Water–energy–food nexus: toward exploring the connection between ancient and modern science in the Indian context

Abstract

In ancient India, air, water, and solar radiation were represented as gods and key components of water availability, food production, and energy generation. However, the linkage between modern science and ancient wisdom has not been well explored. In this study, we estimate the water–food–energy potential using various openly available data products and perform statistical analysis to establish the scientific basis behind the geographical alignment of eight prominent Shiva temples along the 79° meridian east to the north, known as the Shiva Shakti Aksh Rekha (SSAR). Results indicate a strong correlation between the SSAR belt and water–energy–food productivity potential, where 18.5% of the area can produce 44 × 10⁶ tons of rice annually. With an estimated renewable energy generation potential of 596.6 GW, the SSAR belt could significantly contribute to India’s target of 500 GW of renewable energy production by 2030. This study highlights the potential role of traditional geographical alignments in supporting future water, energy, and food security in highly populated countries such as India.

Introduction

The growing population mandates progressive socioeconomic development to manage water, energy, and food, as outlined in UN Sustainable Development Goals (SDGs) 2, 6, and 7 (Mondal et al. 2023). This is particularly relevant in a country such as India, where 18% of the global population (Monika, 2017) puts tremendous pressure on these resources. While significant progress has been made in meeting food (Ray and Bhattacharyya, 2020) and energy demands (Praveen et al. 2021), a deeper understanding of the role of water as an enabler of food and energy security, as well as the interactions between these three sectors in complex systems (the so-called water–energy–food nexus), is necessary. In this context, the interactions of various hydrological (S. Khan et al. 2009) and atmospheric variables (Van Dijk and Renzullo, 2011) deserve additional attention for developing well-informed adaptation policies. This can be achieved through the utilization of remote sensing techniques (Sheffield et al. 2018), on-site measurements with the IoT and sensors (Garrido-Momparler and Peris, 2022), and the development of more detailed hydrological and climate models, including the latest machine learning models (A. Roy et al. 2023). However, the linkage between the knowledge existing in ancient and modern eras is often missing. This study aims to bridge that gap by exploring the scientific reasoning behind the alignment of eight major Shiva temples along the same meridian within the context of the water–energy–food nexus (refer to the section “Historical context of the eight temples”). We utilize state-of-the-art methods and high-resolution spatial–temporal hydrometeorological data to perform detailed analyses. More details on the knowledge that existed in the ancient era and its relevance to modern science are presented in the section titled “Linkages between knowledge of ancient civilization, Hinduism, and modern science”.

To our knowledge, no scientific literature has analyzed the rationale behind the alignment of these eight Shiva temples along the same axis (the selection criteria for the temples are described in the section “Rationale behind the selection of temples”). The main objective of this study is to understand the existence of temples, nature, and socioeconomics within the same belt since ancient times and to explore whether there is a scientific basis behind this phenomenon, based on the estimation of water, energy, and food production potential. The selected study region covers around 50% of the Indian subcontinent; extends between 8°N–36.78°N and 75.6°E–83.2°E; and is further divided into three belts: SSAR, SSAR west region (SSARW), and SSAR east region (SSARE), to examine the spatial variation in water availability, energy generation, and food productivity potential. The hypothesis of the study is designed to answer: Did the alignment of eight prominent Shiva temples along the SSAR belt happen to be random, or is there a scientific basis linked to natural resources availability? To investigate this, the methodology section is divided into three subsections. The first subsection analyses the temporal and spatial variability of water and food productivity across different climatic periods (1903–1932, 1933–1962, 1963–1992, and 1993–2022). The second subsection specifically analyses the variability of renewable energy potential across the study belts. The third subsection combines water, energy, and food productivity. A total of nine parameters pertaining to water and food productivity are computed from empirical and physical equations available in the literature, and a suitable classification method is chosen by considering expert opinion. The analytical hierarchy process (AHP) (Saaty, 1990) is employed to run pairwise comparisons between parameters based on the information available in the literature and field investigation. The highest importance is assigned to water availability parameters since most of the ancient civilizations focused primarily on water availability. Least importance was assigned to climate variability due to a lack of solid evidence about climate change during the period of ancient civilizations. These weights are further used in weighted overlay analysis (WOA) (P. Ghosh and Lepcha, 2019; Hassan et al. 2020; Sadeghi et al. 2020; Walke et al. 2012) to generate an overlay map of the water–food productivity potential (WFPPI) for different belts. Additionally, the renewable energy potential index (REPI) was computed by integrating hydropower, wind, and solar power potential to quantify the regional variability in energy potential. Initially, hydropower potential maps are represented as point (vector) data and are not appropriate for use in weighted overlay analyses. To overcome this issue, we develop a hydropower density map that represents the amount of generated power distributed per unit area. The theoretical maximum long-term (1982–2022) average annual wind power is computed to develop the wind energy potential map. The analysis used the global PV panel database (Belsky et al. 2022) to develop a solar energy potential map. The maps are classified based on the expert’s opinion and statistical scaling. Three classified energy potential maps are further used in the equally weighted sum method to compute the renewable energy potential index (REPI), which further overlays the WFPPI and generates the water–energy–food productivity potential index (WEFPPI). Finally, the performance of the WEFPPI was evaluated using the area under the curve (AUC) of the receiver operating characteristic (ROC) curve (Bowers and Zhou, 2019; Jiménez-Valverde, 2012; Pepe et al. 2006). The results showed that there was a positive correlation between the placement of ancient Shiva temples and water, energy, and food productivity potential. These relationships are significantly influenced by factors such as hydrological variables, soil type, geography and position. A comparative analysis between the three belts (i.e., SSARW, SSAR, and SSARE) showed that the SSAR region has great potential for water, energy, and food productivity due to the presence of all those elements, followed by the SSARE and SSARW regions.

This study might be helpful for decision makers and stakeholders in making optimal decisions for identifying the potential locations of renewable energy plants and agricultural farms. In recent decades, the impact of climate variability has also been prominent on the availability of renewable resources and water. Increasing temperatures accelerate the rate of evaporation, and large variations in rainfall magnitude and frequency add a multitude of stress to the various rainfed river basins of the Indian Peninsula (U.K. Singh and Kumar, 2018). An increasing trend in the kharif session temperature was found in the northern and southern regions throughout the last two climate periods, with a <10% significance level. Simultaneously, this region has an extremely high solar power potential. Increasing temperatures cause negative feedback effects on the maximum power of a PV cell and crop yield, which can reduce the overall power generation capacity of a PV power plant and food production (Zhao et al. 2017), respectively. Tamil Nadu’s coastal regions, particularly Rameswaram, have extremely high and very high wind power potential. However, since 2000, tropical cyclones have severely impacted these regions (Nadimpalli et al. 2021; Rao et al. 2020), posing a serious threat to solar power plants and wind turbines. However, the cold desert of Ladakh has great potential for wind and solar power and is well aligned with the focus of the Indian government to increase the renewable power generation capacity of India by the year 2030. As the severity of climate change progresses toward the future, its impact on water, food, and energy resources also increases, causing adverse impacts on the socioeconomy and sustainability of the study region as well as India.

