Assessment of Spatial and Temporal Variations in Runoff Potential under Changing Climatic Scenarios in Northern Part of Karnataka in India Using Geospatial Techniques

Abstract

The northern dry zone of Karnataka in Southern India is frequently affected by drought, and the overdraft of groundwater resulted in declining groundwater levels. In this context, spatial estimation of available runoff potential, planning and adoption of site-specific interventions for in-situ moisture conservation, supplementing irrigation and groundwater recharge are of prime concern. Therefore, spatial runoff estimation models were developed subdistrict-wise for the northern dry zone of Karnataka using the Soil Conservation Service Curve Number (SCS-CN) method and GIS. The estimated runoff was validated using the recorded data and was found satisfactory (R² = 0.90). The results indicated that for major portion of the study area (61.8%), the mean annual rainfall varied spatially from 550 to 800 mm, and the runoff potential ranged from 10.0% to 20.0% of mean annual rainfall from 1951 to 2013. The higher rainfall and runoff potential was observed in the Khanapur subdistrict which lies in the western part of the selected area. It was observed that the number of subdistricts under the low-rainfall category (<550 mm) has increased, whereas the high-rainfall category (>1100 mm) has decreased over the years. Considerable variation in rainfall and runoff potential was observed during above normal, normal and drought years. The runoff generated from most of the study area was below 10.0% of mean annual rainfall in drought year, <30.0% in above normal year and <15.0% in normal year. The northern dry zone of Karnataka is vulnerable to drought and water scarcity, and the runoff potential was estimated under future scenarios using ENSEMBLE data of CMIP 5 to enable planners to design water-harvesting structures effectively. Finally, based on the modeling results, it was found that by 2050s (2040 to 2069), the runoff potential is expected to increase by 20.0% to 30.0% under RCP 8.5 and by 10.0% to 20.0% under RCP 4.5 and RCP 2.6 scenarios. By 2080s (2070–2099), the runoff is predicted to increase by >30.0% under RCP 8.5, by 20.0% to 30.0% under RCP 4.5 and by 10.0% to 20.0% under RCP 2.6, respectively. Even though considerable increase in runoff potential is predicted for the northern dry zone of Karnataka in the coming years, the current runoff potential itself is relatively high, and there is tremendous scope for its harvesting and utilization for in-situ moisture conservation, supplemental irrigation and groundwater recharge to ensure the long-term sustainability of the region

1. Introduction

Greenhouse gases (GHG) emitted by human activities have increased radiative forcing and have been reported to be contributing to an increase in global mean temperature by approximately 0.74 °C over the past century [1]. The estimates of projected temperature increase over the 21st century ranged from 1.8 to 4.9 °C [2]. The Intergovernmental Panel on Climate Change reported that the rise in the Earth’s surface air temperature since the middle of 20th century is probably due to increase in anthropogenic ozone depletion and global warming. Global warming will prompt changes in precipitation and other climatic factors [3,4]. The increase in temperature results in higher evaporation on the Earth’s surface, variations of hydrologic cycle, precipitation, extreme events and soil moisture status [2]. This will affect the balance between water supply and demand in many regions across the world [5]. The measurements made by the Indian Meteorological Department (IMD) showed that the mean annual temperature of India has increased by 0.5 °C during 1901–2003, whereas the maximum temperature increased by 0.7 °C. It is predicted that the climate change may increase the average surface temperature by 2 °C to 4 °C, decrease the rainy days and increase high-intensity rains in some regions of India [6].

Rainfed farming covers 80% of the world’s cropland and produces more than 60% of the world’s cereal grains, thereby generating livelihoods for people [7]. Water is a key challenge for food production in rainfed regions due to the extreme variability of rainfall, long dry seasons, and recurrent droughts, floods, and dry spells. Water availability to enhance yields can be achieved by improving water utilization and uptake by crops [8]. India ranks first among the rainfed agricultural countries of the world in terms of both extent and value of produce [9]. Rainfed agriculture is practiced mainly in arid, semi-arid and dry sub-humid zones and supports 40% of the national food basket with 55% of rice, 91% of coarse grains, 90% of pulse and, 85% of oilseeds [9]. Indian agriculture is vulnerable to climate change, since 58% of the agricultural area is rainfed, and more than 80% of farmers are small or marginal with less adaptive capacity [10]. Rainfall is the most important natural resource, especially in drylands [8]. The annual per capita availability of water in India has reduced to 1654 m³ in 2007 from 5177 m³ per year in 1951. By 2050, it is predicted to decrease to 1140 m³, resulting in a water scarce condition of <1000 m³ per year [11]. According to the Ground Water Board of India, 15% of the blocks are already overexploited, and the rate of exploitation is growing by 5.5% per year [12]. In arid and semi-arid regions, rainfall is generally lower than the evapotranspiration, and its non-uniform distribution results in frequent droughts during the critical growth stages of the crops, and the rainfall usually comes as intense showers, generating heavy surface runoff and uncontrolled erosion [8]. Despite its scarcity, this rainwater is generally poorly managed, and much of it is lost as runoff and evaporation [8]. Hence, considerable efforts are required to make Indian agriculture resilient to the climatic change.

