2.1 Study area

The study was undertaken for the Mahanadi river basin, a major peninsular river in eastern India, which faces climate-induced hydrological extremes (Figure 1). The total catchment area of the basin is 141,589 square kilometer (sq km), nearly 4.3% of India's total geographical area. The basin has three drainage subbasins (upper, middle, and lower) and extends over five Indian states (provinces). While 99.3% of the basin drainage area is in the state of Chhattisgarh (upper riparian state) and Odisha (lower riparian state), only 0.7% is in the state of Madhya Pradesh, Jharkhand, and Maharashtra. The basin elevation ranges from less than 5 m in the lower subbasin, which is mostly deltaic plains to about 1500 m in the upper subbasin, which is mostly hilly. A total of 16 major tributaries join the main stem of the river Mahanadi.

The basin has a subtropical climate, with the average temperature during the summer months being about 29°C and in winters about 21°C. The average annual rainfall of 106 years (1901–2006) is about 1400 mm, out of which about 90% is received during the monsoon (June–September). Almost 52% of the basin has medium-textured soil and the rest has mostly fine-textured soil (CWC, India Meteorological Department [IMD], 2015; CWC, National Remote Sensing Centre [NRSC], 2014). The average annual water resource availability in the basin is about 73 billion cubic meters (BCM) (CWC, 2019) and it has a total surface water storage capacity of about 15 BCM. The utilizable groundwater resource of the basin is about 16.5 BCM (Kumar & Bassi, 2021). There is high interannual variability in the basin yield (runoff), with floods occurring during the wet years in the lower subbasin, and water scarcity is experienced during dry years across the basin (Kumar & Bassi, 2021). These years were identified using the classification by the IMD, which is based on the percentage departure of realized (actual) rainfall from the normal rainfall (30 years average). It is +20% or more for a wet year; between −19% to +19% for a normal year; and −20% to −59% for a dry year (Source: https://mausam.imd.gov.in/imd_latest/monsoonfaq.pdf).

Further, a high proportion of the basin area is under agriculture (about 54%) and forest (about 32%). The CWC maintains 19 water quality monitoring stations, 21 water discharge measuring stations, and 61 meteorological (mostly for recording rainfall) observation sites in the basin (CWC, 2018). Of late, conflict over water sharing between the two major riparian states, that is, Chhattisgarh and Odisha, has been brewing due to water diversion infrastructure planned by the former in the middle basin (Dsouza et al., 2017).

2.2 Estimation of water accounts

The water accounts for the Mahanadi basin were estimated for typical hydrological years, that is, normal, wet, and dry. In India, a hydrological year starts in June and ends in May of the subsequent year. For this study, the spatial average of the annual rainfall data from the IMD stations in the basin for 1990–2019 was used for the classification of normal, wet, and dry years (Source: https://indiawris.gov.in/wris/#/DataDownload). The final selection was made considering the availability of data sets for a hydrological year for estimating various components of water accounts. Following the approach, the Years 2015 (normal year), 2019 (wet year), and 2012 (dry year) were selected. The rainfall analysis is presented in the results and findings section.

Preparing the Water Accounts for River Basins in India where there are substantial anthropogenic alterations is a challenge. The first Water Accounting Framework which was developed by Molden (1997) mainly described water use and depletion at the irrigation system level. It is considered that water inflow to an irrigation system equals depletion, outflow, and change in storage within the system (refer to Equation 1)

()

where InFlow_Irri is the inflow comprised of precipitation and any water imports in the irrigation system, is the amount of water removed from the system for intended (evapotranspiration by growing crops) and untended purpose (evaporation form soil, flows to sink), is the amount of water committed for environmental purposes and also that flows out of the system due to lack of storage, and represents water that is added to the aquifers and the reservoirs over the past stock at the end of the hydrological year in consideration. However, this framework does not cover the entire range of water depletion that takes place in the river basin. These include water depletion by municipalities, industries, livestock, and forestry.

Subsequently, the Water Accounting Plus approach developed by Karimi et al. (2013) covered all possible depletion that can happen in a river basin. However, the approach relies on satellite data, which measures hydrological processes indirectly and thus may have uncertainty and error. The errors in rainfall may give erroneous river and aquifer flows (Karimi et al., 2013).

Another approach that is followed for the river basins with human alterations in European Union (EU) is to estimate the renewable water availability by adding outflow and consumption (abstraction minus return) and deducting storage changes that occur in aquifers and man-made reservoirs. The difference in the inflow and the renewable water availability is that the latter is adjusted for evapotranspiration from the natural vegetation and the change in the natural storage (soil moisture and groundwater). However, in India data on abstraction from rivers for various uses, the return flow, and conveyance and distribution losses to sink is unavailable.

Therefore, the water account for river basins in India has to make use of the evapotranspiration requirement of the vegetation (irrigated and rainfed), and the supply requirements for other uses (domestic, livestock, and industries) to estimate the consumptive water use. In this approach, the conveyance losses to the sink will not be considered as consumptive uses are estimated directly. Moreover, the rainfall data available from different gauging stations in Mahanadi is not adjusted to the area covered by each of them. Therefore, considering a spatial average rainfall directly as a component of inflow will give erroneous results. Considering these aspects, the inflows were estimated by modifying Equation (1) to replace depletion with consumption, and expanding it to cover the consumptive use in all the sectors (refer to Equation 2)

()

The in Equation (2) is the actual consumptive water uses in different sectors, including domestic–rural , domestic–urban (, agriculture–irrigated (), agriculture–rainfed (, forest (, livestock (), and industries (), and evaporation from surface water bodies (). The can be represented as

()

The in Equation (2) is the river discharge at the last drainage outlet (terminal point) in the basin that goes to the sea. The in Equation (2) is the change in storage in aquifers , surface water storages (), and soil moisture (). Thus, the can be represented as

()

Equation (2) after incorporating various components of (Equation 3), and (Equation 4) can be represented as

()

The data used for estimating various components on the right side of Equation (5) and the sources of data are provided in Table 1. The methodology for estimating them is described in detail below.