Details of the study area

The following are the key research questions, and based on that, the study region was chosen: (1) Is there any hidden relationship between the eight well-known Shiva temples and the potential for water–energy–food resources along the SSAR belt? and (2) If yes, how much of an increase/decrease in potential would be compared to the neighboring belts? The total spatial extent of these belts encompasses 50% of the Indian subcontinent. The study region is divided into three belts: (a) the Shiva Shakti Aksh Rekha belt (SSAR), which extends from 8.48° to 35.97°N and 78.12°–80.63°E; (b) the Shiva Shakti Aksh Rekha East belt (SSARE), covers 15.7°–28.64°N and 80.63°–83.2°E; and (c) the Shiva Shakti Aksh Rekha West belt (SSARW), spans from 8° to 36.78°N and 75.6°–78.12°E (Fig. 1). This division facilitate a comparative analysis between the three belts and provide a scientific basis to explore overall, these three regions cover a large part of the Deccan Plateau and are surrounded by the Eastern Ghats, Western Ghats, and the Narmada River. This area is the most ancient geomorphic province and is characterized by aged rocks, both eroded and planar surfaces, along with rivers (Kale, 2014). This was ruled by the Vakataka, Pallava, Chola, Chera, and Pandya dynasties during the time of the Gupta Empire era (Akter, 2020). Two prominent landscapes, granitic and basaltic, are distinguished here. The landscape is predominantly shaped by erosion and features broad open valleys (Kale, 2014). The remaining part of the study region covers a portion of the Indus–Gangetic Plains, an extensive low-relief alluvial plain ~3000 km long and 100–500 km wide that is situated between the Great Himalayas and the Indian Peninsula. This was ruled by the Malava, Abhira, Arjunayana, and Yaudheya dynasties in ancient India (Akter, 2020). The northern part of the study region extends to Ladakh, Himachal Pradesh, the higher Himalayas of Uttarakhand, and the Shivalik range of the Nepal-India border. Furthermore, GIS analysis indicated that the Ganga (G) River basin has the highest contribution, occupying 27%, 33%, and 42% of the areas of the SSARW, SSAR, and SSARE, respectively. The contributions of other basins are represented in Fig. A1. An imaginary boundary surrounds the proposed study area, so no literature has presented descriptions of various hydroclimatological variables. We have used IMD-gridded data (a detailed description is presented in the “Data preparation” section) to determine the spatial average of long-term hydroclimatology. The last 30 years of climatology (1993–2022) indicate average annual rainfalls of 870, 940, and 1130 mm received by the SSARW, SSAR, and SSARE, respectively. In summer, the maximum temperatures increase to 37.25, 37.65, and 38 °C in SSARW, SSAR, and SSARE, respectively, while in winter, the temperatures decrease to 12.46, 12.75, and 11.65 °C in SSARW, SSAR, and SSARE, respectively (Fig. A2).

Fig. 1: Location of eight renowned Shiva temples in the study area with their respective contributor dynasties, and the approximate time of construction.

While the blue dots represent the locations of all the other Shiva temples, the locations of the eight famous Shiva temples are differentiated by deep red temple shape marks within the three belts: SSARW (in light magenta), SSAR (in light green), and SSARE (in light yellow).

A detailed literature review is conducted to establish the strong relationship in terms of resource availability between ancient civilizations, Hinduism, Shiva temples, and modern science to strengthen the main hypothesis of the present study in the subsequent three sections below.

Linkages between knowledge of ancient civilization, Hinduism, and modern science

In Hinduism, the sun is seen as the primary energy source, and is interlinked with wind and water, serving as the well-known renewable energy sources that have existed for millions of years. The integration of earth (soil) and sky is believed to form five elements (panchabhuta): space (akasha), air (vayu), water (jala), fire (agni), and earth (prithvi), which are vital for sustaining life in the universe (S. Kumari and Asthana, 2023; Sankari, 2016). These elements are vital in agriculture, which might be the reason that most ancient civilizations were developed along fertile river banks (Kumar Singh et al. 2020; TV, 2016). Ancient civilizations (Arnold, 2016; Avari, 2016; Hordon, 2011) had considerable knowledge of the hydrological cycle, measurement of hydrological variables, design of hydraulic structures, irrigation, and natural extremes (Kumar Singh et al. 2020). In ancient India, Hindu temples not only functioned as places of worship but also as centers of education, culture, and agricultural support, significantly contributing to local socioeconomic development (Dhanorkar, 2017; Ghonge et al. 2021; Pandya, 2014; Parker, 2010; Trouillet P, 2017; Chandrasekaran et al. 2024; Misra and David, 2021; Saravanan, 2018). The Worship of panchabhuta, or nature in temples, has been prevalent since ancient times. The Rig Veda metaphorically depicts these elements as divine elements (Bernali, 2015; Brown, 1965). The practice of nature worship started even before the Vedic era (2000 BC–300 BC) (Bernali, 2015), and has drawn attention from researchers in the field of religious research (Hewamanage and Sang, 2015). According to Hindu mythology, Brahma, Vishnu, and Shiva stand for the creation, operation, and destruction of the universe (Bhandari, 2022; Pandey and Tiwari, 2018; Sharma, 2023), respectively. The existence of Brahma, Vishnu, and Shiva—represented as the trinity in Hinduism—indicates an enormous quantity of energy(Hewamanage and Sang, 2015), believed to be comparable to Project Trinity, which was carried out by the US Defense Nuclear Agency to test nuclear weapons during World War II (Rodriguez-velasco and Jesus, 2011). According to the Hindu religion, Lord Shiva performs his cosmic dance, the tandava, in which he destroys things in order to create (Bhandari, 2022; Godlaski, 2012), which is believed to resemble Albert Einstein’s concept about mass-energy transformation (Hecht, 2009; Rajendran and Kuthirummal, 2022). Attempts have also been made to draw parallels between Shiva’s dance and modern physics, i.e., the vibration of a particle as a result of energy transfer at the subatomic level (Bhandari, 2022). Many Shiva temples were constructed or established throughout India by kings from various dynasties in different regions at different times, yet they. In this study, our objective is to unravel the scientific reasoning behind the eight prominent Shiva temples in India, all constructed before ~1350 CE and aligned along the 79-degree east-to-north connecting Kedarnath to Rameswaram, named the Shiva Shakti Aksh Rekha (SSAR, https://www.india-info.org/temples/snippets_temples_general.html). However, these temples can be classified into two types, jyotirlinga and panchavuta. In Sanskrit (i.e., an archaic Indo-Aryan language that is the classical language of Hinduism), jyotirlinga refers to the distinctive manifestation of energy and panchabhuta signifies the five fundamental constituents of the cosmos that are crucial for supporting life, namely air, water, soil, sun, and atmosphere. These unique characteristics distinguish them from other Shiva temples.

Historical development of Shiva temples in India

Shiva temple construction increased significantly between 700 CE and 1300 CE (Fleming, 2009). influenced by dynasties such as the Pallavas, Cholas, Pandyas, and Vijayanagar rulers (Chandrasekaran et al. 2024; Karashima, 2014). followed by the Pallava, Pandya, and Vijayanagar dynasties (Chandrasekaran et al. 2024). The Indian temple architecture is mainly divided into two parts, Nagara (i.e., North Indian style) and Dravidian (i.e., South Indian style), where major architectural components are Garbhagriha (i.e., where the idol is placed), Gopuram (i.e., the entrance of the temple complex), Shikhara (i.e., a tall tower above garbagriha), Vimana (i.e., garbagriha plus shikhara) and Mandapa (i.e., a pillared hall attached with garbhagriha)(J. Kaur et al. 2024). Those components are introduced by different dynasties.

• 1. Pallavas (300–900 CE) pioneered the Gopuram and Vimana structures (Chandrasekaran et al. 2024).

• 2. Cholas (900–1300 CE) introduced detailed rock carvings, symmetrical precision, sheer scale and bronze sculptures (Chandrasekaran et al. 2024; Ramesh, 2018).

• 3. Pandyas (600–1400 CE) emphasized colorful sculptured Gopurams and elaborate mandapas (Chandrasekaran et al. 2024).

• 4. Vijayanagar rulers (1350–1600 CE) developed large Mandapas with intricate pillars (Branfoot, 2022; Chandrasekaran et al. 2024; Ramesh, 2018).

