How far are groundwater resource assessments in India reliable?

The article critically reviews the methodology adopted by the Central Ground Water Board (CGWB) for the assessment of groundwater resources in India. The authors developed a water balance equation that simulates the changes in annual and seasonal groundwater storages of an agricultural region, using a water accounting framework; they solved the developed equation using the data of annual and seasonal water level fluctuations for a wet year for Ludhiana district of Punjab to estimate different components of the water balance. The estimated values are used to test the robustness of the CGWB methodology for monsoon recharge. The annual groundwater storage change obtained from solving the water balance equation was compared with the observed change in annual groundwater storage and was found to closely tally. As per the water balance equation, the CGWB estimates of rainfall recharge conform to the annual water level fluctuations recorded by the agency and therefore are not reliable. Hence, we could infer that our methodology is superior to that of CGWB for estimating monsoon recharge. The validated equation was also used to estimate the recharge from rainfall during a dry year. Using these outputs in a simple groundwater balance equation, irrigation return flows were estimated for the wet and dry years. The equation was then extended to the Ahmednagar district of Maharashtra, a hard rock area, for estimating the surface water import during a wet year, the key unknown in the region's water balance. Hence, our methodology can be employed to estimate rainfall recharge for dry years using annual changes in groundwater storage and estimated consumptive uses of water, when the “WLF approach,” Institute for Resource Analysis and Policy used by CGWB, fails, owing to the presence of an additional variable, that is, monsoon draft. For wet years, it can be used to arrive at the consumptive use of water for various purposes.

publications of the agency that present district-wise groundwater resource assessment estimates.
The methodology is known for its simplicity and suitability given the nature of data normally available from the groundwater monitoring program of the state and central government agencies. The methodology also provides scope for continuous improvement, as the norms can be periodically revised and refined for different climatic, physiographic, and hydrogeological regions, based on the information available from detailed case studies of groundwater assessment in different regions of the country. Nevertheless, questions were often raised by scholars and researchers about the robustness of the methodology developed by CGWB from time to time (Dhawan, 1990;Kumar & Singh, 2008;Kumar et al., 2012;Sajil Kumar et al., 2021). Such questions were supported by the apparent mismatch between the estimates of the groundwater development status of different districts and regions, and the situation on the ground with regard to groundwater level trends between years and seasons.
The CGWB methodology for assessing groundwater recharge during monsoon involves estimating recharge either as a fraction of the annual rainfall by multiplying the rainfall with a coefficient (which changes according to the formation characteristics), or the WLF approach which uses the changes in water levels between pre-and postmonsoon and the specific yield of the aquifer, selecting the result from one of the two methods in an arbitrary way (Groundwater Estimation Committee [GEC], 1997). Both have serious limitations. The rainfall infiltration approach does not consider the complex factors such as soil infiltration properties and water table conditions before monsoon that influence the recharge from rainfall during monsoon. The availability of storage space to receive all the infiltrating rainwater is an issue in hard rock formations during the monsoon season (Kumar et al., 2006). Boisson et al. (2014) and Massuel et al. (2014) who studied the performance of tanks in hard rock areas of South India made similar observations, reporting low percolation efficiency of the water stored in the tanks. Deshpande et al. (2016) and Kabir et al. (2022) found that in the hard rock areas of Maharashtra, the monsoon WLFs in the aquifer do not change linearly with the increase in the quantum of rainfall, with the increase in rainfall not resulting in water level rise after a certain stage. Hence, estimating monsoon recharge as a fraction of rainfall does not have much scientific basis, and that premonsoon depth to water levels as well as rainfall significantly influence the WLF in wells, a good indicator of recharge during the season. Further, Kabir et al. (2022) found that infiltration properties of the formation heavily influenced the rate of recharge from rainfall during the monsoon.
On the other hand, the WLF approach only takes into account the groundwater level fluctuations during the monsoon. But it does not consider the water level trends over the year, understanding which is extremely crucial for assessing the storage changes in the aquifer during the year due to recharge (during the nonmonsoon period along with recharge during the monsoon), abstraction, and natural discharge during the nonmonsoon period and for seeing whether they corroborate with the annual water balance estimates. Further, CGWB estimates do not explicitly mention which methodology has been used for specific assessment units. The CGWB estimates of groundwater balance obtained through such methodology are never validated using annual WLF data.
Internationally, Simmers (1998), while reviewing recharge estimation techniques and the spatial/temporal variability in groundwater recharge, stated that the significance of water balance and Darcian approaches should not be underestimated. Simmers (1998) concluded that for regional estimates of recharge, the phenomenon of localized recharge should not be a matter of concern, and instead, a combination of reliable local data, remote sensing, Geographic Information System technology, and geostatistical techniques is required for a better understanding and quantification of recharge.
In this article, we critically review the CGWB methodology for its deficiencies. We then present a water balance equation that simulates the changes in annual groundwater storage and seasonal groundwater storage of an agricultural region, based on a generic water accounting framework, and solve it for the Ludhiana district of Punjab to arrive at different components of the water balance for the region, using the data of annual and seasonal WLFs for a typical hydrologically wet year. The derived values are used to test the robustness of the CGWB methodology for the estimation of recharge from monsoon rains. The data on annual groundwater level fluctuations and seasonal WLFs of the monsoon, recorded by the CGWB and now available from India-WRIS, and the estimates of annual groundwater storage change derived from the same using value of the specific yield of the aquifer were used for this. The water balance equation thus validated was then used to estimate the monsoon groundwater recharge of dry years, during which pumping become rampant rendering the "WLF approach" of estimating "groundwater recharge" invalid. Once again, using the annual WLF data and the validated "rainfall recharge" as inputs, a simple "groundwater balance equation" was used for estimating another important component of the groundwater balance, that is, return flows from irrigation. The methodology was extended to the hard rock district of Ahmednagar in Maharashtra, where the key unknown was surface water import.