Table 1. Data were used for preparing the water accounts for normal (2015), wet (2019), and dry (2012) year.

Components	Data used	Data source	Data source link/Reference
and	District-wise number of people in rural and urban areas	Census of India	Government of India (GoI) (2011)
	Water supply norm for rural and urban areas	GoI	GoI (2013) and Central Public Health and Environmental Engineering Organisation (1999)
, and	District-wise area under forest, crops, and irrigation	Directorate of Economics and Statistics, GoI	https://aps.dac.gov.in/LUS/Public/Reports.aspx
	State-wise specific information on crop sowing and harvesting and soil type	Agriculture Department in the riparian states	Obtained in-person
	Monthly rainfall data	India Meteorological Department	https://indiawris.gov.in/wris/#/DataDownload
	Monthly climate data (maximum and minimum temperature, relative humidity, wind speed, sunshine hours, solar radiation, pan evaporation)	CLIMWAT database, FAO	https://www.fao.org/land-water/databases-and-software/climwat-for-cropwat/en/
	Average monthly actual evaporation from the forest areas	National Remote Sensing Centre (NRSC)	https://indiawris.gov.in/wris/#/evapotranspiration
	District-wise number of different livestock	Livestock Census, GoI	GoI (2019)
	Water requirement per unit of livestock type	Literature review	Pallas (1986)
	District-wise number of industries and water allocation to them	Water Resources Department of the riparian states	Obtained in-person
	Area under water bodies and the evaporation rate	NRSC	https://bhuvan-app1.nrsc.gov.in/thematic/thematic/index.php
	Evaporation rate	Central Water Commission (CWC)	GoI (2006)
	Observed streamflow at the last gauging station on the main stem of river Mahanadi	CWC	https://indiawris.gov.in/wris/#/DataDownload
	District wise total annual groundwater recharge and the total annual groundwater withdrawal	Central Ground Water Board (CGWB)	http://cgwb.gov.in/Dynamic-GW-Resources.html
	Initial and final storage at the end of the hydrological year for the 11 major reservoirs	CWC	https://indiawris.gov.in/wris/#/DataDownload
	Results from the CROPWAT tool	FAO	FAO (2010)

and represent rural and urban domestic water consumptive use respectively. For this, first, the compound annual growth rate (CAGR) of rural and urban population for the districts falling in the Mahanadi basin between the census years 2001 and 2011 was estimated using data from the Government of India (GoI) (2011). The same was used to estimate rural and urban population for the Years 2012, 2015, and 2019 and apportioned in relation to the area of the upper and lower riparian states districts falling in the basin. Overall, 15 districts of Chhattisgarh (upper riparian state) and 23 districts of Odisha (lower riparian state), which fall partially or fully in the basin were considered. The extent of the district's area in the basin ranges frsom 1.5% to 100% (Central Water Commission, National Remote Sensing Centre, 2014). Thereafter, the annual water supply was estimated considering the water supply norm of 55 L/capita/day (lpcd) for rural areas (GoI, 2013) and 150 lpcd for urban areas (Central Public Health and Environmental Engineering Organisation, 1999). The consumptive water use was assumed to be 30% of the supply in urban areas and 70% of the supply in rural areas, and the rest was considered as return flows to the river system (Kumar & Bassi, 2021).

The consumptive water use for the three hydrological years was estimated for both rainfed () and irrigated crops () using the Food and Agriculture Organization's (FAO) CROPWAT tool (FAO, 2010). The CROPWAT enables estimation of the crop water requirement (evapotranspiration), soil moisture, and irrigation requirement based on soil, climate, and crop data. The observed data on climate (including minimum and maximum temperature, humidity, wind speed, sunshine hours, and rainfall) available for two locations (one each in Chhattisgarh and Odisha) was accessed from the FAO CLIMWAT database, and soil type, and date of sowing and harvesting of each crop for a normal, wet, and dry year were obtained from the respective state governments and used as inputs to the CROPWAT. Using the outputs generated by the CROPWAT, the total consumptive water use in irrigated crops was estimated from the crop evapotranspirative demand () and area under each crop in different seasons (monsoon, winter, and summer), which was extracted from the database maintained by the Directorate of Economics and Statistics, GoI. Two other components contribute to water consumption, that is, nonbeneficial evaporation from the exposed soil surface, and nonrecoverable deep percolation (Perry, 2011). Since, most of the upper and middle part of the basin has hard rock aquifers with limited storage potential and shallow groundwater conditions (Kumar & Bassi, 2021), the nonrecoverable deep percolation was considered to be negligible. Also, as paddy is the major crop in the basin, which requires saturated soil and standing water during its entire growth period (Chapagain & Hoekstra, 2011), the nonbeneficial direct evaporation from soil is minimal (Foster & Perry, 2010) and was thus assumed to be part of the crop evapotranspiration itself. For purely rainfed crops, water consumption was estimated considering the or soil moisture (estimated using the CROPWAT), whichever is lower, and the area under each such crop. For such crops, irrigation water (also referred to as blue water) is not applied and they grow utilizing the available soil moisture (also called effective rainfall component or green water), which may or may not meet the demand. Green water use by the forest () was estimated using the average monthly actual evapotranspiration gridded data (0.05 × 0.05) obtained from the NRSC and the area under forest for the three hydrological years, which was accessed from the Directorate of Economics and Statistics, GoI. The forest area is composed mainly of deciduous trees and plants, which in 2019–2020 occupied about 30% of the basin area.