Harvesting the available rainwater in the field using in-situ moisture conservation techniques and water-harvesting structures and making its use effective is crucial for any project to make rainfed agriculture in arid and semi-arid regions of the developing nations more sustainable [8,9,13]. The spatial variability in runoff is very important for planning in-situ moisture conservation practices and water-harvesting structures in catchments or watersheds [14,15,16,17]. Variations in the hydrologic processes under the projected climate scenarios indicated that tremendous efforts are required to develop sustainable water management strategies for river basins of India [18]. The impact of climate change on the discharge of the Krishna River using SWAT with the assumption of no change in land use and land cover over the time showed an increase in annual discharge, surface runoff and base flow during mid-century [19]. In India, the data of surface runoff is available only from limited sites where gauging stations are available. Hence, modeling is important for estimating runoff, especially from ungauged areas for its sustainable development [20,21]. The SCS-CN method is widely used for accurate estimation of runoff from watersheds [22,23] compared to traditional techniques. It accounts for many factors like topographic features, rainfall, soil, land use and land cover and integration of these into the CN parameter [24,25,26]. Many researchers across the globe directly used, modified and evaluated the SCS–CN model [27,28,29]. The SCS-CN method with the aid of GIS helps to determine the runoff temporally and spatially [12,25]. This methodology is embedded in many hydrological models [30,31]. Some popular hydrological models used to estimate the runoff includes MIKE SHE [30,31], Soil and Water Assessment Tool (SWAT) [32], Water Erosion Prediction Project (WEPP) [33], etc. Many studies projected the impact of climate change on water resources, agriculture and other ecosystems [34,35,36]. Researchers reported the effects of climate change on river basins of India using hydrological modelling [18,37,38].

The selected area, the northern dry zone of Karnataka, a part of Southern India, is affected by prolonged dry spells during the crop season, which leads to its low crop productivity [39]. Adopting moisture conservation measures and providing one or two irrigations during critical growth stages can save the crop and improve the yield significantly [40]. Canal irrigation is available in limited areas, and A major portion of the area is rainfed with groundwater exploitation and relatively high runoff potential [41,42]. Hence, estimation of runoff potential and adoption of suitable moisture conservation practices and water-harvesting structures for rainwater harvesting and groundwater recharge are very important for the sustainability of agriculture in this region. Therefore, the present study was undertaken to estimate the runoff potential available in the selected area using the SCS-CN method and GIS. An attempt was also made to observe the variability of runoff spatially to prioritize the potential areas for rainwater harvesting. The runoff was estimated under changing climatic scenarios for the selected subdistricts, since the distribution of rainfall varies spatially and influences the water available for rainwater harvesting. The prediction of future runoff potential gives insight into designing water-harvesting structures and into efficient management of valuable crops and crop lands.

2. Materials and Methods

2.1. Study Area

The area selected for this study, the northern dry zone of Karnataka, lies in Southern India, located between 13°47′ to 17°30′ N and 74°05′ to 77°36′ E (Figure 1a–c) with an elevation ranging from 75 to 1104 m above the MSL. It is a dry semi-arid region covering nine districts of Karnataka and lies in the Krishna River basin [41]. The average rainfall of these districts varied spatially from 480 to 1745 mm in different subdistricts. Only few subdistricts of Belgavi and Dharwad along the western part of the study area receive relatively higher rainfall, and the remaining areas receive 480–800 mm.

The texture of the soil varies from clayey to loamy. Most of the area is under crop land followed by current fallow, waste land, scrub land and degraded forest. The major crops during kharif season (June to September) are green gram, pearl millet, sunflower, pigeon pea and sorghum, and chickpea and rabi sorghum [39] during the rabi season (November to May). The LULC map from the National Remote Sensing Centre (NRSC) Hyderabad, soil map from National Bureau of Soil Science and Land Use Planning (NBSS&LUP), ASTER DEM (30 m resolution), ENSEMBLE data of CMIP 5.0 (0.5° × 0.5°) from ICAR Headquarters and weather data from Indian Meteorological Department (IMD) were used in the study.