| A CRITIQUE OF THE GROUNDWATER RESOURCE ASSESSMENT METHODOLOGY OF CGWB
Groundwater resource planning in India is heavily dependent on the estimates of recharge from monsoon precipitation. CGWB is the only technical agency that comes up with country-wide estimates of annual groundwater replenishment for different assessment units. The methodology followed for groundwater recharge estimation has been under criticism for many years. Dhawan (1990) questioned the reliability of the estimation of the irrigation potential of groundwater resources estimated by CGWB that endangered the scarce groundwater resources in hard rock areas. Kumar and Singh (2008) questioned the methodology for the assessment of the overexploitation of groundwater. Kumar and Singh (2008) critically reviewed both the GEC-1984 andGEC-1997 methodologies and argued that there are several conceptual issues in the estimation of overexploitation of groundwater resources. Kumar and Singh (2008) also illustrated that these methodologies failed to integrate hydrological, geological, hydrodynamic, social, economic, and ethical factors that capture the physical, social, and economic impacts of groundwater overuse. They also argued that there is a lack of reliable estimations of groundwater recharge and drafts to estimate the exploitation. Sajil Kumar and others (2021) argued that the scientific norms set by GEC in 1984 have several limitations regarding the methodology, availability of reliable data, and unit of measurement. They suggested some improvements by considering these limitations with the watershed as an assessment unit with due consideration to the geomorphological and hydrogeological conditions.
The CGWB methodology follows a set of norms established by the GEC to estimate the recharge of groundwater during monsoon and nonmonsoon seasons. It was stated in the GEC and CGWB reports that these norms were set to avoid unreasonably high or low estimates of groundwater recharge values. But these norms with no proper scientific and research backing normalize the effects of extreme rainfall events and in fact led to the misinterpretation of groundwater estimates of the country. For an instance, it was observed in the course of this study that the rainfall received by Ahmednagar district, Maharashtra during the nonmonsoon season is very little and thus the water available for groundwater recharge after the subtraction of effective rainfall is almost zero. But the 2013 CGWB report showed a groundwater recharge of 254.12 million cubic meters (MCM) from rainfall during the nonmonsoon period for Ahmednagar district, Maharashtra. This shows that the estimates of CGWB methodology are questionable and need to be reviewed for their validity regarding the estimations of groundwater resources.
The agency had been developing and refining its methodology for the estimation of groundwater recharge from rainfall and other sources. The CGWB methodology follows a set of norms by the GEC to estimate groundwater resources. The CGWB methodology estimates groundwater recharge from monsoon rainfall by introducing a coefficient to reconcile with the estimates obtained from two different analytical procedures. Known as the "percentage difference," it is the variation of groundwater recharge value, obtained using the WLF method from the recharge value obtained using the "rainfall infiltration factor method" for the monsoon season. The CGWB methodology estimates the recharge from rainfall in the nonmonsoon season by the rainfall infiltration factor method. These norms lack proper scientific and research backing. One major weakness of the CGWB approach is that the specific method used for the estimates for a specific region is not revealed, that is, whether the agency used the WLF approach or the rainfall infiltration approach. This makes the assessment nonfalsifiable.