The consumptive water use by livestock was estimated by apportioning the district-level census data on the livestock type and number for the years 2012 and 2019 (obtained from GoI, 2019), and the voluntary uptake of water per livestock unit for different types of livestock, and the total livestock units for the animal under consideration, for the prevailing climatic conditions (based on Pallas, 1986). For 2015, the CAGR between 2012 and 2019 was used to estimate the livestock population during the year.

For industry, the consumptive water use () was estimated by using the data set on different types of industries in the basin, and the annual water allocation to them. This data was obtained from the Water Resources Department of Chhattisgarh and Odisha. Most of the industries are iron and steel plants and thermal power plants (having cooling towers) that recycle water, as a result only a small proportion is returned to the river system (Chaturvedi et al., 2020). Therefore, it was considered that 80% of the water allocated to the industries is consumed and the remaining is returned to the river system.

Evaporation from the open surface water bodies was estimated by multiplying the spatial data on the total area under surface water bodies during the three hydrological years in the basin with their annual evaporation rate. The evaporation rate from the open surface water bodies was taken as 225 cm/annum (based on GoI, 2006). About 99% of the total 124 thousand surface water bodies have a water spread area of less than 5 sq km and they dry up before the onset of summer (February–May). Only six water bodies (which are reservoirs) have a water spread area of more than 25 sq km and have year-round availability of water. The biggest is the Hirakud reservoir with a water spread area of about 720 sq km at the full reservoir level (CWC, NRSC, 2014). Thus, the evaporation was estimated for 12 months in the case of the six big reservoirs for all the years and only 6 months during a dry year, 7 months during a normal year, and 8 months during a wet year for the other smaller water bodies.

For estimating the change in groundwater storage (), the difference between the net annual groundwater recharge () and the total annual groundwater withdrawal during the three hydrological years was considered. The was estimated by taking a difference of the Central Ground Water Board's (CGWB's) district-level estimates on annual groundwater recharge (from rainfall and other sources including seepage from canals and return flow from irrigated areas) and annual natural discharge or base flow) and apportioning the values as per the area of the districts in the upper and lower riparian states falling in the basin. The total annual groundwater withdrawal was estimated by adding the CGWB's district-level observed data set on annual groundwater abstractions for domestic, livestock, industrial, and irrigation purpose and apportioning it as per the area of the districts in the upper and lower riparian states falling in the basin.

For estimating the change in surface water storage in the major reservoirs (), the difference between the reservoir storage at the start (June 1) and end (May 31) of the hydrological year was used. A total of 11 major reservoirs in the Mahanadi basin were considered. The CWC data on the river discharge at the terminal gauging point (Tikarpara) in the river basin was used for calculating the “outflow” () at the draining outlet of the basin. The data on water storage in reservoirs and river discharge was accessed from the India Water Resources Information System (IndiaWRIS). Change in soil moisture () is considered as nil as the results from the CROPWAT tool show that the soil becomes dry by December during the three considered hydrological years.

2.3 Assessment of river water quality

The preparation of accounts for water quality is still an emerging field, especially in developing economies such as India. SEEA-Water suggests preparing water quality accounts and water emission accounts (United Nations, 2012). While the former describes the state of a particular water body (like the river) in terms of concentration of water quality parameters, the latter describes the quantum of pollutants added to the river by various water use sectors during a year. As the information on the quality of wastewater returned by each sector is unavailable for the Mahanadi basin, both the concentration of selected parameters (water quality account) and the total load contributed by them to the river (water emission account) were assessed using the observed river water quality data.

The water quality assessment helps to plan actions that might be needed to improve water resources. The biological oxygen demand (BOD), nitrogen (nitrates plus nitrites), and electrical conductivity (EC) are considered as parameters with well-documented direct and indirect negative effects on water security for both humans and freshwater biodiversity (Damania et al., 2019; Jayaswal et al., 2018; Wang et al., 2021). The BOD measures the degree of organic pollution in water and is an indicator of river health; nitrogen concentration captures water pollution from excessive nutrient loading in water bodies, primarily caused by agricultural run-off and untreated wastewater; and EC is a proxy for salinity balance and changes to pH levels (Damania et al., 2019). Thus, considering the importance of these three parameters in the maintenance of ecosystem functions and human health, they were considered for the water quality assessment.

The water quality data sets were obtained from the CWC for the first water quality monitoring station (Rajim) on the river main stem, the one in the middle subbasin (Basantpur), and one at the terminal gauging point (Tikarpara) in the basin for the typical hydrological years. The water quality data sets were available as monthly average values for the 14 parameters, out of which data for the three selected parameters were used. These data are in open access and be obtained through the India Water Resources Information System or IndiaWRIS. In the first set of analyses, variation on average monthly values of each parameter with respect to observed monthly streamflows during the three hydrological years (normal, dry, and wet) was assessed. This was used to determine the flow conditions during which BOD and nitrogen values equal or exceed the acceptable limit, the minimum water quality requirement as per the Bureau of Indian Standards (BIS) specification 2296:1992 for use of the inland surface waters for drinking water (with only disinfection) and outdoor bathing purpose (Source: http://117.252.14.242/rbis/india_information/water%20quality%20standards.htm).