Linkage between temple and resource availability

A thorough literature review revealed limited direct archeological evidence linking ancient temples to resource availability, but several findings suggest an indirect connection. For instance, carvings of maize and sunflowers in the 12th–13th century Hoysala temples in Karnataka indicate knowledge of agricultural fertility, implying the presence of suitable soil, water, and climate conditions (Johannessen and Parker, 1989; A. K. Singh and Nigam, 2017). Similarly, sculptures of 70 plant species found at the Jambukeswara temple in Tamil Nadu, as well as in other Buddhist, Jain, and Hindu temples, suggest large-scale cultivation in temple lands (Chandrasekaran et al. 2024; A. K. Singh, 2023; A. K. Singh and Nigam, 2017; Trouillet P, 2017). While staple crops like paddy and wheat have left no direct imprints, historical records confirm their significance in both ancient and modern India (Nene, 2005). It is possible that specific crops were cultivated under temple supervision. Since agriculture requires essential resources such as water, solar energy, and labor, these findings suggest a historical link between temple sites and resource abundance in their surrounding regions.

Rationale behind the selection of temples

The rationale for selecting the temples in this study is based on three criteria:

• 1. Temple Prominence and Maintenance: Selected temples have well-defined areas, minimal external land-use interference, and are well-maintained, indicating their historical and spiritual significance. The higher the area, the higher its economic value, and vice versa.

• 2. Architectural Features: Indian temple architecture is mainly divided into two groups: Dravidian style, or South Indian style, and Nagar style, or North Indian style. A temple should have a vimana (i.e., shikhara plus garbhagriha) or gopuram (i.e., a tall, heavy-weight temple entrance tower).

• 3. Symbolism of Water, Energy, and Food: Temples must represent the classical elements/water/energy/food.

All eight temples chosen for this study satisfy these criteria and form the basis for further analysis.

Historical context of the eight temples

• 1. Kedarnath Jyotirlinga (Uttarakhand): The temple is believed to date back over 3000 years based on Mahabharata references, but scientific dating suggests it was built before ~850 CE, as indicated by engraved inscriptions (Chaujar, 2009).

• 2. Ramanathaswamy Jyotirlinga (Tamil Nadu): Linked to the Hindu epic Ramayana and its Mandapa pillars align with Vijayanagar architecture (~1350–1600 CE) (Branfoot, 2022; Henry, 2019; V. Kumar, 2024).

• 3. Jambukeswara (Tamil Nadu): Representing the Jala Lingam (Water) (Sankari, 2016), it dates back to ~500 BCE (A. K. Singh and Nigam, 2017), aligning with the transition from the Megalithic period to the Sangam age (Madhava Soma Sundaram et al. 2020).

• 4. Chidambaram Nataraja (Tamil Nadu): Representing Akasha Lingam (Sky) (Sankari, 2016), early worship of a wooden Shiva idol (~600 CE) evolved into a bronze Nataraja (~700–900 CE)(Srinivasan, 2004).

• 5. Thiruvannamalai (Tamil Nadu): Representing Agni Lingam (Fire) (Sankari, 2016), the temple was built with granite (~900 CE), with later Chola influences (Chandrasekaran et al. 2024; Ravisankar, 2020; Sankar, 2019).

• 6. Ekambaranatha (Tamil Nadu): Representing Prithvi Lingam (Earth) (Sankari, 2016), the temple predates the Pallava period (~600 CE) (Jayapal and Amirtham, 2016).

• 7. Mallikarjuna Swamy Jyotirlinga (Andhra Pradesh): Situated near the Krishna River, it was built by the Cholas (~900 CE) (Karashima, 2014).

• 8. Sri Kalahasti (Andhra Pradesh): Representing Vayu Lingam (Air) (Sankari, 2016), it was likely constructed by the Pallavas (600–850 CE) based on structural comparisons (Mitra and Sadhukhan, 2020).

Data preparation

Data preparation involves analyzing nine parameters: average annual rainfall, length-order density, drainage density, soil-aggregate stability index, annual average soil erosion, cation exchange capacity, trend of ISMR, and minimum and maximum temperature during the Kharif season, are categorized into three groups to derive insights into the spatial status of water availability, soil fertility, and climate variability. Three renewable energy resources, solar, wind, and hydropower, are considered to analyze the spatial distribution of renewable energy resources and their generation potential. The preparation of these parameters requires various types of remote sensing (RS)-based data products. The Shuttle Radar Topographic Mission (SRTM) 90 m DEM (https://csidotinfo.wordpress.com/data/srtm-90m-digital-elevation-database-v4-1/) is used to generate various topographical aspects, such as elevation, slope, and stream network. The physical and chemical soil parameters are downloaded from the International Soil Reference and Information Centre (ISRIC, https://soilgrids.org/) at a 250 m resolution. Daily precipitation data at a ${0.25}^{^\circ }\times {0.25}^{^\circ }$ grid and maximum and minimum temperature data at a ${1}^{^\circ }\times {1}^{^\circ }$ grid resolution were downloaded from the Indian Meteorological Department (IMD, https://www.imdpune.gov.in/). Monthly precipitation with spatial resolution of ${0.5}^{^\circ }\times {0.5}^{^\circ }$ for the last glacial maximum (LGM) and mid-Holocene was obtained from the processed ecoclimate database (https://ecoclimate.lncc.br/about/). The MODIS NDVI product, which has a spatial resolution of 250 m and a temporal resolution of 16 days, was obtained from 2001 to 2022 and analyzed and downloaded from the Google Earth Engine (https://doi.org/10.5067/MODIS/MOD13Q1.061). Monthly global runoff data with 0.5° × 0.5° resolution from 1902 to 2014, developed by researchers (Ghiggi et al. 2019) and available for download from an open-source data archive, were utilized. Daily wind speed data with ${0.5}^{^\circ }\times {0.625}^{^\circ }$ resolution from 1982 to 2022 are downloaded from the Prediction of Worldwide Energy Resources developed by NASA. Average annual global horizontal irradiance (GHI) and temperature data were downloaded from the Global Solar Atlas (GSA) and the National Solar Radiation Database (NSRDB). The results of this analysis are represented by various spatial maps of water, energy, and food productivity indices. While the focus has been on the alignment of eight Shiva temples, the obtained results are further validated by considering 43 additional Shiva temples falling on the three chosen belts. The locations of the Shiva temples were extracted randomly as vector (point) data using Google Earth Pro. The complete analysis was performed in Python 3.10 and ArcMap 10.3.

Methodology

As mentioned previously, this study utilizes nine parameters as selection criteria¹ and conducts a pairwise comparison (AHP) between them to evaluate the weights for each criterion. These weights are subsequently applied in a multi-criteria decision-making (MCDM) technique to create an overlay map that represents the water–food productivity potential index (WFPPI). Hydropower², wind³, and solar⁴ power potential maps are generated to compute the renewable energy potential index (REPI) using the equal-weighted overlay method, which further overlays the WFPPI and generates a water–energy–food productivity potential index (WEFPPI) map for the study region. The study further evaluated the reliability of the WFPPI, REPI, and WEFPPI maps using ground truth data. The detailed methodological flowchart is depicted in Fig. 2. Further, an extensive literature review is conducted to understand the influence of wealth, construction material, spirituality, social and political situation on the Shiva temple construction. While all data sets used in this study can represent topography, morphology, land cover and climatology of the recent past, obtaining the data pertaining to the ancient environment is difficult. However, a detailed literature survey and statistical analysis are carried out to understand the topography, morphology (see “Archeological evidence of ancient river valley civilization” for details), and land cover of ancient times⁵. The paleoclimatic analysis is also conducted to analyze ancient climatology⁶.