| Objectives
The objective of the study is to develop an alternate methodology for the estimation of groundwater recharge from rainfall and irrigation return flows, based on a water balance equation that can simulate the annual and seasonal changes in groundwater storage of an agricultural region, using a water accounting framework. The framework uses the data on annual as well as seasonal fluctuations in water levels, estimates of various inflows into the system, and consumptive uses of water or outflows from the system. After solving the equation to derive various components of the water balance, the same equation is used to check the reliability of the methodology adopted by the CGWB for groundwater resource assessment in India, particularly recharge from monsoon rainfall, by checking whether the values of change in annual groundwater stock derived from their estimates of "monsoon and nonmonsoon recharge" and "abstraction" conform to the annual WLF data recorded by the same agency.

| Methodology
We first empirically test the robustness of the CGWB methodology vis-à-vis the "estimate of groundwater recharge from rainfall" that it yields, using a water balance equation that can simulate the annual and seasonal changes in groundwater storage of an agricultural region, based on a water accounting framework that contains the "rainfall recharge" as a component. This exercise is carried out to identify a methodology that can produce the most reliable estimate of "rainfall recharge," for which the data on "annual change in groundwater storage" was used. The validity of the CGWB methodology is tested by checking whether the values of change in annual groundwater stock derived from their estimates of "monsoon and nonmonsoon recharge" and "abstraction" conform to the annual WLF data recorded by the same agency.
Again, with the help of data on annual groundwater storage change, the validated "rainfall recharge" was used to solve a simple groundwater balance equation to estimate another critical component of the annual water balance in irrigated areas, that is, the irrigation return flow. The availability of data on annual as well as seasonal groundwater level fluctuations in a digital form (now on the India-WRIS platform) enables us to do this triangulation. The equation uses the basic concepts in hydrology and geohydrology.
The water balance equations constructed on the basis of the water accounting framework are as follows: (1) SW IMP is the imported surface water, GWR RAINFALL the groundwater recharge through rainfall, ET c , crop evapotranspiration, R e the effective rainfall, C u the industrial and domestic consumptive use, S Δ the aquifer storage change, and − ET R c e the irrigation consumptive use. The groundwater balance equation is as follows: where GWR GROSS is the gross groundwater recharge that is equal to GWR rainfall + RF, GWR GROSS the groundwater draft and RF the return flow.
The water balance equation that is based on the water accounting framework (Equation 1) was first applied to the Ludhiana district of Punjab, which is largely an alluvial plain. Then two extreme hydrological years were selected on the basis of the rainfall data obtained from India-WRIS for the past 10 hydrological years. The hydrological year with the highest rainfall was considered the wet year and the one with the lowest rainfall was considered as the dry year.
To test the validity of the equation, the wet year was considered normal in wet years, no pumping is expected during the monsoon, and because of this, the seasonal change in storage would actually correspond to the effective monsoon recharge.
The monsoon recharge for the wet year was estimated as the product of WLF during the monsoon season, the geographical area of the district, and the corresponding specific yield of the aquifer (called the "WLF approach". The groundwater recharge from rainfall in the nonmonsoon season was estimated by subtracting effective rainfall (obtained from CROPWAT) from total rainfall and then multiplying that value with the geographical area of the district. The sum of these two values yields total recharge from rainfall.
The industrial and domestic consumptive use of the district was obtained from the CGWB reports. The irrigation consumptive use of the district was estimated using CROPWAT by giving the inputs of meteorological data, soil data, cropping pattern, percentage of area sown by each crop, and duration of each crop for the corresponding year. Later, the annual storage change for the Ludhiana district was estimated using Equation (1) for a wet year from the rainfall recharge values obtained from both CGWB methodology and the "WLF approach" as explained above.
These estimated annual storage change values derived using both methods were then compared with the actual annual storage change obtained from annual groundwater level fluctuation data provided by India-WRIS to see which estimate was closer to the latter. Further, the methodology thus validated was later extended to estimate the water balance for the dry year and to arrive at the rainfall recharge estimates for Ludhiana, and rainfall recharge and return flows for both the wet and dry years for the Ahmednagar district of Maharashtra.
The return flow to the aquifer was estimated for both locations using a simple groundwater balance (Equation 2), with the estimates of rainfall recharge, and the groundwater draft as available from CGWB, as inputs.