In the next set of analyses, the BOD load and nitrogen emission at the three monitoring stations on the river during a hydrological year was assessed. It was estimated by adding the product of average monthly values of each parameter and the observed monthly streamflows. It can be represented in the form of Equation (6) as

()

where is the annual emission or load (in case of BOD) of each selected parameter in million grams, is the average monthly concentration or value of each parameter in mg/L, is the observed monthly streamflows in million cubic meters, and n is the total number of months in a hydrological year for which data sets are available.

In the case of the EC, which represents a material's (water in this case) ability to conduct electric current, the values were first converted to total dissolved solids (TDS) equivalent using the correlation between the EC and the TDS. Both EC and TDS are used to describe salinity levels. For the natural waters having EC up to 500 µ℧/cm, the conversion factor is 0.55 (Walton, 1989), that is, TDS (mg/L) equals 0.55 multiplied by EC (µ℧/cm). Thereafter, the TDS load was estimated using Equation (6).

2.4 Selection and computation of water accounts derived indicators

Some specific water accounts linked indicators that help to improve understanding of the basin water resources and conditions, and water management achievements have been proposed (Hunink et al., 2019; Karimi et al., 2013; Pedro-Monzonís et al., 2016). For this study, water accounts derived indicators covering hydrological, physical, and environmental aspects were computed to identify the variables that are critical for improving the water management at the river basin scale. The choice of indicators was guided by the latest policy discourse adopted by the Indian national government, which is to develop river basin management plans based on a better understanding of the problems related to water scarcity and water quality at the basin scale (Bassi, Beetlestone, et al., 2020). For this, it is important to track key water accounts derived indicators to analyze water use and water quality trends. (Schmidt et al., 2017). Thus, the approach to select water accounts derived indicators focused on those that help to monitor: pressure on the freshwater resources; water productivity; and water quality.

For assessing pressure on water resources (in terms of quantity), a hydrological indicator “water consumption index (WCI)” was used, which is the ratio of water actually consumed to the total renewable water resources (Margat, 1996). The WCI computation does include the portion of the withdrawn water that is returned to the river or aquifer (as return flow). Thus, it is better than other indicators to assess the pressure on water resources which simply provide the ratio of the total volume of abstraction or available volume for abstraction to the total renewable water. The two variables that influence the WCI are the actual annual water consumption in various sectors and the amount of annual renewable water resources in the basin. The latter is computed by deducting green water consumption (evapotranspiration by rainfed crops and forests) from the estimated inflow. Also, the usefulness of water consumption estimates in decisions concerning improving interstate allocation of water was discussed.

To prioritize the allocation of available water to the economic subsectors (agriculture, industries, etc.) and to determine whether allocated water has been used efficiently, water productivity is an important physical indicator. It is scale-dependent. While at the basin scale the water productivity is computed by using the value added by the activity to the economy per cubic meters (cu m) of water consumed (Kijne et al., 2003), at the field scale it is computed by either considering the production (physical) or estimating the net returns (economic) from the activity (say agriculture) per cubic meters of water consumed (Kumar, 2010). In this study, we estimated both the basin-level water productivity of agriculture and livestock and industrial sectors (in INR/cu m) that are the major water consumers and the field-scale water productivity of the major irrigated crops in economic terms (in INR/cu m). While the former is important in terms of decision-making concerning prioritizing water allocation among different economic activities, the latter is useful for assessing whether the best use (in terms of productivity) of the allocated water is made in the agriculture sector.

The value added (in INR) by the agriculture and livestock and industrial sectors in the Mahanadi river basin was estimated using the district level (a total of 38 districts in the upper and lower riparian states) contribution to the domestic product (in INR) during the three hydrological years. This data was accessed from the website of the Directorate of Economics and Statistics of Chhattisgarh (Source: http://descg.gov.in/pdf/publications/latest/ST_DP_CG_2019-20.pdf) and Odisha (Source: http://www.desorissa.nic.in/pdf/estimates-sdp-back%20series-31.05.2022.pdf). The values were apportioned in relation to the area of each of the 38 districts in the Mahanadi river basin. For comparison among the hydrological years, the values were adjusted to constant prices using the wholesale price index of India. The water consumed by different economic activities was derived from the constructed water accounts. The ratio of the value added (in Indian Rupee) to water consumed (in cu m) for the agriculture and livestock and industrial sectors provided the basin-level water productivity estimates for these sectors.

The economic water productivity of the major irrigated crops was estimated by taking the ratio of net return per crop (in INR) to the water consumed by each such crop (in cu m). The net return was estimated by taking the difference between the input cost of cultivation and gross returns per unit of land for each crop. The input cost includes all cash expenses, rent paid in case the land is taken on lease, and the opportunity cost of family labor. The data on these variables were extracted from the database maintained by the Directorate of Economics and Statistics, GoI for all the Indian states (Source: https://eands.dacnet.nic.in/Cost_of_Cultivation.htm).

For monitoring the water quality, we discussed the importance of the water quality index (WQI). WQI gives a single value indicator for water quality based on several water quality parameters (Kankal et al., 2012; Uddin et al., 2021), which makes it easier to monitor it and can assist water managers prepare strategies to improve the quality of river water. Since a WQI based on four parameters (pH, dissolved oxygen, BOD, and fecal coliforms) is already used to classify water bodies in terms of their water quality in India, we focussed on the importance of considering nitrogen and EC (as highlighted in Section 2.3) in such water quality assessment.

2.5 Assumptions and data limitations

The consumptive water use estimates were prepared considering the state of Chhattisgarh and Odisha as about 99.3% of the basin drainage area is in these two states. Further, due to the restriction imposed by the COVID pandemic, water quality data set for the hydrological year 2019–2020 (wet year) was yet to be released by the CWC. The other nearest wet year was 2007–2008, which would not have been an ideal selection for the overall water accounts assessment. Hence, the water quality for the wet year was not assessed.