Fig. 2: Workflow diagram of the proposed framework.

The five panchabhuta elements (P1–P5) are the main sources of water, energy, food productivity, and climate variability, where shakti, or energy, is represented by solar, wind, or hydropower potential, which primarily depend on the sun, air, and water, respectively. The REPI is computed using S1–S3 via the equally weighted overlay method. Another nine secondary parameters (S4–S12) are used in the AHP-MCDM technique to generate the WFPPI, which is further overlayed with the REPI to generate the WEFPPI.

Development of the water and food productivity potential index (WFPPI)

Three water availability parameters⁷, long-term average annual rainfall (P_a), length-order density (LORD), and drainage density (D_d)(Arabameri et al. 2018; Das and Vema, 2023; B. Ghosh and Mukhopadhyay, 2021; Meshram et al. 2023), are employed to capture spatial water availability. P_a represents the average annual atmospheric water that a region receives in the form of precipitation. On the other hand, a high D_d represents a greater availability of river water without considering stream order. While watersheds with higher drainage density may indicate greater availability of water (Dragičević et al. 2019), this definition alone is not sufficient to provide a clear idea about the actual quantity of available water. To address this issue, we introduce a novel morphological parameter called the length-order density (LORD), developed based on the continuity equation. The concept behind this parameter is that higher-order streams are formed by the confluence of several lower-order streams, resulting in a larger volume of water compared to lower-order streams. The fertility of the soil⁸ was estimated using the annual average soil loss (Soil_loss) (Renard and Ferreira, 1993), soil aggregate stability index (SSI) (Nguemezi et al. 2020), and cation exchange capacity (CEC) (Ketterings et al. 2007). The Mann‒Kendall (MK) (Junaid and T, 2021) test was used to calculate the trends in the Indian Summer Monsoon Rainfall (ISM_prc), minimum temperature (T_min), and maximum temperature (T_max) during the Kharif season to capture the spatial variability in climate⁹.

Development of an AHP matrix

The analytical hierarchy process (AHP) is one of the most widely used consistent weight generation techniques; this method involves pairwise comparisons between criteria (Raju et al. 2020; Tian et al. 2018). Questionnaire surveys are conducted involving a total of 70 pilgrims and 58 villagers (i.e., decision-makers and experts) from three temples of Tamil Nadu and Kedarnath. A pairwise comparison matrix (A) is formed based on subjective judgments of decision-makers and Saaty’s nine-point rule (Saaty, 1990). One could download the questionnaire database using the given link in the “Data availability” section.

$$A=\,\left[{a}_{i,j}\right]$$

(1)

${a}_{i,j}$ represents the importance of the ith criterion to the jth criterion. where the rows and columns are represented by i and j, respectively. Weights (w) for each criterion are generated from the AHP matrix using the eigenvector method (S. Sarkar and Biswas, 2022).

$${k}_{i,j}={a}_{i,j}/\mathop{\sum }\limits_{i=1}^{n}{a}_{i,j}$$

(2)

$${w}_{j}=\mathop{\sum }\limits_{j=1}^{n}{k}_{i,j}/n$$

(3)

$$A\times w={\lambda }_{p}\times w{,{here}\,\lambda}_{p}{\rm{represent}}\;{\rm{principal}}\; {\rm{eigenvalue}}$$

(4)

The ratio between the consistency index (CI) and the random consistency index (RI) reflects the consistency of the proposed AHP matrix. If the consistency ratio (CR) is ≤0.1, the evaluated weights can be used for further analysis; if not, the entire process is repeated until the CR becomes less than or equal to 0.1 (Arabameri et al. 2018).

$${{{CI}}}=\,\frac{{\lambda}_{{{p}}}-n}{n-1},\,n\,{{{is}}\; {{the}}\; {{number}}\; {{of}}\; {{parameters}}}$$

(5)

Multi-criteria decision making (MCDM)

Weighted overlay analysis (WOA) is one of the most popular MCDM techniques due to its mathematical simplicity, and it is used in various disciplines to make decisions by considering several criteria (Al-Anbari et al. 2018; Das and Vema, 2023; B. Ghosh and Mukhopadhyay, 2021). The WOA reclassifies parameters based on their capacity to represent the best alternative. For example, annual precipitation is classified into four classes using the quantile method, where the 3rd to 4th quantiles represent the maximum rainfall value and the lowest probability of facing water stress, respectively, and are assigned the highest rank, while the 1st to 2nd quantiles are assigned the lowest rank. The classification of all the parameters is presented in Supplementary Table A1. The weighted overlay index for each pixel was calculated using the following equation (Eq. (6)), where I_{WFPPI, j} is the WFPPI for the jth pixel, w_i is the weight for the ith criterion (i.e., parameter), and a_{i, j} represents the reclassified value of the jth pixel in the ith criterion. A higher I_WFPPI indicates a high potential for water availability and food productivity.

$${\rm{I}}_{{\rm{WFPPI}},{\rm{j}}}=\mathop{\sum }\limits_{{\rm{i}}=1}^{\rm{n}}{\rm{w}}_{\rm{i}}\times {\rm{a}}_{\rm{i},{\rm{j}}}$$

(6)

The WFPPI was classified using the Gaussian distribution or normal distribution approach. The ranges are defined as follows: 0 to μ−2*(σ): very low (VL); μ−2*(σ) to μ−σ: low (L); μ−σ to μ: moderate (M); μ to μ + (σ): high (H); μ + (σ) to μ + 2*(σ): very high (VH); and μ + 2*(σ) <: extremely high (EH). If the spatial distribution of the WFPPI did not follow a normal distribution, the quantile approach was used for classification.

Renewable energy potential index (REPI)

Three types of renewable energy are considered in the renewable energy potential index (REPI)¹⁰. The theoretical potential of hydropower and wind power is calculated by considering a standardized specification of mechanical equipment and theoretical concepts presented in various studies (Arthur et al. 2020; Jain et al. 2020; Ragheb and Ragheb, 2011; Ohunakin and Akinnawonu, 2012; Tefera and Kasiviswanathan, 2022). The solar power potential⁴ is computed based on the power generation capacity of a PV cell in a particular region. The global horizontal irradiance (GHI), the temperature coefficient of maximum power (K_P), the maximum power under standard test conditions (P_max (STC)) and the module temperature (T_m) (Al-Ezzi and Ansari, 2022; Jain et al. 2020) are the major influencing factors. To date, most studies have considered the long-term daily average GHI (Jain et al. 2020). The estimate of the long-term average daily maximum GHI is used to determine the maximum power generation capacity of a PV cell (Fig. A3c). A renewable energy potential index (REPI) map is generated by overlaying a reclassified spatial map of solar, wind, and hydropower potential using equal weights. Higher REPI values indicate greater potential for producing renewable energy, and vice versa. The water energy food productivity potential index (WEFPPI) is derived by superimposing the WFPPI on the REPI. A weight of 0.5 is assigned to each index, indicating equal importance for both energy generation and water–food productivity. The classification technique proposed for WFPPI is used for REPI and WEFPPI.