| Data sources
The data required for the study has been collected from multiple sources. The rainfall data and time-series data on groundwater level fluctuations for the period from 1996 to 2019 were obtained from India-WRIS, a web-enabled water resources information system, which is a memorandum of understanding between the Central Water Commission, Ministry of Jal Shakti and Indian Space Research Organization. The minimum temperature, maximum temperature, humidity, wind speed, and sunshine hours were obtained for both Ludhiana and Ahmednagar stations from CLIMWAT, which is a climatic database that is used in combination with CROPWAT. The groundwater draft and rainfall infiltration factor for both districts were referred from the CGWB report and GEC-1997 report, respectively. The cropping pattern, major crops, and area sown under each crop of Ludhiana district were acquired from Statistical Abstract (2015)  The normal annual rainfall of Ludhiana district and Ahmednagar district is 678.8 and 595.7 mm, respectively. Ludhiana district represents a typical plain topography, whereas Ahmednagar has a diversified topography with western hills, central plateaus, and basins between the plateaus.
Ludhiana district has Indo-Gangetic alluvium and the soil is sandy loam to clayey with alkaline nature. The major drainages of the district are the Sutlej River, its tributaries, and Budha Nallah. The district is predominantly occupied by the Quaternary Alluvium geological formation. The subsurface geological formations comprise sand, silt, clay, and kankar in various proportions. The major water-bearing formations are sand and gravel. The storativity value ranged from 4.3 × 10 −4 to 6.98 × 10 −4 and the transmissivity value ranged from 628 to 1120 m 2 /day in the Ludhiana district.
Ahmednagar district has shallow gray to deep black soils. About 95% area of the district is underlain by hard rock Deccan trap basalt formations, which occurred as basaltic lava flows from intermittent fissure-type eruptions. A small area comes under alluvium along the banks and flood plains of major rivers. The district lies partly in the Godavari basin and partly in the Bhima subbasin of the Krishna River basin, draining its Northern and Southern parts, respectively. The transmissivity varied from 2 to 357 m 2 /day in the Deccan trap basalt formation and 21 to 598 m 2 /day in alluvium formations, respectively.

| RESULTS AND DISCUSSION
In this section, the results of the analysis of long-term trends in average depth to water levels for the districts of Ludhiana and Ahmednagar are presented and inferences are drawn by comparing the two districts that represent two distinct geological settings. Thereafter, the results obtained from the application of the water balance equation based on the water accounting framework vis-àvis the annual groundwater storage change using both the CGWB estimates of rainfall recharge and rainfall recharge estimated from the WLF approach are presented and discussed first for Ludhiana for a wet year. On validation of the water balance equation, the same was used to estimate groundwater recharge for a dry year in Punjab. The groundwater balance equation was subsequently used to estimate the irrigation return flows, based on the estimated draft available from CGWB to show how the return flow changes from a wet year to a dry year.

| Comparing groundwater level trends in
Ludhiana and Ahmednagar Figure 1 shows the graphical representation of the long-term trends in average (depth to) water levels covering pre-and postmonsoon situations in Ludhiana and Ahmednagar, respectively. As it is clear from the figure, there is a consistent long-term lowering of water levels followed by a slight recuperation in Ludhiana. Effectively, there was a marked decline of around 9.50 m, over a period of 22 years. However, in the case of Ahmednagar, there was no significant long-term trend in water levels.
But in the case of Ahmednagar, there was a very sharp seasonal fluctuation in water levels, with the water level rising during the monsoon, followed by a sharp fall after the monsoon. The largest rise in water level during monsoon was 8.16 m, and the lowest was 0.39 m ( Figure 2). There was not a single year in which the WLF during monsoon was negative. Such seasonal fluctuation was not very sharp in the case of Ludhianathe highest rise in water level was only 3.30 m (in 2010), whereas the highest estimated drop in water levels during the monsoon was 4.23 m (in 2004).
Such a differential trend could be attributed to the difference in the formation characteristics. In the case of Ludhiana, with deep alluvial formations bearing water, in the absence of good rains (resulting in very limited recharge), farmers could pump out water from the groundwater reserves during the monsoon season leading to a lowering of water levels below the premonsoon levels, whereas in Ahmednagar, as there is no groundwater stock, pumping during the monsoon would be just limited to the dynamic groundwater (monsoon recharge). The WLFs during the monsoon of different years for both locations are presented in Figure 2.
The average of monsoon WLF for Ludhiana was −0.29 m for the 22-year period, whereas, in the case of Ahmednagar, the average was 3.77 m for the 24-year period. Although the WLF during monsoon indicates the net storage change (MONSOON RECHARGE-WITHDRAWAL), the relatively higher fluctuation in water levels in Ahmednagar (even in good rainfall years when pumping during the monsoon season is unlikely) is because of the low specific yield of the aquifer (WLF = net storage/specific yield).