3.1 Rainfall analysis

The mean annual rainfall in the basin, based on the IMD data from 1990 to 2019, is 1453 mm. The maximum average annual rainfall of 2199 mm was observed in 2019 and the minimum of 1000 mm in 2000. Though there is no substantial spatial variability in the rainfall (CWC, NRSC, 2014), it exhibits high interannual variability, as expressed by the fact that the coefficient of variation in rainfall magnitude is about 20%. Further, analysis of rainfall data with respect to the percentage departure of actual rainfall magnitude from the normal rainfall indicates that there were 21 normal years, 4 dry years, and 5 wet years during the last 30 years for which analysis was performed (Figure 2). While the average rainfall departure from the normal range (−19% to +19%) during dry years was about −5%, for wet years, it was about +16%. Hence, the difference in rainfall magnitude between dry and normal years is not substantial, whereas it is during the wet years. For preparing water accounts, the most recent normal (2015), dry (2012), and wet (2019) year was considered. The average annual rainfall was 1511 mm during the normal year, 1170 mm during the dry year, and 2119 mm during the wet year. The selection of the normal year was also influenced by the availability of other data sets required for preparing water accounts.

3.2 Streamflow and groundwater resources

The long-term average annual streamflow at the terminal point of the basin, calculated using observed river discharge data from 1990–1991 to 2019–2020, is about 44,350 million cubic meters (MCM). The maximum flow was 110,576 MCM during 1994–1995 and the minimum was 14,548 MCM during 2016–2017. The flow at 75% dependability (i.e., minimum flow expected in 75 out of 100 years) was estimated to be 29,000 MCM. These are the flows after meeting the existing water demands (for which water supply infrastructure exists) and including the return flows that drain from the basin to the sea (Bay of Bengal). The streamflow at the terminal gauging site during the normal year was 28,796 MCM, whereas it was 26,985 MCM and 45,308 MCM during dry and wet years, respectively. The estimated annual net groundwater recharge in the basin was 13,633 MCM during the normal year, and 13,896 and 17,593 MCM during dry and wet years, respectively. The estimated net recharge during the dry year was comparable to the normal year as it received higher irrigation return flow and seepage from canals carrying surface water during the nonmonsoon months.

3.3 Consumptive water use in various sectors

The estimated population in the basin in 2012 was 39.4 million, 41.6 million in 2015, and 44.9 million in 2019. While the rural population is increasing at a CAGR of 1.3%, the average annual growth rate in the urban population is higher, at 3.4%. Nevertheless, the rural-to-urban population ratio was 3.0 in 2019. The total estimated water supply for domestic uses during the normal year was 1182 MCM. It was 1106 MCM and 1294 MCM during dry and wet years, respectively. Overall, water consumption was 609 MCM during the normal year, 575 MCM during the dry year, and 658 MCM during the wet year.

The estimated gross area under crops in 2012–2013 (dry year) was 6.6 million ha, which decreased marginally by 3.58% in 2015–2016 (normal year) and by 7% in 2019–2020 (wet year). Nevertheless, the percentage of irrigated area to total cropped area was highest during a wet year (40%) and lowest during a dry year (35%. Almost 75%–80% of the cropped area was under paddy in all the years. Other major crops include maize, wheat, pulses (mainly gram, and pigeon pea), sugarcane, fruits and vegetables, groundnut, mustard, soyabean, and linseed. The total estimated consumptive water use was 16,699 MCM for rainfed crops and 13,286 MCM for irrigated crops during the normal year. The same were 16,442 and 13,732 MCM, respectively, during a dry year, and 18,138 and 14,253 MCM during a wet year.

The forest area was about 5.6 million ha in 2012–2013 (dry year) and 2015–2016 (normal year), which increased only marginally (by 0.4%) in 2019–2020 (wet year). The estimated consumptive water use was 44,737 MCM during the normal year, 44,881 MCM during the dry year, and 45,055 MCM during the wet year. In all the considered hydrological years, consumed water (mainly green water available as soil moisture) for forest areas was high in comparison to other sectors as they need water year-round for their growth.

The estimated livestock (cattle, buffaloes, sheep, goats, and pigs) population in the basin was 17.8 million in 2012. Since then, it has decreased to 13.3 million in 2015, and 11.0 million in 2019. Cattle constitute about 64% of all livestock during the three hydrological years under consideration. The estimates on livestock consumptive water use were 112.8, 149.2, and 92.9 MCM during normal, dry, and wet years, respectively.

The total water allocation to the industries in the basin was 2639 MCM in 2012 and 3048 MCM in 2015. However, due to the closure of certain industrial units, it decreased marginally to 2915 MCM in 2019. Most of the industries in the basin are thermal power plants and iron and steel plants. The estimated consumption of water for industrial use during a normal year was 2438 MCM. It was 2111 MCM in the dry year and 2332 MCM in the wet year.

3.4 Water accounts: Quantity

The water accounts for the normal, dry, and wet year are presented in Figure 3a–c. In the normal year, the total inflow in the basin was 120,604 MCM. Out of which, 65% was consumed in the domestic, agriculture and livestock, forest, and industrial sectors, and about 6% was lost as evaporation from the surface water bodies. These together constitute the consumed water. Further, about 7% of the total basin inflow remained in the aquifers (as a change in storage) after accounting for all the withdrawals in a year. However, as per the estimates of storage change in the reservoirs, it was found that about 1531 MCM was released over the total inflow from the previous year's water stored in the reservoirs during the normal year. The outflow that drains into the sea was about 28,796 MCM (24% of the total basin inflow).