The area under the curve (AUC)¹¹ of the receiver operating characteristic (ROC) curve was used to validate the spatial distributions of water, energy, and food productivity potential in the three belts. Furthermore, stakeholders are often interested in the highest potential locations, however, cumulatively the high (H), very high (VH), and extremely high (EH) potential regions cover 56.6% study area, hence choosing a location within a large extent for development activity is a difficult task and reduces the willingness of Investors. Where the very high and extremely high pinpoint the highest 12% potential region, reduces the degree of uncertainty in selecting a suitable location for development activity. Hence, each index map underwent binary classification into two distinct categories: (1) Positive: This class included extremely high (EH) and very high (VH) potential; (2) Negative: This class included very low (VL), low (L), moderate (M), and high (H) potential. Based on the hypothesis of the present study, the sites of Shiva temples may have abundant water resources, high energy generation, and high food productivity potential; these data are utilized for ground truth validation. A higher AUC indicates a robust link between water availability, energy generation, food productivity potential, and the location of ancient Shiva temples throughout the study region. We computed the area under the curve (AUC) for three previously obtained indices in the present study, which are discussed in the “Results and discussion” section.

Influence of wealth, construction material, spirituality, political and social pivots

In ancient India, the location and growth of temple-oriented activities mainly depended upon several factors, such as the wealth of the ruling dynasty, spirituality, political and social pivots, and construction materials(Mitra and Sadhukhan, 2020), which are discussed below.

• 1. Availability of building materials: The south Indian states have abundant granite resources(Indian Bureau of Mines, 1990; Mahajan and Malokar, 2024; Ronald et al. 2018), which were preferred for temple construction due to their durability, weather resistance, and esthetic appeal(Bharasa and Gayen, 2021). In contrast, northern and central Indian temples were often built using basalt and sandstone(G. Kaur et al. 2019). Additionally, seismic activity in northern India(Ruchi et al. 2021) may have discouraged the construction of tall and heavy structures.

• 2. Wealth of the ruling dynasty: Southern dynasties, such as the Cholas, Pandya, and Pallavas, had strong economic foundations, supported by trade and maritime connections(Rajagopal et al. 2019). This financial stability enabled large-scale temple construction. Northern dynasties, while militarily powerful, faced frequent invasions(Hassan Aali, 2015; M. Kumari, 2022; K. Roy, 2021; Sinha, 1955), which may have redirected resources away from temple architecture.

• 3. Spirituality, politics, and social pivots: The Chola rulers were deeply influenced by Shaivism, leading to the construction of many Shiva temples(Karashima, 2014), followed by the Pallava, Pandya, and Vijayanagar dynasties(Chandrasekaran et al. 2024). Temples were also used as political and economic centers, reinforcing the authority of rulers and integrating local communities(Balambal, 2005; Heitzman, 1991; Karashima, 2014). Invasions by Afghan, Mughal, and Khalji rulers led to the destruction of many northern temples, further contributing to the disparity in temple distribution(S. N. Khan and Bilal, 2020; Zarin Ansari et al. 2018).

These factors collectively suggest that South India had more favorable conditions for temple architecture, explaining the high concentration of temples below the Deccan Plateau.

Archeological evidence of ancient river valley civilization

Rivers have made significant contributions to the growth and preservation of human civilization from ancient times. Archeological evidence suggests that the Harappan, Mohenjo-Daro, Mesopotamia, and Egyptian civilizations evolved along the fertile banks of the Indus, Tigris-Euphrates, and Nile rivers, respectively(Jansen, 1993; Macklin and Lewin, 2015; Miller, 1984; R. Pal, 2018). Similarly, in the southern SSAR belt, the Archeological Survey of India (ASI) discovered an ancient Tamil civilization of 200 BC, near the bank of Vaigai River Valley of Tamil Nadu, interestingly, its timeline coincides with the Sangam dynasty’s period (an ancient Tamil civilization) (Ramakrishna et al. 2018). Some studies strongly argue that a Neolithic civilization also developed across the Vaigai River Basin around 5511–5147BC (Ramkumar et al. 2022). Recent archeological excavation of Adichanallur in Tamil Nadu claimed the existence of an ancient Tamil civilization on the bank of the Porunai river between 1260 and 1051 BC (Tamil Nadu Department of Archeology, 2021). Similarly, in north India, 10-8 thousand years old Mesolithic artifacts and 45,000 years old Middle Paleolithic artifacts of Jalaun district claimed ancient civilization’s existence in the northern SSAR belt (I. B. Singh, 2005). Evidence of all those river valley civilizations is buried and well-preserved around their place of origin from at least 7000 to 200 BC. All this archeological evidence indicates that there was no abrupt transformation of the river course or change in topography and geomorphology within the study’s time frame. Additionally, geomorphological studies indicate that the Indian subcontinent’s topography was shaped between the Precambrian and Quaternary periods, spanning 1000 to 1 million years ago, far beyond the scope of this study (Kale, 2014). Based on the above information, the river courses are found to be stable, at least in the present study timeline.

Results and discussion

WFPPI map constructed using the WOA

We analyzed the spatial variability in the water-food productivity potential across the three belts using the AHP and WOA, considering nine parameters as criteria. Their spatial representation is depicted in Figs. A4, A8, A9, and A13, A14. IMD gridded data are utilized to calculate the average annual precipitation (P₃₀) for each 30-year time horizon between 1903 and 2022 to determine the atmospheric water contribution to surface water resources (Fig. A4). A strong linear relationship between the P₃₀ in various climate periods and the annual average precipitation map suggested by earlier research (Fig. A5) supports the accuracy of the spatial maps (Rao and Mqoley, 1970). The 120-year average annual precipitation (P₁₂₀) is considered to determine the spatial distribution of annual precipitation across various river basins within the study area. The P₁₂₀ in the Western Ghats basin (WG) of the SSARW is 2270 ± 734 mm, of which the Arabian Sea branch of the Indian Summer Monsoon Rainfall (ISMR) accounts for 70–80% (Dixit and Tandon, 2016; S. Ghosh et al. 2016). For the majority of the river basins in the SSARW belt, P₁₂₀ is between 599 ± 73 and 950 ± 193 mm. Although the Ganga (G) basin receives less rainfall than the WG and Extreme South basins (ES), it has the highest volume of precipitation, ~156 ± 50 BCM (Fig. A6c), because of its highest area coverage in the SSARW (Fig. A1). While the WG and ES contributed just 5% and 2% (Fig. A1), respectively, to the SSARW, the spatial variability in rainfall in these basins was significantly greater than that in other river basins (Fig. A6a), indicating a discontinuity in the western ghat wall effect. In the northern region, during the winter, the westerly jet stream is divided into two parts by the Tibetan Plateau (TP) (Liang et al. 2015). As a result, the Indus and North Himalayan basin (I&NH) receives significantly high and low amounts of rainfall, with a 60% coefficient of variance (C_V) (Fig. A6b). Overall, the entire SSARW region receives an annual rainfall of 861 ± 470 mm, equivalent to 655 ± 358 BCM of water (Fig. A6j and l). Most of the basins of the SSAR belt receive annual rainfall between 787 ± 118 and 1302 ± 156 mm, except for the I&NH River basin, which is influenced by the TP and receives very little rainfall of 390 ± 185 mm (Fig. A6d). Annually, the SSAR belt receives an average rainfall of 960 ± 324, which provides 592 ± 200 BCM of fresh water (Fig. A6j and l). The interbasin variability in long-term average annual rainfall decreases as we move from west to east (i.e., from SSARW to SSARE), and this variability varies linearly with basin area (Figs. A6k and 7), likely due to two factors: (1) The distance from the western ghat increases, weakening the rainfall shadowing effect (Ashfaq, 2020), which leads to a more uniform distribution of rainfall. (2) The SSARW belt covers the highest area of the TP, followed by the SSAR and SSARE belts. As the contribution of the TP decreases, the spatial variability also decreases in the northern region. On average, annually, 1188 ± 199 mm of rainfall is received by the SSARE belt (Fig. A6j), which supplies 393 ± 66 BCM (Fig. A6l) of water to the Ganga (G), Godavari (GO), Krishna (K), Mahanadi (M), Narmada–Tapti–Mahi (NTM), and South East (SE) River basins. Lumped analysis indicated that the SSARE belt had the highest P₁₂₀ and lowest C_V, whereas the SSARW belt had the largest volume of precipitation, and the SSAR belt was between these two belts. As a result, the probability of supply-side water stress in the SSARE belt is lower, followed by that in the SSAR and SSARW belts. The same is indicated by the spatial map of P₃₀ for the four different climate periods (Fig. A4). This large quantity of precipitation generates a significant volume of runoff flowing through natural drainage or river systems. Annually, the Indian River system contributes 1953 km³ of surface water, of which 690 km³ of water is utilizable (R. Kumar et al. 2005). As the proposed study area covers 50% of the Indian subcontinent, it can generate 976 km³/year of water, of which 345 km³/year is utilizable. Drainage patterns play a significant role in the spatial variability of flow regimes. The confusion of large numbers of drainages represents high drainage density and higher flow (Chenini et al. 2010). A threshold of 1 × 10⁵ is assigned to the flow accumulation raster to generate a river network with a sufficient drainage area, and the drainage density is computed for each individual pixel by considering a circular area with a 40 km radius around each pixel. The statistics indicate that there is no significant difference between the average Dd and its spatial variability within the three belts (Table A3). The geographical location of the G basin and topographical features of the Himalayas influence most of the tributaries of the G River to confluence within the stretch of the SSAR belt, which generates the highest Dd of 1.26 km/km² compared to that of other belts (S. Singh et al. 2019) (Fig. A8c). Additionally, 63%, 82%, and 97% of the basins of the SSARW, SSAR, and SSARE belts contributed to the Bay of Bengal (BOB), while 32%, 14%, and 3% of the area contributed to the Arabian Sea (AS), respectively. The high contribution in the Bay of Bengal indicates that as the river system approaches the east, the order of streams increases. The G and GO basins formed the highest-order streams within the SSAR belt, which overlapped with the high D_d, leading to a higher LORD in the SSAR belt than in the other belts (Fig. A8d).