2.2. Estimation of Runoff Using SCS-CN Method

The rainfall grid data (0.25° × 0.25°) from the Indian Meteorological Department (IMD) for the period from 1951 to 2013 were used. The slope map generated from DEM showed that around 70% of the selected area had a slope of less than 5% (Figure 2) and hence, the SCS-CN method was suitable for determining the runoff potential of this region [12]. The SCS-CN method (SCS 1972) uses daily rainfall for estimating the runoff potential and is given below.

$$\mathrm{For}P>0.2S$$

(1)

$$Q=\frac{{(P-0.2S)}^2}{P+0.8S}$$

$$Q=0\mathrm{For}P\le 0.2S$$

(2)

where P = precipitation (mm); Q = surface runoff (mm); and S = potential maximum retention or infiltration (mm). S is related to curve number (CN) as shown in Equation (3).

$$S=\frac{25400}{CN}-254$$

(3)

where CN = curve number.

CN is a function of land use, treatments, soil and antecedent moisture conditions (AMC) of the area and varies between 0 and 100. Even though standard tables are available, in this study, CN was selected based on data from nearby or similar areas. According to soil characteristics, the soils were categorized into four major hydrologic groups, namely, A (low runoff potential), B (moderately low runoff potential), C (moderately high runoff potential) and D (high runoff potential). The antecedent moisture conditions (AMC) considered were AMC I (5-day antecedent rainfall <35 mm), AMC II (5-day antecedent rainfall >35 mm) and AMC III (5-day antecedent rainfall >52.5 mm) [43]. The thematic layers were prepared in ARCGIS and used as input while estimating the runoff. The thematic layers such as ASTERDEM, rainfall, slope, soil map and LULC map were intersected in ARCGIS, and the SCS-CN method was applied (Figure 3). The runoff potential was determined with the SCS-CN method coupled with GIS [43,44]. The observed daily runoff data for the period from 2012 to 2015 from All India Coordinated Research Project for Dryland Agriculture, Vijayapura were used for validation of the model (Figure 1c). The runoff obtained was intersected with the catchments generated in GIS, and catchment-wise, runoff volume was determined for planning the interventions.

2.3. Rainfall and Runoff Analysis

The runoff was estimated subdistrict-wise for the period from 1951 to 2013 using SCS-CN and GIS to enable the stake holders to make use of the data for subdistrict-wise planning. The Mann–Kendall test and Sen’s slope were used to find the trend of rainfall and runoff [45,46]. Rainfall and corresponding runoff for 63 years in each subdistrict was divided into three classes of 21 years, namely, 1951–1971, 1972–1992 and 1993–2013 and were analyzed for its spatial and temporal variability over the years. The annual rainfall pertaining to the subdistricts of selected districts was categorized into drought, above normal and normal years [47]. The years having annual rainfall > +19% were classified as above normal, from −19% to +19% as normal and < −19% as drought years [12,48]. The spatial and temporal variation of runoff corresponding to these three categories of years was analyzed for all subdistricts in the selected districts.

2.4. Rainfall and Runoff under Changing Climatic Scenarios

Many researchers studied climate change impact on river flows using RCP scenarios [49,50]. A Representative Concentration Pathway (RCP) is a greenhouse gas concentration trajectory adopted by the IPCC for its fifth Assessment Report (AR5) in 2014. Four pathways have been selected for climate modeling namely, RCP 2.6, RCP 4.5, RCP 6.0 and RCP 8.5. In the present study, the daily minimum temperature, maximum temperature and rainfall of CMIP 5 (Ensemble data) for RCP 2.6, RCP 4.5 and RCP 8.5 scenarios were used as the data set. Rao et al. [51] and Chary et al. [52] also used same data set/ensemble means from a number of climate models belonging to CMIP 5 for the four RCPs viz, RCP 2.6, RCP 4.5, RCP 6.0 and RCP 8.5 for climate change scenarios. The runoff potential corresponding to the baseline (BL) (1976–2005), 2020s (2010 to 2039), 2050s (2040 to 2069) and 2080s (2070 to 2099) was estimated using the SCS-CN method and GIS with the assumption that no change in the land use pattern may occur over the time. The runoff determined was analyzed to find its variability in the future in order to provide the planners with an insight into designing water-harvesting structures for efficient management of crops and cropping systems.

3. Results

3.1. Spatial Variation of Rainfall and Runoff at Northern Dry Zone of Karnataka

The estimated runoff was validated using the recorded data from All India Coordinated Research Project for Dryland Agriculture, Vijayapura. A linear regression analysis of the measured and estimated runoff was carried out, and an R² value of 0.90 was obtained which showed a good match between the observed and simulated values (Figure 4). Hence, this model could be successfully used for estimating the runoff under various scenarios including climate change. Considerable spatial variability in the annual average rainfall was observed across the subdistricts in different districts (

Table 1. Spatial variability of mean annual rainfall (from 1951 to 2013) in the northern dry zone of Karnataka.

District	Annual Rainfall (mm)
Vijayapura	534–707
Belgavi	532–1745
Bagalkot	523–648
Gadag	501–1083
Koppal	540–707
Bellary	462–650
Davengere	422–1494
Raichur	469–613
Dharwad	605–810

).