The water accounts for the dry year were very similar to that of a normal year except for the change in reservoir storage and outflow draining the basin. Out of the total basin water inflow (118,953 MCM), the sectoral (domestic, agriculture and livestock, forest, and industry) consumption was 65%, evaporation losses were 5%, and 7% was net addition to the groundwater reserves (after accounting for annual natural discharge and total annual abstractions) at the end of the hydrological year. About 754 MCM was released from the reservoirs (water stored in previous years) over the total inflow received by them during the dry year. This is almost half of the water released during the normal year as reservoirs did not have sufficient storage to meet summer water demands during the dry year. The outflow was estimated to be 26,985 MCM, which is 23% of the total basin inflow.

The wet year receives substantially higher water inflow (149,366 MCM) than the normal and dry year. Out of this, 54% was consumed in various sectors, 5% was lost as evaporation from the surface water bodies, and 10% is available in the aquifers and reservoirs (as storage change) at the end of the hydrological year. Unlike the normal and dry years, reservoirs show a positive trend in terms of change in storage. Further, a high volume of outflow (30% of the total inflow) was available during the wet year.

The CWC estimates for the water resources availability in the Mahanadi basin is about 73,000 MCM (CWC, 2019). These are long-term average annual estimates of blue water availability for the period 1984–1985 to 2014–2015. A hydrological year-based water accounts prepared and presented in this study are much more comprehensive as they provide estimates on various components of water accounts which the CWC assessment does not provide fully. For instance, water consumption by rainfed agriculture and forests was not computed by CWC. Moreover, the assessment by CWC is undertaken only intermittently, and it is not used to compute the water accounts based indicators. This study computes some relevant water accounts based indicators that can be monitored and used to identify elements that need to be altered for improving water resources management at the river basin scale.

3.5 Water quality assessment

The average values of EC, BOD, and nitrogen concentration during a normal and dry year for the three monitoring locations (on the upper, middle, and terminal gauging point on the river main stem) in the Mahanadi basin are presented in Table 2. For all the cases except one, the observed maximum BOD and nitrogen concentration and EC value during a hydrological year were within acceptable limits. The only exception is at Rajim, where the maximum BOD concentration during a dry year was higher than the acceptable limit.

At Rajim, the variation in values of BOD for the water samples collected on a monthly basis with respect to the observed monthly streamflows during a dry year is presented in Figure 4. Rajim, which is the first water quality monitoring station on the main stem of river Mahanadi, presents a particular case. It does not have year-round river flows as is the case with two other water quality monitoring stations. In the dry year, the BOD concentration increased in August during the low flow condition, but then it substantially reduced thereafter even though the flows also reduced further. This means that there was a major pollution episode during August that deteriorated the river water quality. However, post-August the available flows in the river were able to reduce the BOD in the river.

The estimates on emissions or loads at the three monitoring stations are presented in Table 3. In both normal and dry years, the total emissions (load in case of the BOD) in the river increase from the monitoring stations in the upper subbasin (Rajim) to that in the lower subbasin (Tikarpara). Further, the BOD, nitrogen, and TDS increased during the dry year for all the locations (except for the terminal location). Nevertheless, the values of BOD, nitrogen concentration, and EC did not breach the acceptable limits (refer to Table 2). Thus, it can be concluded that the river has a huge dilution capacity, and despite the substantial increase in emission or load in its lower reaches in comparison to upper reaches, the values of BOD, nitrogen, and EC remain within the acceptable values as per the BIS.

The CWC does a seasonal water quality trend analysis of rivers in India. The last one was undertaken from 2011 to 2015 (CWC, 2020). However, they do not compare the monthly value of water quality parameters with the streamflow as has been undertaken in this study. The latter provides a useful information to the water managers on the natural assimilation capacity of the river and when there is a need to initiate actions to improve river water quality. For instance, during a sudden increase in BOD at Rajim as presented in Figure 4.

4.1 WCI

The computed value of the WCI for the Mahanadi river basin is about 46% during the dry year and it reduces to 43% during normal year and 37% during the wet year. The WCI is computed using the water accounts estimate on the total amount of water consumed across various sectors and the total renewable resources in the basin following the method explained in Section 2.4. Going by the threshold of the WCI followed elsewhere in the world (for instance in the EU, where it is referred to as water exploitation index plus) (Faergemann, 2012), the Mahanadi river basin is under severe water scarcity during the normal and dry year (WCI value is equal to or greater than 40%), and also during the wet year (WCI value above 20%). Though the WCI does not factor in the differences in the quantum of water required per unit of production (agricultural, industrial, etc.) between different climatic settings such as in the arid tropics and temperate regions (Kumar et al., 2020), and between different seasons in a year (Pedro-Monzonís et al., 2015), it provides precautionary thresholds above which the ecosystem services would likely be affected by the water abstractions, especially where the water consumption trends seem to continue increasing (Alcamo et al., 2000).

The variables that influence the WCI are water resources consumed and the inflow received in the basin during the hydrological year. Since the inflow in the Mahanadi basin depends largely on the rainfall during the particular hydrological year, which cannot be managed through human intervention (presently there is no water import or export), the scope for improving the WCI rests in reducing the blue water consumed in different sectors. The estimated water accounts for the Mahanadi basin show that water consumed by irrigated crops (57%–60% of the blue water consumed during the 3 years) and lost as evaporation from the surface water bodies (28%–31% of the blue water consumed during the 3 years) are the most crucial variables for managing the water demand.

