in Chakradharpur. Maximum total runoff in sub-watersheds was observed in 2001, 2005, 2010, 2015 and 2017. The majority of sub-watersheds affected or facing high runoff were observed in 2017 with 35 sub-watersheds facing very high to high runoff and most affected sub-watersheds lying in the upper or central parts of catchments. Because of high runoff, reduced soil productivity rendered lower productivity of agricultural land and resulted in frequent agricultural droughts in these areas (Tirkey et al. 2018). Climate change induced negative impacts on vegetation pattern in Rabi and Kharif seasons in the study areas (Tirkey et al. 2018; Chaudhary & Pandey 2019; Mahto et al. 2021) corroborating the fact that watersheds with high direct runoff volume are more vulnerable to drought and water scarcity.

A large part of the study area is covered with agricultural land (2,457 km²). Vegetation pattern and growth may indicate vegetation health and water presence in any region. Variation in the vegetation health over a 17-year period from 2000 to 2017 for the month of October (peak growth period during kharif season) is shown in Figure 9. Maximum area of the watersheds possible was affected by poor vegetation due to high runoff, and maximum variation was seen in the years 2001, 2005, 2010, 2015 and 2017. The result obtained by combining drought indices; that is, Vegetation condition index, temperature condition index and vegetation health index, reflect drought zones (Singh et al. 2003; Amalo & Hidayat 2017). A drought severity map (Figure 10) prepared in the present study showed that study area exhibits a major three zones with severe drought, moderate drought and drought-free zone. Severe drought conditions were identified in drought years and the majority of the watersheds affected by severe drought with poor vegetation health conditions mainly lie in the Ranchi, Lohardaga, Gumla and Khunti.

Maximum temperature exhibited an increasing trend in these areas with reduced rainfall during the last 37 years (Chaudhary & Pandey 2019) possibly leading to reduction in agricultural growth and triggering recurrent drought in these areas.

Variation among different components is taken in this study; that is, rainfall, direct runoff, total runoff and drought zones, as shown by first-factor loading matrix in Table 6 where the eigenvalue of the first two components is greater than 1. Large eigenvalues mean a large amount of variation in the data whereas small eigenvalues mean less variation in the data set (Arefin et al. 2020). The total cumulative value is 84.20%, which indicates good correlation. Meshram & Sharma (2017) remarked that when correlation coefficient values > 0.9, the parameter is strongly correlated, followed by good correlation > 0.75, moderate correlation > 0.6 and poor correlation (<0.6).

Correlation matrix between the four components; that is, drought zones, total runoff, direct runoff and rainfall (Table 7), shows that the first component, drought zone, is strongly correlated with total runoff with correlation value 1.00 and poor correlation existing between direct runoff (0.17) and rainfall (0.32). The second component; that is, total runoff, shows good correlation with drought zone with a value of 0.71 and poor correlation with direct runoff (0.57) and rainfall (0.07). The third component is direct runoff, which shows moderate correlation with rainfall (>0.65) and poor correlation with drought zone (0.17) and total runoff (0.37), the last and fourth component is rainfall, which shows good correlation with direct runoff (0.69) and poor correlation with drought zone (0.32) and total runoff (0.30).

From the above result, it was found that some components have strong correlation, some components have moderate correlation and some components have poor correlation. Based on this analysis, it was not possible to recognize a significant component for this correlation study. Therefore, the first factor loading matrix was rotated to get a reliable correlation. Rotated factor loading matrix from the first-factor loading matrix was obtained by post-multiplying the transformation matrix with the selected component. Table 8 gives the result of the rotation matrix. Rotation matrix analysis shows first PCA component values; that is, drought and total runoff, were strongly correlated as compared to other factors (S.F-7). Therefore, total runoff and drought zone are the most important factors that can be taken for further priority measurement for erosion and water conservation measurement. Based on results from the study, total runoff volume for 23 sub-watersheds of upper catchments (3,652 km²) required high prioritization as compared to 18 sub-watersheds (764 km²) in lower catchments (Table 9).

The present study provided sub-watersheds wise drought severity and runoff analysis for the time period of 17 years (2000–2017) and is the first descriptive study aimed at sub-watershed prioritization in the study area in Jharkhand state using remote sensing and GIS techniques. The methodology adopted for this study provided a proper structured layout to give an idea to locate high runoff and drought-prone areas, which require watershed prioritization to assist watershed management against recurrent droughts. The output maps are very informative as rain gauges are sparsely available to give ideas about exact runoff generated by different watersheds (Sarangi et al. 2005). Recurrent drought and water scarcity is an existing problem in arid and semi-arid regions in India and in other developing countries. An effective management and forecasting system is lacking because of deficient source, unavailability of high resolution satellite data and tenuous subsisting capacity of people living in the societies (Zade et al. 2005). Maps generated by using optical satellite data and MODIS data were a low-cost approach that was suggested in these areas for mapping high runoff zones, drought-prone zones and for watershed prioritization. Prior studies have also been done based on geospatial technologies using optical and MODIS data products covering aspects of climate change and drought vulnerability mapping (Chaudhary & Pandey 2019).