The mean annual rainfall varied spatially from 480 to 1745 mm and it ranged from 450 to 2600 mm over the years during 1951 to 2013 (Figure 5a), and runoff varied from 4.4% to 30.1% of rainfall (Figure 5b). The rainfall from a major portion of the area (61.8%) ranged from 550 to 800 mm, and runoff lied between 10.0% and 20.0% of the annual rainfall. Higher rainfall (>1500 mm) and more runoff (>30.0%) was generally observed in the Khanapur subdistrict of the Belgavi district, which is characterized by undulating topography. High runoff occurred due to intense rainfalls even in other subdistricts. Out of the nine districts, only 4.5% of the area had very high runoff more than 30.0%, 22.7% of the area had high runoff ranging from 20.0% to 30.0% and 7.9% of the area had runoff less than 10.0% of annual rainfall (Figure 5b). Ahmadi et al. [53] evaluated the effect of land use changes on runoff over 15 years in the Haraz River basin located in Hyrcania using remote-sensing data and GIS analyses. The annual precipitation of the region was 665 mm, and the estimated runoff was 9.4% of precipitation in 1996 and 9.6% of precipitation in 2011. Adoption of in-situ moisture conservation measures like modified crescent bund and coconut husk burial treatments reduced the annual runoff (22.3% and 20.4% of the annual rainfall compared to 36.9% of the annual rainfall in control), soil loss (47.0% and 49.0% of control) and nutrient loss in cashew garden grown on steep slopes in southern Karnataka with annual rainfall from 3000 to 3500 mm [54]. Ningaraju et al. used SCS-CN and GIS for estimating the runoff from ungauged watershed, Kharadya mill in Karnataka with annual rainfall of 749 mm, and the obtained runoff varied between 35.47 and 240.16 mm from 2003 to 2013 [55]. For a similar watershed, Rawat and Singh [56] reported a mean annual rainfall of 1107.7 mm and an estimated runoff of 238.3 mm during 2002–2015. Parvez and Inayathulla [57] estimated the surface runoff from upper Cauvery Karnataka by the SCS-CN model, and it varied from 170.12 to 599.84 mm in the study area, when rainfall rates were 1042.65–1912 mm.

3.2. Long-Term Variability of Rainfall and Runoff in Different Sub-Districts

Canal irrigation facilities are available in some subdistricts where tanks/reservoirs/water bodies are present. Hence, crops depend mainly on the rainfall availability and its distribution. Considerable temporal variation in rainfall and runoff potential was observed across the subdistricts (Figure 6a,b). While rainfall was high in Belgavi, there was more runoff in the Raichur subdistrict in some years due to the high-intensity rainfall that occurred. The Mann–Kendall test and Sen’s slope showed a significantly decreasing trend (p < 0.01) of rainfall and runoff at Belgavi at a rate of 9.8 mm/year and 0.13% of annual rainfall/year, respectively. The other two locations did not show any significant trend for rainfall and runoff, but a slight increase in rainfall and runoff was observed. In order to determine the long-term variability of rainfall and runoff in different subdistricts, 63 years (from 1951 to 2013) of rainfall and runoff data were divided into three classes of 21 years each, namely 1951–1971, 1972–1992 and 1993–2013, respectively. The number of subdistricts under the low-rainfall category (<550 mm) increased, and the high-rainfall category (>1100 mm) showed a decreasing trend (Figure 7a–c). An increasing trend of runoff ranging from 0.5% to 1.0% of annual rainfall was observed in most subdistricts (Figure 7d–f) during the 63-year period. A decreasing trend of rainfall was observed at Belagavi over the years (Figure 7a–c) as obtained in the Mann–Kendall test and Sen’s slope.

The rainfall in different subdistricts of Vijayapura ranged from 627 to 688, from 605 to 707 and from 534 to 697 mm during the three time periods (

Table 2. Spatial variability of mean annual rainfall and runoff over the years (from 1951 to 2013) in the northern dry zone of Karnataka.