Paddy (rice) is the main crop in the basin, accounting for 80% of the gross cropped area and 84% of the irrigated area in 2019–2020 (wet year). It is grown in three seasons, that is, monsoon (June–October), winter (August–January), and summer (January–June). The summer variety is fully irrigated. However, the irrigation requirement of summer paddy is about 2.5–5 times higher than the monsoon and winter paddy, the highest being in the dry year (Figure 5). Thus, scope exists for reducing the water consumption by replacing the area under summer paddy with crops that utilize less water but without compromising on the financial returns to the farmers (such as horticultural crops). Some of the vegetables that are grown during the summer months in the basin though not on scale include bottle gourd, bitter gourd, green chilli, and brinjal. Additional field water saving can also be achieved as most of the horticultural crops are amenable to microirrigation systems such as drip technology that can reduce the nonbeneficial evaporation from the exposed soil surface and thus reduce the irrigation requirement, especially during summer months. However, for this to happen, farmers need to be incentivised by providing the subsidy for adopting drip irrigation technologies. Further, to ensure that water is saved at the basin scale as well, farmers should be discouraged to utilise the water saved by drip technology for increasing area under irrigation during the summer months. For this, surface water supplied for irrigation by the water resources department and the groundwater directly abstracted by the farmers', which presently is heavily subsidized, needs to be priced volumetrically. Further, the price can be kept high during the summer months to dissuade farmers from growing water-intensive crops such as summer paddy.

Concerning the evaporation from the open surface water bodies, various approaches adopted under pilot projects in India can also be tried in the Mahanadi basin. Some of the methods used to reduce evaporation include: creating windbreakers to reduce the air movement over water; covering the water surface with fixed or floating material that can trap the air and prevent the release of water vapor to the atmosphere; reducing exposed water surface; and, storing water in aquifers (CWC, 2006). Almost 99% of the total water bodies in the Mahanadi river basin have a water spread area of less than 5 sq km and only six big reservoirs have a water spread area of more than 25 sq km, the highest being 720 sq km for the Hirakud reservoir. The subtropical climate with good overall rainfall conditions offers an opportunity to use native tree species as windbreakers, especially for small water bodies such as tanks. For the large reservoirs, partial covering of the water surface with the floating material is an option. However, the choice of floating material (thermocol boards, photovoltaic solar panels, etc.) to cover the reservoirs should be based on the study of technical feasibility of covering a particular reservoir, economic and environmental viability of the various material options, and their effectiveness in terms of reducing evaporation. Substantial water savings can be obtained by reducing evaporation from the six big reservoirs. Using the estimates of water accounts (evaporation loss from water bodies estimated using the methodology discussed in Section 2.2), the water loss can be reduced by 627 MCM/annum even if one-fourth area of the six big reservoirs are protected from evaporation.

4.2 Water productivity

The estimate of water accounts for the Mahanadi river basin shows that the consumed water in agriculture (for rainfed and irrigated crops and livestock use) is much higher than in the industries (thermal power plants, and mining and quarrying are the major ones). It was 36%–37% of the total water consumed (blue plus green water) during all the years for which water accounts were estimated. However, the water productivity (economic terms) in agriculture was much lower than industries for all the years (Table 4). Though it has been argued that such assessments can be used as a basis for reallocation of water from low- to high-value uses (Molden et al., 2003), water reallocation should not compromise the needs of activities with low-valued uses like crop production on which small and marginal farmers livelihoods' depend and also for environmental purposes (Kumar, 2010). Therefore, to determine the real economic gains from the intersectoral water reallocations, social, environmental, and political costs need to be considered too (Giordano et al., 2021). The social cost can arise due to reduced crop yields and earnings for farmers on account of less availability of irrigation water, the environmental cost can be a reduction in the irrigation return flow in basins where they contribute significantly to river flow during the lean season (nonmonsoon months) and aquifers, and the political cost may arise because of discontent among farmers due to transfer of water from agriculture to industries. Nevertheless, interventions to improve water use productivity in agriculture are needed in the Mahanadi basin and the scope for it in irrigated crops was analyzed further.

The net economic water productivity (in INR/cu m) for the major irrigated crops in the Mahanadi basin is presented in Figure 6. Overall, there are 18 irrigated crops in the basin, but paddy and gram constitute about 92% of the total irrigated area. Gram has the highest net economic water productivity in all the years among the major irrigated crops. In the case of paddy, the net economic water productivity of the variety grown during monsoon is the highest and that of summer paddy is lowest in most cases. During a dry year, the gross return from the sale of summer paddy is unable to cover even the input cost, resulting in losses to farmers. Hence, it gives a negative net return per unit of water consumed. It appears that farmers are taking the risk of growing summer paddy in the anticipation of good return even with the low yield as the price they can get for it is higher than the other two varieties, which are grown in the monsoon and winter months. Nevertheless, our analysis clearly shows that the summer paddy is not an attractive option for farmers during the dry year and for making efficient use of water.

In Section 4.1, we have discussed the scope of horticultural crops in water saving at the basin scale. Presently, groundnut is another cash crop that is grown during the summer months in parts of the basin where agroecology is suitable for its growth. The growing duration (4 months) of summer groundnut matches with the summer paddy. It has an average irrigation water requirement of 400 mm which is about 2.5 times less than the summer paddy. The net economic water productivity of summer groundnut varies from INR 2.6/cu m in a dry year to INR 9.0/0.37 cu m in a wet year. Thus, it can provide much better returns to the farmers in comparison to the summer paddy. Further, if the entire area under summer paddy (360,674 ha during 2019–2020, which was a wet year) is replaced by groundnut, the field level water saving will be about 1758 MCM. Thus, along with the horticultural crops, it is another option that can replace high water-consuming summer paddy without compromising on the financial returns to farmers.