It was observed that reduction in rainfall at upper and central catchments subsequently made these zones prone to water deficit. High direct runoff was estimated in these areas having low rainfall value because of undulating topography and hard rock terrain geomorphology and the presence of massive hard rocks of Schist, Gneissic Complex, consisting dominantly of granites, which form the basement rocks in the Koel River Basin.

High runoff among the majority of sub-watersheds in drought years over larger parts of the study area have been noticed in the present study. The SCS CN method has also been used by other researchers in different parts of India, which shows results similar to present outcomes (Patil et al. 2008; Tirkey et al. 2013; Pandey & Stuti 2017; Pradhan & Joshi 2019). The quantitative estimates from the rainfall anomaly index revealed drought and non-drought years with drought in the major parts of the study area identified in 2001, 2005, 2010, 2015 and 2017 and the larger numbers of sub-watersheds occupied upper catchments with high total runoff volume.

Vegetation yield is one of the parameters which require sufficient water quantity and proper soil nutrients to give sufficient growth. The deteriorating effect on vegetation health in an area clearly indicates inadequate water content and high erosion prospect (Crouse 2018).

The sub-watersheds wise vegetation health and drought study helped to validate the result obtained from the study of vegetation condition as maximum area of watersheds facing poor vegetation condition and the same sub-watershed are highly affected by drought and poor vegetation health. The result obtained from satellite derived drought indices used in this study as VCI, TCI and VHI played a key role to demarcate the catchments facing severe drought. The sub-watershed wise drought analysis has become essential to monitor severe drought facing areas in the past two decades (2000 to 2017) to understand the changing climatic variability witnessed in the area. The results indicated that maximum numbers of sub-watersheds of upper and central catchments are facing severe and recurrent drought as compared to the sub-watersheds lying in the lower catchments, which received sufficient rainfall and have low runoff potential.

PCA analysis is a more suitable, well-known and widely used method for its simplicity to choose more influencing parameters (correlated parameters) that are required for watershed prioritization. Strong correlation was seen between the total runoff and drought regions (obtained from drought indices). Based on PCA studies, we have demarcated the numbers of sub-watersheds that need high priority for water resource management to combat drought in the area.

Advantages of this study: sub-watershed wise runoff and drought analysis was intended as a first assessment along the parts of Jharkhand state. It is quite difficult to estimate runoff volume at different locations because of less availability of rain gauge stations. Results from this study helped to estimate accurate total runoff volume generated by each sub- watershed over a period of seventeen years. This study amalgamated drought influencing parameters to highlight the areas that require urgent watershed management and need implementation of water harvesting structures such as check dams, farm ponds and nala bunds to reduce drought severity and increase ground water potential. The upper and central sub-watersheds in Ranchi, Lohardaga, Gumla and Khunti are mainly affected and more prone to severe water scarcity and drought effects. Watershed prioritization obtained from this study would help district administrators to implement judicious planning and management of soil and water resources. The geomorphological setup of the area being similar to other parts of Jharkhand would enable use of present outcomes for use in other parts to mitigate water scarcity and recurrent drought conditions.

The authors would like to thank the United State Geological Survey (USGS) for providing optical satellite data like Sentinel-2 and MODIS data. Authors also would like to acknowledge the National Bureau of Soil Science and Land Use Planning (NBSS & LUP) for soil data, NCEP CFSR and Indian Meteorological Department (IMD) for rainfall data, and Survey of India (SOI) for providing the topographical map. First author acknowledges the receipt of financial support under the DST INSPIRE Fellowship (DST/INSPIRE/03/2016/002057) from the Ministry of Science and Technology and Department of Science and Technology, Government of India.

Note Applicable (this manuscript does not involve the use of any animal or human data or tissue).

Not applicable (the manuscript does not contain data from any individual person).

Not applicable (the manuscript does not contain data from any individual person).

Authors confirm that all data and materials as well as used software in this study support our published claims and comply with field standards. The data that support the findings of this study are available from the corresponding author upon reasonable request.

All authors contributed to the study conceptualization and design. Material preparation, data collection and analysis were performed by Stuti Chaudhary. Supervision, validation, review and editing were performed by Prof Arvind Chandra Pandey.

The authors declare that they have no competing interests.

This Research was supported by the DST INSPIRE Fellowship (DST/INSPIRE/03/2016/002057) from the Ministry of Science and Technology and the Department of Science and Technology. Stuti Chaudhary received this fellowship.

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