| The water balance of Ludhiana, Punjab
The rainfall data of the Ludhiana district of Punjab for the past 10 hydrological years, that is, from June 2010-May 2011 to June 2019-May 2020 was checked to identify the extreme rainfall years. The annual rainfall data for the corresponding hydrological years is represented in the form of a bar graph as shown in Later, the daily rainfall data for both the wet and dry years was obtained from the relevant data sources as explained in Section 3.3. The values of the rainfall infiltration factor and specific yield of the district were taken as 0.22 and 0.12, respectively. As discussed earlier, the net groundwater storage change during the monsoon season in the wet year was considered as the groundwater recharge from rainfall during the monsoon.
This groundwater storage change was estimated using groundwater fluctuation data obtained from India-WRIS. The groundwater recharge from rainfall during the nonmonsoon season of the wet year was estimated by subtracting the value of effective rainfall (CROPWAT) from total rainfall and then multiplying the obtained value with the geographical area of the district as 40.17 MCM. The groundwater recharge from rainfall for the wet year estimated by CGWB methodology and the WLF approach proposed by us were 646.85 and 1000.52 MCM, respectively. The irrigation consumptive use of the district was obtained from CROPWAT by giving an input of meteorological data, soil data, cropping pattern, percentage of area sown by each crop, and duration of each crop. The irrigation consumptive use of the district for the wet year from CROPWAT was estimated to be 1478.96 MCM. The surface water import was taken as 717.33 MCM (Kaur et al., 2009). The industrial and domestic use was obtained from the CGWB (2013) report as 106.05 MCM for the wet year. Then, the value of annual storage change was estimated by recharge and abstraction estimations of both methodologies using Equation (1).
The annual storage change values estimated by both methodologies were then compared with the net annual storage change obtained from WLF data of India-WRIS. The results are shown in Table 1, from which it was evident that the annual storage change value obtained from CGWB methodology showed a decline in groundwater storage to an extent of 220.82 MCM. This did not corroborate with the average annual fluctuation in water levels for that year (0.58 m), which corresponds to a positive storage change of 267.06 MCM. Whereas the annual storage change obtained by using the recharge and abstraction estimates from the methodology proposed by us showed an increase in groundwater storage to the tune of +97.50 MCM. The graphical representation of the comparison of these annual storage change values is shown in Figure 4. This clearly showed that the methodology proposed by us is more reliable and robust than the CGWB methodology. Further, the return flow to the aquifer for the wet year was estimated as 2570.67 MCM using Equation (2).
Subsequently, the validated methodology was employed to arrive at the rainfall recharge for the Ludhiana district for the dry year, as monsoon WLF data cannot yield reliable estimates of rainfall recharge during drought years owing to the probably significant rate of pumping that can occur during the rainy season in such years. The groundwater recharge due to rainfall during nonmonsoon season was obtained from the effective rainfall method as 26.08 MCM. The annual storage change for the dry year of the district was estimated from the WLF data of India-WRIS as −1043.53 MCM. The irrigation consumptive use value obtained from CROPWAT was found to be 2341.58 MCM. The industrial and domestic consumptive use was obtained from the corresponding CGWB report as 163.30 MCM (CGWB, 2017b).
T A B L E 1 Parameters of water balance and water accounting of Ludhiana district, Punjab.

Sl. no.
Parameter Values The annual rainfall recharge for the dry year was estimated to be 744 MCM, based on the values of annual groundwater storage change (−1043.53 MCM) estimated using the average annual WLF for that year (−2.25 m), the estimates of consumptive use of water for various sectors for the year and the surface water import. The groundwater recharge from rainfall during monsoon was estimated to be 717.92 MCM, based on the value of groundwater recharge from rainfall during nonmonsoon season obtained from CROPWAT as 26.08 MCM. However, the net storage change during monsoon season was found to be −352.50 MCM. This clearly showed that there was an excess withdrawal of groundwater through pumping for irrigation during the monsoon season owing to insufficient amount of rainfall. Then, the return flow to the aquifer for the dry year was estimated to be 1759.53 MCM using Equation (2). All the results are presented in Table 1. Figure 5 depicts a graphical representation of important parameters in water balance and water accounting of the Ludhiana district, Punjab.