District	Mean Annual Rainfall (mm)			Mean Annual Runoff (% of Rainfall)
	1951–1971	1972–1992	1993–2013	1951–1971	1972–1992	1993–2013
Vijayapura	627–688	605–707	534–697	6.1–16.7	7.1–20.5	7.6–18.3
Belgavi	547–1927	532–1958	537–1748	5.4–32.1	5.9–30.0	5.9–30.1
Bagalkot	528–647	553–602	523–648	6.0–21.4	6.0–17.2	7.4–18.2
Gadag	625–745	501–687	519–1083	5.3–18.2	4.6–16.2	6.7–17.0
Koppal	569–707	540–707	556–646	6.0–22.9	7.3–22.4	8.6–23.8
Bellary	462–650	469–648	479–645	5.8–15.0	7.1–16.4	8.1–16.5
Davengere	422–675	443–952	500–1494	5.2–19.7	7.0–17.0	6.7–23.6
Raichur	593–713	610–692	469–667	6.0–20.0	7.7–21.5	9.2–18.8
Dharwad	616–807	605–767	648–801	5.3–17.4	5.1–15.3	6.0–17.8

). Runoff varied from 6.1% to 16.7%, from 7.1% to 20.5% and from 7.6% to 18.3% of rainfall during 1951–1971, 1972–1992 and 1993–2013, respectively. It is evident from Table 2 that the mean annual rainfall in low-rainfall areas decreased over the years, and intense rains have resulted in more runoff. Figure 5a shows that the Belgavi district consists of low-, medium- and high-rainfall areas. The rainfall and runoff showed a decreasing trend at Belgavi (Figure 7a–f) as obtained by the Mann–Kendall test and Sen’s slope. At Bagalkot, the rainfall and runoff has decreased and then increased over the years (Table 2). At Gadag, even though the rainfall in high-rainfall subdistricts increased, the runoff showed a decreasing trend. At Koppal, the rainfall has decreased, whereas the runoff potential increased due to high rainfall intensities.

In Bellary subdistricts, no considerable variation in rainfall was observed, whereas the runoff has increased over the years. In case of Davengere subdistricts, considerable increase in rainfall and runoff was observed. At Raichur, the rainfall and runoff increased during 1972–1992 and then decreased. In different subdistricts of Dharwad, the rainfall and runoff showed a decreasing trend and then increased. Considerable variability in the subdistrict-wise rainfall and runoff shows the need for planning of location-specific soil and water conservation interventions.

3.3. Variability of Rainfall and Runoff during above Normal, Normal and Drought Years

The rainfall from most of the area ranged from 800 to 1400 mm in above normal years, from 450 to 650 mm in normal years and from 230 to 550 mm in drought years (Figure 8a–c). Normal rainfall at Vijayapura district ranged from 540 to 620 mm. Out of 63 years (1951–2013), 19.0% to 27.0% of the years were above normal rainfall years with a runoff of 11.5% to 25.7%; 62% to 73% were normal rainfall years with a runoff of 7.1% to 17.5% and 6.3% to 11.1% were drought years with a runoff of 2.7% to 8.0% in different subdistricts of Vijayapura (

Table 3. Temporal variation of rainfall and runoff during above normal, normal and drought years in the northern dry zone of Karnataka.