4.3 For surface water quality

The water quality assessment of the Mahanadi River indicates that the BOD, nitrogen concentration, and TDS values increase from the upstream to the downstream stretch of the river during most of the years for which water accounts were estimated (refer to Table 3). The Chhattisgarh part of the river basin has no operational wastewater treatment (Central Pollution Control Board [CPCB], 2015a) and is discharging all its wastewater without any treatment. Odisha has about 13 wastewater treatment plants with an operational capacity of 158 thousand cu m/day, which is only about 12% of the total sewage generated per day (CPCB, 2015a). Although the concentration of all three water quality parameters is within the acceptable limits for drinking water, with only disinfection and outdoor bathing purpose for most of the flow conditions, the scenario may change in the future with the likely increase in water demand and wastewater generation from the domestic and economic sectors and contribution of nutrient-laden agricultural runoff. Therefore, it is necessary to strengthen the existing wastewater treatment capacity and also undertake regular monitoring of wastewater quality in the basin, which can further support the water quality assessment in the future.

In India, two main approaches are used to classify water bodies in terms of their water quality. The first approach is strictly for rivers, wherein any river stretch having a BOD greater than the acceptable limit for drinking water with only disinfection and outdoor bathing (i.e., more than 3 mg/L) is declared as polluted (CPCB, 2015b). However, the priority in terms of restoration of water quality is given to a river stretch having a BOD exceeding 30 mg/L. The second approach is based on the WQI. The WQI is computed for all types of surface water bodies; a score of below 50 classifies them as having a poor water quality (CPCB, 2001). The WQI used in India considers only four parameters, namely, pH, dissolved oxygen, BOD, and fecal coliforms. In both approaches, nitrogen and EC are left out, which along with BOD, are considered as parameters with well-documented direct and indirect negative effects on water security for both humans and freshwater biodiversity (Damania et al., 2019). In the water quality assessment presented in this study, these three parameters were considered, and perhaps, the WQI index used in India should be modified to include nitrogen and EC as additional parameters. Thus, there is a need to revisit the water quality parameters used to classify water bodies as having good or poor quality water and to enable proper identification of polluted river stretches.

4.4 For resolving interstate dispute over water sharing

In India, planning and development of interstate river basins are undertaken within the administrative boundaries of the state and not by considering the natural hydrological system, that is, the whole river basin. This has resulted in disputes among riparian states on water sharing, especially where water demand has gone up remarkably and during dry years when the overall water availability reduces, with each riparian state asking for more water than their allocated share (Richards & Singh, 2002; Bassi, Beetlestone, et al., 2020; Bassi, Schmidt, et al., 2020).

Disputes related to interstate water sharing have surfaced in the Mahanadi river basin as well (Dsouza et al., 2017). Odisha, which is a lower riparian state, has accused Chhattisgarh of planning several projects for diverting water just before it enters the Hirakud reservoir, which is the major source of water supply for domestic uses, irrigation, and industries in Odisha.

The blue water consumption in the two riparian states was analyzed to find out the scope for water reallocation between them. As per the water accounts prepared for the Mahanadi river basin, out of the total blue water consumed in the basin (excluding evaporation from the water bodies), about 67%–72% is in Chhattisgarh and 28%–33% in Odisha for all 3 years. In terms of average blue water (provided through irrigation) consumed per unit area of irrigated land (major water consumer), it is higher in Odisha (6098 cu m/ha) than in Chhattisgarh (5724 cu m/ha). The reason being in Odisha most of the paddy, which is the major crop in the basin, is harvested in winter, whereas in Chhattisgarh, it is harvested during autumn, the former requiring more water during its growth.

Further, Chhattisgarh does have insufficient infrastructure for water conveyance as existing infrastructure provides less water than the actual water requirement (Kumar & Bassi, 2021). Based on the calculation from the observed terminal flow, it is clear that during all the years for which water accounts were prepared, a substantial proportion of water outflow to the sea (refer to Figure 3c). Thus, scope exists for building additional water diversion infrastructure in the upstream (Chhattisgarh). However, it should be planned after taking into account the water requirement of Odisha and the downstream environmental water needs of the basin. As per the water accounts, the overall blue water consumption (irrigation, domestic uses, livestock, and industries) in Odisha varies from 4880 MCM during the wet year to 5494 MCM during the dry year. Concerning environmental flow (e-flow), the National Green Tribunal (NGT), a judicial body handling environmental matters in India, based on a study commissioned in six river basins, has directed that a river should maintain a minimum of 15%–20% of its average lean season (November–March) flow. However, the e-flow is no longer considered only a minimum flow but the quantity, timing, and quality of water flow required to sustain freshwater and estuarine ecosystems and the human livelihoods and well-being that depend on these ecosystems (Arthington et al., 2018). Nevertheless, using the estimates based on the NGT guideline, the average monthly minimum flow requirement of the Mahanadi River is 69–91 cu m/s, which was fulfilled during all the hydrological years considered for this study. This is the minimum flow that should remain throughout the year and was computed using the CWC 30 years (1990–1991 to 2019–2020) monthly river discharge at the terminal gauging station on the Mahanadi River. The same data set was also used for computing terminal outflow (refer to Section 2.2). Thus, the water consumption analysis and the minimum flow (representing e-flow) requirement provide an estimate of the downstream water requirement during the typical hydrological years. To these, water conveyance and transmission losses should also be added.

Nevertheless, based on the observed trends, the amount of additional water that can be diverted upstream would change in the future as the downstream water requirement would increase due to the growth of population, and economic activities. Therefore, water allocation from the new diversion infrastructure to the downstream areas needs to be revisited at regular intervals (say every 10 years), also taking into account the expected socioeconomic growth in the basin.