| The water balance of Ahmednagar, Maharashtra
The procedure for the selection of two extreme hydrological years for the Ahmednagar district was the same as that of Ludhiana. The rainfall data from June 2010-May 2011 to June 2019-May 2020 for the district were collected from India-WRIS and were analyzed to identify the extreme rainfall years. The rainfall data for the corresponding hydrological years is presented in Figure 6. From Figure 6 As the methodology proposed by us was found to be more reliable and robust than the CGWB methodology, the same was extended to the water balance and water accounting of Ahmednagar district, Maharashtra.
In the case of Ahmednagar district, the key unknown was the surface water import for irrigation. Consequently, Equation (1) was used to estimate the surface water import to Ahmednagar district based on the values of recharge from rainfall obtained by net storage change of groundwater during monsoon season, annual storage change obtained from the WLF data of India-WRIS, and the consumptive use of different sectors in a wet year. The irrigation consumptive use estimated by CROPWAT for the wet year of the district was found to be 3720.92 MCM. The domestic and industrial use was referred from the CGWB report 2014 (as on March 2011) as 34 MCM for the wet year of the district. The specific yield and rainfall infiltration factor of the district were taken as 0.02 and 0.1, respectively. The groundwater recharge from rainfall during the monsoon season for the wet year was taken as the net storage change during the monsoon season, which was found to be 2291.93 MCM. The groundwater recharge from rainfall during the nonmonsoon period was taken as zero as Ahmednagar district has an insignificant amount of nonmonsoon rainfall and a negligible amount of rainwater available for groundwater recharge during the nonmonsoon season. The surface water import was estimated to be 2687.91 MCM for the wet year of the district using Equation (1).
Subsequently, this estimated value of "surface water import" was used for the water balance of the dry year to estimate the recharge from rainfall. The irrigation consumptive use estimated by CROPWAT for the dry year of the district was found to be 5291.08 MCM. The domestic and industrial use was referred from the CGWB report 2017 (as on March 2011) (1), based on the values of annual storage change, consumptive use for various sectors, and the surface water import obtained from the water balance of the wet year. But the net change in storage during the monsoon season of the dry year as per WLF data was estimated to be 1911.24 MCM, even without incorporating the pumping that might have occurred during the season owing to droughts. This means, if the volume of pumping is considered, the actual recharge during the monsoon season would be even higher. This anomaly could be due to the underestimation of groundwater draft (provided by CGWB estimates) and overestimation of surface water import during the dry year. The return flow to the aquifer for the wet year was calculated using Equation (2) as 376.21 MCM. The return flow for the dry year was estimated to be −1457.29 MCM. But the return flow cannot be negative. This anomaly again could be due to the underestimation of groundwater draft. All the values are presented in Table 2. Figure 7 depicts a graphical representation of important components of the water balance of Ahmednagar district, Maharashtra.

| FINDINGS
Comparative analysis of long-term water level trends of Ludhiana and Ahmednagar showed some distinct patterns. Ludhiana witnessed a consistent long-term decline in water levels along with sharp variation in monsoon WLF between years with fluctuation becoming negative in some years. Contrary to this, Ahmednagar witnessed high yearto-year variation in monsoonal WLFs with no significant long-term trends in water levels, though in none of the years, the monsoonal WLF was negative.
The groundwater recharge values estimated as per the water balance equation developed by us based on the water accounting framework for the selected wet year of Ludhiana district, Punjab was equal to 1000.52 MCM. Interestingly, the annual storage change value for the wet year for the Ludhiana district of Punjab obtained using the  CGWB methodology suggested a decline in groundwater storage to an extent of 220.82 MCM. This did not corroborate with the average annual fluctuation in water levels for that year (0.58 m) which corresponds to a positive storage change of +267.06 MCM. Whereas the annual storage change obtained from the methodology proposed by us using the recharge and abstraction estimates showed an increase in groundwater storage to the tune of +132.84 MCM. This is much closer to the annual groundwater balance estimates available from India-WRIS data. This suggests that our methodology performed better than the one by CGWB in estimating the groundwater balance of Ludhiana.
Subsequently, the validated water balance equation was employed to arrive at the rainfall recharge for the Ludhiana district for the dry year, as monsoon WLF data cannot yield reliable estimates of rainfall recharge during drought years owing to the probably significant rate of pumping that can occur during the rainy season in such years. The annual rainfall recharge for the dry year was estimated to be 740.87 MCM, based on the values of annual groundwater storage change (−1043.53 MCM) estimated using the average annual WLF for that year (−2.25 m), the estimates of consumptive use of water for various sectors for the year and the surface water import. Accordingly, the return flow to the groundwater for wet T A B L E 2 Parameters of water balance and water accounting of Ahmednagar district, Maharashtra.