Along with water availability, soil fertility is another important parameter that strongly influences food production potential. Three parameters are considered in this study: Soil_loss, SSI, and CEC to counteract soil fertility. The RUSLE model is employed in this study due to its simplicity and ability to capture pixel-by-pixel water-induced soil erosion potential. The spatial average soil erosion potentials within the SSARW, SSAR, and SSARE were 35, 32, and 28 ton/ha/year, respectively, with a greater width of the uncertainty band and a C_V of 140% for all belts, reflecting the significant spatial heterogeneity of soil erosion. The accuracy of the RUSLE model in explaining the state of soil erosion has been demonstrated by the similar spatial patterns of soil erosion observed in prior research (S. C. Pal et al. 2021). Since the LS and P-factors are functions of slope, increasing slope bias within the RUSLE model leads to extremely severe soil erosion potential in mountainous regions (Fig. A9). Despite being a nonlinear function of soil texture (Das and Vema, 2023), the K factor exhibits a good linear relationship with the amount of silt and sand in the soil (Fig. A10b and c). With a high silt content and little sand, the central area mostly exhibits high soil erodibility, which enhances soil erosion. The Himalayan range contains 40% of the soil organic carbon (SOC) of the Indian subcontinent, and a lower temperature reduces the rate of SOC decomposition (Bhattacharyya et al. 2000), which explains the substantial geographical diversity of the SOC content, with 75% C_V. The influence of clay and silt content in SSI is not significant because of less spatial heterogeneity (Fig. A10g and h, Table A3). Within the study area, the Himalayas and some regions of the Western Ghats have the highest SSIs. The SSI statistics reveal that the SSAR and SSARW belts have larger SSIs—6.7 ± 5.3% and 6.62 ± 5.4%, respectively—due to the presence of the Himalayas in the northern region. A good linear relationship between the clay content and CEC indicates the ability of negatively charged clay particles to absorb positively charged cations (Ketterings et al. 2007), which increases the CEC concentration in the central part of the study region (Fig. A8a). The spatial distribution of CEC showed the absence of a significant difference between the spatial heterogeneity of the three belts (Table A3).

The food production in India during the Kharif season is mostly dependent on the ISMR (Bapuji Rao et al. 2014). Cyclonic circulation brings precipitation to monsoon-prone areas of the central Indian basin, including the GO, C, and K River basins. In the recent climate period (1993–2022), the ISMR (in June, July, August, and September) exhibited a declining trend in the northern region, while in the southern region, it increased (Fig. A13d). This indicates an increase in the frequency of extreme precipitation (Chaluvadi et al. 2021) and drought (Parajuli et al. 2021; Ul Shafiq et al. 2019) in the southern and northern regions, respectively. The trend of the ISRM during the preceding three climatic periods (1903–1932, 1933–1962, and 1963–1992) indicates that low rainfall occurred in the regions of major surplus basins, such as M, GO and WG, while the G basin exhibited an increasing trend with 1–5% significance (Fig. A3a–c). This could be the result of the formation of an anticyclonic circulation over central India, which led to a divergence of moisture flux (S. Ghosh et al. 2016). Stable temperatures play a crucial role in agricultural yield. Previous research carried out by Bapuji Rao et al. (2014) concluded that an increase in the minimum temperature would decrease the Kharif-paddy yield. This study argues that although the trends of minimum and maximum temperatures in the Kharif session are highly heterogeneous in space–time, the extreme southern and northern parts of the study region show increasing trends for both minimum and maximum temperatures in all climate periods except 1951 to 1962 (Fig. A14). Hence, it is obvious that the average temperature follows an increasing trend, as supported by the findings of previous studies (Bal et al. 2016). The increasing trend in temperature clearly indicates the occurrence of climate change in this region. However, the degree of similarity between ancient and present climate, land use, and topography is essential to establish a relationship between ancient temples and the water–energy–food nexus of the recent past.

The climate over the Indian sub-continent is highly influenced by its geographical location and topography. The topography we observed today was formed between the Precambrian and Quaternary periods around thousands of millions to one million years ago (Kale, 2014) and is far from the timeline of the present study. Furthermore, archeological evidence of various river bank civilizations argued that the river courses are stable at least from ~5000 BC. The lumped analysis indicates a weakened phase ISMR with a reduction of annual precipitation by 106 mm in the Last Glacial Maximum (LGM) (Rashid et al. 2011). In the mid-Holocene period, strong ISMR increased average annual precipitation by 113 (Anoop et al. 2013; Gupta et al. 2024). Where the coefficient of variance (C_V) in the LGM and mHol along the SSARW, SSAR, and SSARE are 58.2–54.9%, 38.7–40.7%, and 17.0–15.5%, respectively. Reduction in C_V from west to east and spatial correlation coefficient of 0.88 (p-value ~ 0) indicate high similarity between precipitation patterns in the ancient and previous centuries (Fig. A12). Earlier studies claimed that between 1 and 1200 CE, the ISMR anomaly average yielded above-normal precipitation in Tamil Nadu, some parts of Telangana and Andhra Pradesh, and Uttarakhand, where the eight famous temples of Lord Shiva are currently located. At the same time, below-normal rainfall is observed in some regions of Uttar Pradesh (UP) and Madhya Pradesh (MP) along the SSAR belt (A. Sarkar et al. 2024). Rainfall and anthropogenic activities (i.e., LULC change) are two major factors responsible for soil erosion. Between 1 and 1500 CE, forest cover and density in India were ~1.98 × 10⁶ km² and ~0.6 km²/km² started declining continuously under the stress of anthropogenic activities and reached ~0.81×10⁶ km² and ~0.25 km²/km² by the year 2023 (Fig. A15). The same is reported in the Indian State of Forest 2023 (Ministry of Environment Forest and Climate Change, 2023). This reduction was influenced by the increased cropland density from ~0.2 to ~0.64 km²/km². Dense forest cover could reduce soil erosion in ancient times compared to the previous century.