Domain Districts	Sub-Districts	Mean Rainfall (Pmean) (mm) and Runoff (Qrange) (% of Rainfall)
		Above Normal Year			Normal Year			Drought Year
		P mean	Q range		P mean	Q range		P mean	Q range
Davengere	Honnali	2150.9	21.0	32.9	865.9	7.2	13.2	531.0	5.1	9.2
	Harihar	895.0	14.4	22.0	591.4	12.6	18.1	413.9	8.2	13.7
	Channagiri	883.6	12.6	20.0	595.7	6.8	11.7	362.9	2.8	5.2
	Davengere	722.6	11.4	19.0	467.7	10.3	16.3	246.2	7.0	12.1
	Harpanahalli	814.4	10.7	17.8	514.4	6.2	10.8	267.2	1.6	3.2
	Jagalur	678.7	12.8	20.2	463.3	7.4	12.9	230.0	5.9	10.0
Raichur	Deodurg	968.6	15.2	23.6	572.6	9.6	15.9	330.1	5.6	10.1
	Lingsugur	800.9	11.0	19.0	524.8	7.9	13.7	313.2	3.1	7.0
	Manvi	967.4	16.7	26.0	553.2	12.3	19.7	285.3	7.0	11.6
	Raichur	997.1	14.4	23.1	605.7	10.4	17.1	371.9	7.0	12.2
	Sidhnur	939.3	18.6	27.6	567.7	8.2	14.1	301.5	3.7	7.0
Dharwad	Navalgud	1157.9	14.9	23.2	635.3	9.3	15.3	378.6	3.5	5.9
	Dharwad	1101.5	12.4	19.8	696.3	7.1	12.5	310.5	7.2	12.6
	Hubli	1238.3	22.5	32.0	612.2	9.0	15.1	391.2	4.5	7.8
	Kundgol	1216.4	12.8	20.6	667.0	5.0	9.3	393.6	2.1	4.4
	Kalghatti	1258.2	15.4	23.1	717.1	6.4	11.6	425.1	3.4	5.6
Bellary	Huvinahadagalli	791.8	10.6	17.3	556.3	7.8	13.3	324.1	1.0	2.6
	Hagaribommanahalli	893.7	15.6	23.5	594.1	9.0	14.7	332.1	3.0	6.2
	Kudligi	681.9	13.7	21.5	454.2	7.5	12.9	230.4	1.0	2.0
	Sandur	864.4	14.9	23.0	606.3	6.8	11.8	310.2	0.4	1.5
	Hospet	808.8	12.3	19.6	547.5	7.4	12.9	304.8	2.9	5.9
	Bellary	788.5	12.0	19.6	507.9	7.9	13.6	279.8	3.6	6.9
	Siruguppa	872.1	15.3	23.5	581.5	8.4	14.8	387.8	3.7	8.1
Vijayapura	Indi	869.9	13.1	21.8	590.1	9.3	15.8	283.1	2.7	4.8
	Vijayapura	842.0	15.8	24.3	545.0	10.7	17.5	320.5	4.2	8.0
	Sindgi	1039.1	16.5	25.7	619.9	10.2	17.0	344.5	3.2	6.6
	B Bagevadi	898.5	15.5	23.7	571.2	9.9	16.7	358.3	3.1	6.2
	Muddebihal	886.6	11.5	19.5	540.2	7.1	12.5	336.4	3.0	6.0
Belgavi	Khanapur	2646.1	28.1	40.6	1613.0	18.2	28.7	999.7	11.7	20.5
	Sampgaon	1160.7	15.4	23.8	646.3	6.6	12.0	415.0	4.1	7.5
	Belgavi	1383.5	14.9	24.1	815.5	9.2	15.6	517.6	2.6	5.4
	Parasgad	1081.6	16.2	24.2	588.2	9.3	15.3	376.4	5.9	10.0
	Ramdurg	837.6	13.2	20.8	549.3	8.6	14.3	330.4	4.1	7.3
	Hukeri	1091.4	16.5	25.6	664.2	10.8	17.4	374.5	3.4	6.0
	Gokak	1091.4	11.9	19.5	629.0	5.0	9.5	356.8	1.6	3.4
	Chikodi	861.3	10.3	17.8	570.2	6.5	11.4	332.0	2.6	5.4
	Raybag	752.7	12.2	19.9	490.3	7.4	12.4	277.9	6.2	9.7
	Athni	764.2	15.6	23.9	520.4	7.1	12.2	318.3	4.8	9.0
Bagalkote	Jamkhandi	759.6	14.2	21.7	523.2	8.0	13.6	329.9	4.1	7.9
	Mudhol	783.3	10.4	18.3	518.1	5.9	11.0	307.7	0.9	2.2
	Bilgi	847.2	16.9	26.3	573.2	10.0	16.4	286.4	2.0	4.7
	Bagalkote	796.8	11.4	18.8	535.3	8.5	14.2	293.6	4.1	7.3
	Badami	806.3	13.6	21.7	565.3	8.9	14.6	331.6	2.1	4.4
	Hungund	859.0	17.1	26.1	590.2	9.5	16.0	356.7	4.5	8.8
Gadag	Nargund	981.7	13.5	21.2	610.0	7.6	13.0	385.9	1.1	2.4
	Ron	900.8	15.4	24.6	604.6	8.6	14.4	385.2	3.7	7.3
	Gadag	800.8	11.2	18.6	543.0	5.6	10.3	291.9	2.4	4.9
	Mundargi	742.0	14.8	23.1	505.7	8.8	15.0	268.4	3.0	5.0
	Shirhatti	2305.6	26.2	37.7	707.3	5.2	9.5	465.1	2.9	6.0
Koppal	Kushtagi	826.8	14.3	22.5	532.8	10.1	16.6	324.5	4.8	7.7
	Yelbarga	840.2	11.3	19.2	570.3	7.7	13.4	350.1	3.5	6.8
	Koppal	953.5	20.5	30.0	623.2	13.1	20.3	316.7	3.0	6.5
	Gangawati	864.6	14.3	22.2	510.3	9.0	14.8	294.6	3.3	6.2

).

More drought years were experienced in a few subdistricts of Davengere, Dharwad and Belgavi, and above normal years were also higher in Davengere and Belgavi. Considerable spatial variation in rainfall and runoff were observed during different rainfall years (Figure 8a–f). During above normal years, the rainfall in some subdistricts of Belgavi exceeded 2600 mm. In most subdistricts, the runoff was below 30% of the rainfall in above normal years, 15% in normal years and 10% in drought years (Figure 8d–f).