Sl. no.
Parameter Values and dry years for the district was estimated to be 2570.67 and 1759.53 MCM, respectively. The methodology yielded results that clearly showed a sharp reduction in irrigation return flows during the dry year, a clear indication of its appropriateness. Thus, the methodology proposed by us was found to be more robust and reliable than the CGWB methodology. The same methodology was hence extended to Ahmednagar district which had a much more complex water balance, with surface water import as a major unknown component. Based on an estimate of annual rainfall recharge, the annual change in groundwater storage, and total consumptive use of water, the methodology was applied to arrive at the estimate of surface water imported to the district for irrigation during the wet year. The "surface water import" for the year was estimated to be 2687.91 MCM. The same value was used for deducing the groundwater recharge from rainfall using the water balance equation for the dry year. The rainfall recharge during the dry year was estimated to be 1509.52 MCM. But the net change in storage during the monsoon season as per WLF data was estimated to be 1911.24 MCM, even without incorporating the pumping that might have occurred during the season owing to droughts. This anomaly could be due to the underestimation of groundwater draft (provided by CGWB estimates) and overestimation of surface water import during the dry year. The return flow for the wet year for the district was estimated to be 376.21 MCM.
The CGWB methodology for assessing groundwater recharge during monsoon involves estimating recharge either as a fraction of the annual rainfall by multiplying the rainfall with a coefficient, whose value is decided by the formation characteristics, or the WLF approach, and selecting the result from one of the two methods in an arbitrary way. Both have serious limitations. First of all, the depth of the water table before the onset of the monsoon (Deshpande et al., 2016;Kabir et al., 2022) and the soil infiltration properties (Kabir et al., 2022) have a potential effect on the recharge from rainfall infiltration. While the WLF approach takes into account the groundwater level fluctuations during the monsoon, it does not consider the water level trends over the year, which indicate the storage changes in the aquifer during the year due to recharge (during the nonmonsoon period along with recharge during the monsoon), abstraction and natural discharge during the nonmonsoon period. As a result, the annual water balance estimates are never cross-checked. Further, CGWB estimates do not explicitly mention which one of the two methods has been used for a specific assessment unit. The CGWB estimates of groundwater balance obtained through such estimates are never validated using annual WLF data. Our analysis shows that the CGWB estimates of groundwater balance or the net change in groundwater stock (for Ludhiana, Punjab) were not tallying with that corresponding to the observed change in annual water levels for the district and the difference was in an order of magnitude.
Our methodology considers two water balance equations, one simulates the annual water balance, using a generic water accounting framework, and the other simulates only the changes in groundwater storage over the year (gross recharge − gross abstraction = change in groundwater stock). For solving the annual water balance equation, the WLF during monsoon and the specific yield of the aquifer were used to estimate the monsoon recharge, for years when the pumping during monsoon is expected to be nil or negligible, which helps to get reliable estimates of recharge. Further, by solving the annual water balance, considering all inflows (surface water import, net groundwater recharge from rainfall) and outflows (consumptive water use in all sectors including agriculture, over and above the use of soil moisture), we are able to derive the annual groundwater storage change, which is then compared with the observed change in annual groundwater storage, as available from water level monitoring and the results could be validated as our results were found to closely tallying with the observed annual water level changes. Hence, our methodology is superior to the CGWB methodology for estimating the monsoon recharge. The simple groundwater balance is then used to estimate the return flows from irrigation, which completes the water balance.