The classification of parameters is represented in Table A1. We assign an importance factor to each pair of parameters in the AHP matrix, taking into account the knowledge of ancient civilizations in water resource management and food production, as presented in various studies and expert judgments (Table A2). The maximum eigenvalue of 10 is computed to calculate the consistency index (CI), and a value of 1.46 is chosen for the random index (RI) based on the number of criteria or parameters. Finally, we compute a consistency ratio (CR) of 8.9% by taking the ratio between the CI and RI; since it falls below 10%, the consistency is satisfied, allowing us to further utilize weights in the WOA. The weights generated from the AHP matrix indicate the influence of each parameter on water availability and food productivity potential, with P_a having the highest contribution of 26.5%, followed by LORD, D_d, Soil_loss, SSI, CEC, ISM_prc, T_max, and T_min. The final WFPPI for the four different climate periods indicated that the SSAR belt had 1.75–2.13 and 9.14–9.37 Mha of EH and VH water–food productivity potential regions, respectively, in all climate periods except 1933–1962 (Fig. 3a). This may be due to a reduction in annual rainfall across the east coast region of Tamil Nadu (Fig. A4b). Annual rainfall plays an important role in the production of kharif and rabi staple foods (i.e., rice) in India. The previous 10-year annual average food production record of India indicates that the largest planted area and highest yield were generated from rice cultivation, followed by wheat, corn, and millet (IPAD, https://ipad.fas.usda.gov/). Cumulatively, the EH and VH water–food productivity potential regions cover a 25.51 Mha area and have the capacity to produce 102 × 10⁶ tons of rice, accounting for approximately 57% of India’s annual rice production, with shares of 24.6%, 17.1%, and 15.3% belonging to the SSAR, SSARW, and SSARE belts, respectively.

Fig. 3: Bar diagram of various potential classes of the WFPPI and WEFPPI.

The geographical distribution of various potential classes (i.e., VL: Very Low, L: Low, M: Moderate, H: High, VH: Very High and EH: Extremely High) of a WFPPI and b WEFPPI within the SSARW (R1), SSAR (R2), and SSARE (R3) belts, where C1 (1993–2022), C2 (1963–1992), C3 (1933–1962), and C4 (1903–1932) represent the four climate periods. The bar diagram represents the potential areas of a water–food productivity and b water–energy–food productivity in million hectares (Mha).

REPI and renewable energy potential

The study estimates the theoretical maximum power outputs of a wind turbine, hydropower, and solar PV cell to examine the regional distribution of RE. The Modern-Era Retrospective Analysis for Research and Applications, version 2 (MERRA-2) reanalysis wind speed product with a 0.5° × 0.625° spatial resolution available in the Prediction of Worldwide Energy Resources (POWER) portal is used to calculate the long-term annual and daily average gridwise power generation capacity of a wind turbine. Although technological advancements have enhanced the installed capacity of wind turbines, there is still a lack of records worldwide. Therefore, we consider average dimensions, such as rotor diameter and hub height, from the records of wind turbines in India for the financial year (FY) 2017–2018 (Charles Rajesh Kumar et al. 2019) rather than from any specific wind turbine model. The best PV cell deployed in this study area, after comparing various parameters of PV cells listed in the global PV cell database, was used (Belsky et al. 2022) to analyze the geographical distribution of PV power production. In addition, hydropower is highly spatially constrained, as it exists only along streams. Moreover, it has a higher capacity factor than wind turbines and solar PV cells (Petrakopoulou et al. 2016). A monthly global runoff time series with a 0.5° × 0.5° spatial resolution is employed to compute 90% dependable flow at the location with a minimum power generation capacity of 50 kW. Spatial maps of the three RE potentials are further classified based on spatial statistics and expert opinion. In India, hydropower plants with a maximum capacity of 25 MW are considered to be small hydropower plants (SHPs). Our study revealed that the proposed region has 9383 potential SHP plant locations with a 22.4 GW power generation capacity, which highly correlates with the potential for small hydropower generation in India estimated by the Ministry of New and Renewable Energy (MNRE)(Mishra et al. 2015; MNRE, 2017). 8.63 and 13.05 GW potential are concentrated along the SSAR and SSARW belts, respectively. 413 potential sites are located within the SSARE belt, having a potential of 730 MW. The MNRE’s projected state-level potential for SHP in India is used to validate the possible potential of the field. The proposed SHP potential and the number of potential locations (states with more than 90% of the area within the study region considered for comparison) show a good relationship with an estimated value reported in the state-level MNRE annual report for 2022–2023 (MNRE, 2023), with a Pearson correlation coefficient (Sedgwick, 2012) (PCC) of 0.85 and 0.82 and a p-value of <0.05. The capacity of possible hydropower locations decreases as we move eastward (Figs. 4c, A16), which is primarily influenced by topography and river flow. A PCC > 0.75 with a p-value of 0.001 indicates the presence of a causal relationship between river flow and hydropower capacity, which also indicates that most of the river basins have flow-induced hydropower potential (Fig. A17a and b). Head-induced hydropower potential is observed in the SSAR-occupied K basin, the SSARE-occupied NTM basin and the SE River basin (Fig. A17c and d). The rugged terrain of the Himalayas, combined with an abundance of water from snow and glacial ice (Li et al. 2016), generates significant hydropower potential (Schwanghart et al. 2018). However, in the southern region, the high rainfall and elevated western Ghats provide significant hydropower potential (Khare et al. 2023). This may be the case; our results suggested 5629 possible hydropower locations with a cumulative capacity of 26.3 GW in the SSARW belt. The lack of a high head and notably low flow velocity in the eastern river system creates large fertile deltas along the coast of the BOB (Ramkumar et al. 2016), which may be the cause of extensive rice cultivation (Gurunadha Rao et al. 2011; Wischnath and Buhaug, 2014) and concurrently decrease hydropower generation potential. With these, the SSAR belt contributes 13.9 GW of capacity and 3761 potential hydropower sites. In the SSARE belt, the flat terrain of the Indo-Gangetic floodplain (IGP) (Kanhaiya et al. 2017) prevents it from producing enough water flow and head, and the water resources of the eastern coastal region and central India are primarily dependent on precipitation. In this area, rainfall-induced river flow is essential for producing hydroelectric power (Khare et al. 2023). All these factors could explain why the hydropower potential of the SSARE belt was lower than that of the other two belts. Simultaneously, it spreads along the most downstream region of the G and Go Basins, where it receives a high flow volume. The presence of a sufficient head along the stream and the contribution of the large upstream basin area produced ~98% of the hydropower capacity of the SSARE belt (Fig. A17e). The majority of the possible hydropower locations can be found in the lower-order nonperennial stream of the SSARE-occupied G, upper M, upper NTM, and lower GO basins (Fig. A17e).

Fig. 4: Power generation potential across the three belts for different renewable energy sources and potential classes.

a Solar, b wind, c small hydropower, d total (solar + wind + small hydro), e very low potential, f low potential, g medium potential, h high potential, i very high potential, and j extremely high potential.