3.4. Prioritization of Areas for Rainwater Harvesting

The mean annual runoff potential obtained using the SCS-CN model was intersected with the catchments generated in GIS, dissolved and catchment-wise runoff volume was generated for planning soil and water conservation interventions (Figure 9). In areas with less runoff volume generated, in-situ moisture conservation needs to be adopted. Similarly, areas with surplus runoff potential after in-situ moisture conservation need to be harvested in water-harvesting structures and may be utilized for supplemental irrigation using efficient application methods or for groundwater recharge. It was found that 41.0% of total area with a runoff volume <10,000 m³ and catchment size varying from 0.1 to 30.6 ha (average size of 1.08 ha) needs in-situ moisture conservation treatments like conservation furrow, contour bunds, compartmental bunding, broad bed furrow (BBF), ridge and furrow, contour cultivation, etc. to retain the moisture in the soil. Around 21.0% of the area had runoff volume ranging from 10,000 to 20,000 m³ with catchment size of 0.58 to 52.7 ha (average size of 11.3 ha) and needs priority for water-harvesting/groundwater recharge structures in addition to in-situ moisture conservation measures. Finally, 32.0% of the area had high runoff potential (>20,000 m³) with catchment size varying from 1.17 to 89.98 ha (average size of 17.1 ha) and needs adoption of water-harvesting interventions as top priority. The map generated (Figure 9) will help planners decide on the site-specific interventions based on the runoff potential available for harvesting.

3.5. Variability of Rainfall and Runoff under Changing Climatic Scenarios

Runoff potential was estimated under changing climatic scenarios for the northern dry zone of Karnataka. Most of the subdistricts had rainfall ranging from 460 to 800 mm and runoff varying from 6.7% to 20.0% of the mean annual rainfall during the baseline period (1976–2005) (Figure 10a). Under changing climatic scenarios, an increasing trend of rainfall and runoff was predicted w.r.t the baseline period in all the scenarios (Figure 10, Figure 11 and Figure 12). Under RCP 2.6, more than a 10.0% to 20.0% increase in runoff was expected from 85.0% and 89.3% of the area by the mid (2050s) and end of the century (2080s), respectively (Figure 10b–d). In case of RCP 4.5, a 10.0% to 20.0% increase in runoff was expected from 69.2% of the area by 2050s and a 20.0% to 30.0% increase from 59.2% of the area by 2080s (Figure 11a–c). Under RCP 8.5, runoff was expected to increase by 20.0% to 30.0% from 61.7% of the area and >30.0% from 86.4% of the area by 2050s and 2080s, respectively, and a substantial increase in rainfall was also predicted in almost all the subdistricts (Figure 12a–c). The results of the study also give immense scope for rainwater harvesting in the future. Similarly, Gardner [58] assessed the effect of climate change on the mean annual runoff in the United States, and Fowler et al. [59] developed a framework for model improvement and simulated runoff under changing climatic conditions for the Harvey River at Dingo Road in the southwest of Australia.

The impact of climate change in the northern dry zone of Karnataka with the assumption of no change in the LULC over time showed considerable increase in surface runoff during the middle and the end of the present century when RCP could be 4.5 and 8.5, respectively. When planning soil and water conservation interventions, rainfall and runoff of the 2020s and the 2050s are important compared to those of the 2080s (2070–2099), which is very far. Therefore, the probable increase in runoff is expected to vary spatially from 10.0% to 30.0% across different subdistricts. The existing runoff potential itself is relatively high, and there is substantial scope for its harvesting and utilization for supplemental irrigation or groundwater recharge for climate-resilient agriculture in this area.

4. Discussion

Estimation of runoff and planning of suitable interventions subdistrict-wise for supplemental irrigation and groundwater recharge are of prime concern for sustainable management of rainfed agriculture in the northern dry zone of Karnataka in India. In view of the limited number of gauging stations and non-availability of sufficient runoff data, a spatial runoff estimation model (subdistrict-wise) was developed using the SCS-CN method and GIS. Higher rainfall and runoff were observed only in a few subdistricts which are close to the west coast, and runoff from a major portion of the area ranged from 10.0% to 20.0% of the annual rainfall. Runoff is expected to increase considerably under different future scenarios. Under RCP 2.6, more than 10.0% to 20.0% increase in runoff is expected by the mid (2050s) and end of the century (2080s). In case of RCP 4.5, 10.0% to 20.0% increase in runoff is expected by the 2050s and 2080s. Similarly, under RCP 8.5, runoff is expected to increase by 20.0% to 30.0% and by >30.0% by the 2050s and the 2080s, respectively. According to the fourth assessment report of the Intergovernmental Panel on Climate Change (IPCC), the average river discharge and availability of water may increase by 10.0% to 40.0% near the high latitudes and in some wet tropics. It may reduce by 10.0% to 30.0% in dry tropics and dry areas at mid-latitudes, part of which are already in water-stressed conditions [35,36]. Previous studies reported that the annual water potential and hydrological extremes in the Krishna basin will increase under future climatic scenarios. They predicted an increase in annual discharge by 13.8 m³ and 27.8 m³ under RCP 4.5 and RCP 8.5, respectively, in the Krishna basin due to hydrological extreme events as compared with that of the past records of 1985–2005 [37]. Another study using the SWAT model for the Krishna basin highlighted that in the mid-century (2044–2070), the surface runoff and water yield were projected to increase by 20.0% and 6.8%, respectively, under the RCP 4.5 scenario, and by 39.0% and 29.0%, respectively, under the 8.5 scenario. By the end of century, the mean surface runoff and water yield were projected to increase by 64.6% and 56.2% using the RCP 8.5 scenario, and by 22.4% and 10.0% under RCP 4.5 scenario, respectively [38]. These results indicate that surface runoff and water yield are more sensitive to climate change. Similarly, Mishra and Lilhare [18] projected a continuous increasing trend of water balance components along with rainfall and air temperature in both RCP 4.5 and 8.5 scenarios of CMIP5 models in the Krishna River basin. Rainfall may increase by 8.0% to 20.0% under RCP 4.5 and 10.0% to 40.0% in the case of RCP 8.5 by the end of the century [38]. Similarly, the surface runoff and streamflow were projected to increase by 20.0% to 55.0% and 20.0% to 60.0% under RCP 4.5 and 35.0% to 120.0% and 40.0% to 120.0% under RCP 8.5 by the end of the century, respectively.