| CONCLUSIONS AND POLICY
The CGWB methodology follows a set of norms set by the GEC-1997 to estimate the groundwater resources of India. The CGWB methodology estimates groundwater recharge from monsoon rainfall by introducing a coefficient to reconcile with the estimates obtained from two different analytical procedures. The coefficient is the variation of the groundwater recharge value, obtained using the WLF method from the value obtained using the "rainfall infiltration factor method" for the monsoon season, expressed in percentage terms. The CGWB methodology estimates the recharge from rainfall in nonmonsoon season by the rainfall infiltration factor method. The norms considered for estimating the values of different variables included in the methodology (such as that for "recharge coefficient" and "return flow coefficient") are, however, not based on any measurements or detailed studies undertaken in different climatic, physiographic and geohydrological settings and are instead arbitrary.
One major weakness of the approach is that the specific method used for the estimates for a particular region is not revealed. This makes it difficult to check the reliability of the estimates. We first empirically test the robustness of the CGWB methodology vis-à-vis the "estimate of groundwater recharge from rainfall" that it yields, using a water balance equation developed by us based on a water accounting framework that contains the "rainfall recharge" as a component. The validity of the CGWB methodology is tested by checking whether the values of change in annual groundwater stock derived from their estimates of "monsoon and nonmonsoon recharge" and "abstraction" conform to the annual WLF data recorded by the same agency.
The analysis presented in this article shows that the groundwater recharge estimation methodology being followed by CGWB for groundwater resource assessment and planning yields unreliable results, because of the overdependence on rainfall and infiltration coefficient, as shown by the test performed by us using the water balance equation that uses the inflows and outflows from the system and the "annual storage change in the aquifer." The values of change in annual groundwater stock derived from CGWB's own estimates of "monsoon and nonmonsoon recharge" and "abstraction" were not found to be conforming to the annual WLF data recorded by the agency. Needless to say, given the mammoth size of India's groundwater economy and the heavy dependence of a large proportion of the rural and urban population on groundwater wells for drinking and domestic needs and that poor assessment of the resource means putting the water security and livelihoods of several hundreds of millions of people is at risk, as knowledge about resource availability is key to any management decision-making. Therefore, the country cannot afford to continue using this methodology for resource development and management planning. Therefore, the development of an alternative methodology was imperative.
Therefore, a new method was derived, keeping in view the scientific data available with the CGWB and validated by it. When the annual groundwater storage change obtained from solving the water balance equation derived by us was compared with the observed change in annual groundwater storage, as available from water level monitoring, the results were found to closely tally with the observed annual water level changes and the results could be validated. Hence, we could infer that our methodology is superior to the CGWB methodology for estimating the monsoon recharge. The simple groundwater balance is then used to estimate the return flows from irrigation, which completes the water balance.
It can be concluded that the water balance equation developed by the authors using a generic water accounting framework can be safely employed to estimate groundwater recharge from rainfall in dry years with the help of data on the annual change in groundwater storage and consumptive use of water for irrigation and other purposes when the "WLF approach" fails owing to the presence of a new variable in the water balance equation because of groundwater pumping for irrigation during the monsoon season in drought years, In wet years when there will be no groundwater pumping for irrigation during the monsoon, the water balance equation can also be used to arrive at the consumptive use of water for various purposes, with the help of data on annual groundwater level fluctuation, and estimated rainfall recharge, as well as groundwater recharge from rainfall.
Further, with the availability of good estimates of groundwater draft, the recharge estimates can be used to arrive at irrigation return flows, an important component of the water balance in canal-irrigated areas, when used in a simple groundwater balance. For this, extensive monitoring of groundwater use in different geohydrological environments will be crucial. While it will be practically impossible to monitor the several million wells that are used for agriculture using water meters, proxy methods will have to be employed. In the case of electric pumps, the data on the energy used for pumping groundwater, the average efficiency of the pump sets, and the average depth to the water table can be used to arrive at the groundwater draft. Current meters can be installed in sample electrified wells for estimating pump efficiency. In the case of wells run by diesel engines, current meters can be installed to sample wells to estimate the efficiency of the pumps, based on the known values of pumping depth and pump capacity. This, along with primary data on hours of pumping and pump capacity obtained from sample agro wells, secondary data on the number of wells, and well monitoring data on depth to water levels can be used to estimate the groundwater draft. This way, we can produce good and regular assessments of groundwater development status so that appropriate evidence-based policies at national and state levels could be formulated to improve the situations in the future.
Exercises like the one attempted here, if carried out for a larger number of groundwater assessment units, can help convince the official agencies (such as the CGWB and state groundwater departments of India) of the need to use the alternative methodology proposed here for estimating groundwater recharge and annual and seasonal balance. A reassessment of groundwater resource condition in India using the new methodology itself can lead to a change in the policy outlook for groundwater in the country, as the assessment outcome match more closely with the ground realities.

DATA AVAILABILITY STATEMENT
Data that support the findings of this study are openly available at https://indiawris.gov.in/wris/ and http://cgwb. gov.in/.