Researchers have conducted studies to determine the geographic distribution of photovoltaic cell power generation capacity in India, and their findings suggest that, in addition to hydropower, the southern, western, and interior regions of the country are conducive to the development of solar and wind power plants (Jain et al. 2020; Lu et al. 2020). In December 2013, the Power Grid Corporation of India Ltd. (PGCIL) released a report concentrating on harvesting solar and wind power from hot and cold desert regions, including Rajasthan’s Thar Desert, Gujarat’s Rann of Kutch, Jammu & Kashmir’s Ladakh region, and Himachal Pradesh’s Lahul and Spiti valley. By 2050, the generation of 220–450 GW of power is possible from solar and wind power on just 5–10% of the wasteland to meet the growing demand (PGCIL, 2013). The findings of the present study strongly support the results of previous research as well as government policies (Jain et al. 2020; Lu et al. 2020; NITI Aayog, 2015; PGCIL, 2013). The spatial map of solar and wind power potential indicates that the coastal regions of Tamil Nadu and Ladakh (Lohan and Sharma, 2012; Palaneeswari, 2018) have extremely high power generation potential, where a single wind turbine and a 380 W_P (watt peak) PV module can produce a minimum of 2.11 MW and 280 W_P (at 1.30 p.m., under maximum GHI) power, respectively. The first criterion for choosing a location for wind or solar power plants could be the power generation potential; high-potential zones are always of interest to investors (Malik et al. 2020). To analyze the practical implementation of solar and wind power plants, we make a few assumptions: (1) Only 5% of an area (PGCIL, 2013) from an extremely high and very high-power potential zone with a capacity factor (CF) >70% is considered suitable for installing a power plant. (2) For the wind power plant, we deploy four wind turbines within a 250 m × 250 m area. In reality, it is not possible to reach 70% CF for a wind or solar power plant. As mentioned earlier, the estimate is limited to the theoretical power generation potential (Trenkel-Lopez and Matthews, 2018). Utilizing the above assumptions in the proposed framework, we determine 580 GW_P of solar power and 242 GW of wind power potential within the study region, which is equal to 77% and 34%, respectively, of the potential of solar and wind power in India estimated by the MNRE (2023). It is also noted that out of the total power generation, the SSAR belt has 72% and 69% shares of solar and wind power, respectively, followed by the SSARW and SSARE belts, respectively. This study revealed sufficient energy potential in these three belts, which could support India’s 500 GW RE-dream by 2030 (NITI Aayog, 2015), where the SSAR belt alone could provide valuable contributions. However, a thorough analysis is necessary to comprehend the practical implementation of renewable energy facilities, including the aspects of political, economic, environmental, and other constraints (David and Brian Vad, 2014; Jain et al. 2020; A. Kumar et al. 2010; Malik et al. 2020; Poullikkas, 2009; Taleghani and Kia, 2005; Tefera and Kasiviswanathan, 2022; Ye et al. 2017).

We also analyze the potential for wind and solar power using the gridded approach in addition to evaluating the potential for hydropower generation through a location-based study. The overlay analyses, which represent each parameter as a layer, find that single-point data are less useful due to their lack of spatial coverage. We compute the density of the hydropower potential to express the hydropower potential as a function of space. Classified maps of solar, wind, and hydropower density are employed in an equally weighted overlay analysis to generate the REPI by eliminating bias between the three REs. Using the normal distribution classification method, we further classified the REPI into six classes. A region with a high REPI value indicates a high RE potential, and vice versa. The geographic heterogeneity of the REPI reveals that the enormous potential of solar and wind power generation along the SSAR belt (Fig. 4a, b) encompasses a 3.4 Mha area with an extremely high REPI (Fig. 4j). After the SSAR belt, the SSARW belt has the second-highest potential for producing solar and wind energy. It also has the highest potential for hydropower (Figs. 4c, A16), which has very high and high RE potentials over 3.6 and 50.7 Mha of land in the belt, respectively (Fig. 4h, i).

WFPPI and REPI are used to create the final WEFPPI map for each climate period (Fig. 5f–i). From the previous discussion, it is evident that the SSAR belt has a significantly high solar and wind power generation potential; consequently, it has extremely high and very high water–food productivity potential. As a result, this region contains extremely high to very high potential for water–energy–food productivity in all climate periods (Fig. 3b).

Fig. 5: Spatial maps of four climate periods.

a–d WFPPI, e REPI, and f–i WEFPPI. VL: very low, L: low, M: moderate, H: high, VH: very high and EH: extremely high potential.

Performance evolution of the water, energy, and food productivity indices

The WFPPI, REPI and WEFPPI were validated against the locations of the saliva temples, and the area under the curve (AUC) values for all the climate periods ranged from 0.84 to 0.89 (Fig. 6a, c, Table A4). These values demonstrate the existence of positive correlations between water, energy, food, and the placement of ancient Shiva temples along the SSAR belt, as well as how well those three indices explain positive and negative potential classes. Moreover, the absence of a significant difference in the area under the curve (AUC) indicates that temporal variation was not highly significant in the present analysis. This could be attributed to several reasons; for instance, the spatial distribution of precipitation did not significantly change within the four climate periods, as we assigned the highest weight to annual precipitation because water is a crucial factor for food production in ancient India. Additionally, the RUSLE-R factor, which is a function of monthly and annual rainfall, also maintains the same spatial distribution during all climate periods.

Fig. 6: The ROC curve and AUC for different climate periods.

a WFPPI, b REPI, and c WEFPPI.

Conclusion

Although a thorough literature review revealed limited archeological evidence directly linking ancient temples to resource availability. The findings of the present study suggest that the site of ancient Shiva temples is strongly influenced by soil fertility, water availability and energy potential, which in turn improves socioeconomic development in terms of high food production and technological advancement. Furthermore, notable similarities between ancient and modern climatic, topography, and geomorphological features support the hypothesis of the study. It is well understood that ancient civilizations acquired knowledge of various aspects of water resources (Kumar Singh et al. 2020; TV, 2016). Moreover, renewable energy drivers such as blowing wind, flowing water, and incident solar radiation are of concern (Delyannis, 2003; Malville and Singh, 1995; Parmar, 2023; Shamasastry, 2015), but very limited scientific literature explains the utilization of these drivers in large-scale agriculture and power generation in ancient times (Sorensen, 1995). As a number of empires have ruled in India for various centuries, they suppressed the flow of knowledge from one century to another. Based on the findings of the present study, we can infer that the availability of abundant resources for food, water, and energy across the SSAR belt played a crucial role in the socioeconomic development of the Indian subcontinent in ancient times, which influenced ancient decision-makers to build all these expensive and famous Shiva temples across the SSAR belt with the objective of improving regional socioeconomic status. Besides water, energy, and food, several other factors are essential for temple construction activity in ancient times, such as the wealth of the ruling dynasty, spirituality, political and social pivots, and construction materials (Mitra and Sadhukhan, 2020). Availability of those factors in ancient South India enhanced their architectural activities compared to other parts of the country, and that could be the reason for the clustering of temples below the Deccan plateau. The enormous potential of RE and its smooth transition from fossil fuels have strengthened the opportunity for new jobs in India (Muttitt and Gass, 2023; Rajesh Kumar and Majid, 2020). Consistent policy measures with transparent linkages between administrations and investors are key parameters for presenting India as a global leader in the renewable energy sector in the coming decades (Rajesh Kumar and Majid, 2020). These index maps provide insights into historical and existing resource availability, which can assist decision-makers, innovators, investors, industries, project developers, and stakeholders in identifying regions with high potential for renewable energy generation and agricultural productivity. However modern climate variability must be accounted for in detailed implementation strategies, the findings serve as a foundation for exploring sustainable development opportunities in the SSAR region, particularly in resource planning, conservation efforts, and adaptation measures for climate-resilient agriculture. While a strong relationship between the WEF nexus and the axis chosen in which the prominent eight Shiva temples lie was found, it is challenging to directly compare the results with those of the ancient era due to the unavailability of enough data. However, the additional 43 Shiva temples constructed subsequently in the three selected belts confirm our findings of higher water–energy–food productivity potential.