In the selected study area, which is a part of the Krishna River basin, the current runoff potential itself is relatively high, and there is substantial scope for rainwater harvesting and its utilization. Hence, rainwater harvesting and its optimal utilization for supplemental irrigation using efficient application methods is recommended as one the climate-resilient adaptation strategies for the northern dry zone of Karnataka. This methodology could be utilized by planners for the estimation of runoff potential and planning of interventions in similar watersheds/catchments with limited gauging stations or with no gauges. The runoff potential predicted for the future scenarios showed considerable increase in runoff potential in the near future and more scope for planning and adoption of suitable soil and water conservation interventions for this dry zone.

5. Conclusions

The rainfed areas in northern dry zone of Karnataka in Southern India are facing erratic rainfall, water scarcity, soil erosion and prolonged dry spells, which results in reduction in crop yield or even crop failures. This region needs site-specific management of natural resources like soil and water for its long-term sustainability. Planning and adoption of suitable soil and water conservation interventions needs data of runoff potential available in the region. Because of the limited availability of runoff data, SCS-CN and GIS was adopted to estimate the runoff potential for planning the interventions under the present scenario and climate change scenarios. The results indicated that 61.8% of the study area had mean annual rainfall varying spatially from 550 to 800 mm, and the runoff potential ranged from 10.0% to 20.0% of mean annual rainfall during 1951–2013. The higher rainfall and runoff potential of more than 30.0% was observed in limited subdistricts that lie along the western part of the selected area. The Mann–Kendall test and Sen’s slope showed a significantly decreasing trend of rainfall and runoff at Belgavi, the high-rainfall area during the 63-year period. It was also observed that the number of subdistricts under the low-rainfall category (<550 mm) has increased, whereas the high-rainfall category (>1100 mm) has decreased over the years. Considerable variation in rainfall and runoff were observed in high, medium and low rainfall regions as well as in above normal, normal and drought years. The runoff volume was estimated catchment-wise and prioritized the regions for adoption of different soil and water conservation interventions. The results indicated that 41.0% of the area had a runoff volume <10,000 m³ and needs in-situ moisture conservation treatments like conservation furrow, contour bunds, compartmental bunding, broad bed furrow (BBF), ridge and furrow, contour cultivation etc. to retain the moisture in the soil. Around 21.0% of the area had runoff volume ranging from 10,000 to 20,000 m³ and needs water-harvesting/groundwater recharge structures along with in-situ moisture conservation measures. Around 32.0% of the area had high runoff potential (>20,000 m³) that needs adoption of water-harvesting interventions as top priority. The generated map will help planners decide on the site-specific interventions based on the runoff potential available for harvesting. The runoff potential estimated under future scenarios will enable planners to design water-harvesting structures effectively. Based on the modeling results, it was found that by the 2050s (2040 to 2069), the runoff potential was expected to increase by 20.0% to 30.0% under RCP 8.5 and by 10.0% to 20.0% under RCP 4.5 and RCP 2.6 scenarios, respectively. By the 2080s (2070–2099), the runoff was predicted to increase by >30.0% under RCP 8.5, by 20.0% to 30.0% under RCP 4.5 and by 10.0% to 20.0% under RCP 2.6, respectively. Hence, considerable increase in runoff potential is expected for the area in the coming years. The study indicated that current runoff potential itself is relatively high and there is tremendous scope for its harvesting and utilization for in-situ moisture conservation, supplemental irrigation and groundwater recharge to ensure the long-term sustainability of the region.