Understanding the influence of teleconnections on Indian summer monsoon rainfall and agricultural productivity: a case study of Haryana, India (1980–2023)

India, home to around 1.4 billion people, is centrally located in South Asia, where the Indian summer monsoon (ISM), commonly known as the southwest monsoon, plays a pivotal role in the country's climate and socio-economic dynamics. The ISM involves a complex interaction between land, atmosphere, and ocean. The rainfall brought by the southwest monsoon is crucial, contributing 80% of the annual precipitation and serving as the primary water source for India's agriculture-dependent communities (Chakraborty et al. 2002; Ding & Sikka 2006). The yearly variation in the Indian Summer Monsoon Rainfall (ISMR) is approximately 10% of its average seasonal value. This fluctuation significantly influences agriculture, power production, and industrial output, profoundly affecting the nation's overall economic health (Gadgil & Kumar 2006). Due to its intricate nature, the variability of the ISM has become a critical and socially significant research topic. This is primarily because the ISMR impacts nearly 20% of the world's population. Given India's vast expanse, there is an inherent inconsistency in ISMR across its regions. Consequently, even during the best monsoon years, some parts of India experience scanty rainfall, while others might face floods even during the worst monsoon years. However, it is essential to recognize the regional disparities in ISMR variability. Owing to its high variability in both inter-annual and intra-seasonal timescales, accurate prediction of the ISMR remains a formidable challenge (Goswami et al. 2006). Hence, precise and timely seasonal forecasts of the ISMR are crucial for the nation's socio-economic development.

Every year, while the monsoon season delivers vast amounts of rainfall to most areas in India, certain regions still grapple with various types of droughts. This uneven distribution of the ISMR is largely influenced by the country's topography. Regions like the windward slopes of the Western Ghats and the northeastern mountainous areas often receive abundant rainfall due to their orography. However, the spatial distribution of ISMR can change dramatically on a daily basis. Favourable atmospheric conditions dictate that the ISMR often occurs in bursts throughout much of India. Heavy rainfall events are typically linked to atmospheric disturbances, such as depressions or cyclonic storms, which originate from warm waters in nearby oceans and typically last around 3–4 days as they traverse the country. The frequency of these heavy rainfall events can impact the overall intensity of the ISM, which is vital for areas like agriculture, water resource planning, disaster management, and the country's economic progression. Notably, years of bountiful food grain production align with those of abundant rainfall during the monsoon season, while years with scant rainfall often correlate with reduced food grain yields (Guhathakurta & Rajeevan 2008).

Despite extensive scientific research, the ISM remains a complex system, often leaving experts puzzled with myriad unanswered questions. Questions about its onset, intensification, progression, and recovery from break periods remain largely unanswered. Even as India has made notable strides in industrial growth since gaining independence, the national economy remains deeply anchored in agriculture and food production, essential for sustaining its vast and expanding population. However, recent advancements have been noted in simulating and predicting the ISM across various time scales using global climatic models. This progress has been accompanied by a deeper comprehension of the climate's response to significant monsoon teleconnections such as El Niño, La Niña, and the southern oscillation index.

As per the 2014 AR5 report from the IPCC, the El Niño-Southern Oscillation (ENSO), which is the primary tropospheric variability in the tropical Pacific, exhibits a significant correlation with the ISM. This correlation raises pressing concerns regarding the ability of diverse agroecological regions in the Indian subcontinent to ensure food security. The growing anomaly in seasonal rainfall during the ISM has profound impacts on agro-ecosystems throughout various parts of India. Jha et al. (2016) documented changes in how different agroecological regions in India responded to drought conditions. In their assessment of extreme rainfall risk across Indian agro-ecosystems, they observed a distinct climate pattern in the upper Gangetic plains: while cloud cover increased over time, actual rainfall decreased. The climatic and agro-ecosystem patterns vary significantly across regions, spanning from the northern Indo-Gangetic plains to central and eastern India, the southern peninsula, and the western regions of the country.

Numerous research findings consistently indicate that in years marked by diminished ISMR, the agricultural output of food grains during the kharif season tends to be lower, as evidenced by studies conducted by Webster et al. (1998), Selvaraju (2003), and Krishan Kumar et al. (2004). The summer of 2009 stood out as a period of severe drought, resulting in a substantial 14% reduction in rice production, according to the CCAP report in (2010). Additionally, instances of heavy and extreme rainfall events have been observed to give rise to floods, causing significant harm to crops, as discussed by Goswami et al. (2006). Annamalai et al. (2013) put forth the hypothesis that a notable rise in sea surface temperature (SST) over the Indo-Pacific warm pool, potentially linked to escalating concentrations of greenhouse gases, might explain the eastward-westward shift in monsoon patterns. This shift manifests in increased rainfall in the western tropical Pacific and diminished precipitation over South Asia. Roxy et al. (2015) made the noteworthy observation that the thermal contrast between land and sea has weakened, likely attributable to the heightened warming of the Indian Ocean. This diminished contrast has implications for and weakens the Hadley circulation, consequently diminishing rainfall in South Asia. Consequently, the region encounters more frequent occurrences of drought conditions, contributing to a significant reduction in grain yield.

Haryana state experiences a subtropical, semi-arid to sub-humid continental climate with a monsoon pattern, where about 75% of the annual rainfall occurs during the ISMR. The teleconnections like the ENSO and Indian Ocean Dipole (IOD) are believed to influence the year-to-year and decade-to-decade variations in the ISMR. The present study was conducted with the objective to investigate the impacts of the interaction of ENSO and IOD teleconnections on the ISMR of Haryana and subsequently on crop productivity. Additionally, the study seeks to understand the complex interactions between temperature, rainfall, and the productivity of kharif (monsoon) crops.

In this research, daily rainfall data spanning 44 years, from 1980 to 2023, were acquired from the India Meteorological Department at a resolution of 0.25 × 0.25°, as documented by Pai et al. (2014). The collected data underwent analysis at various time scales for comprehensive assessment. Additionally, the climatology of atmospheric variables was explored using the fifth-generation ERA5 reanalysis data from the European Centre for Medium-Range Weather Forecasts. This high-resolution dataset, featuring a spatial resolution of 0.25 × 0.25° and 37 vertical levels from the surface at 1,000 hPa to the top level at 1 hPa, has been available on a daily basis since 1979 to the present, as detailed by Hersbach et al. (2020). Throughout the paper, this dataset will be referenced as ERA5.

The study classified the positive, negative, and dormant phases of the ENSO, namely La Niña, El Niño, and neutral years, throughout the research period. This classification was based on the Oceanic Niño Index (ONI), which employs a 3-month running mean of the SST anomaly in the Niño 3.4 region (5°N–5°S and 170°W–120°W) of the equatorial Pacific Ocean. To be classified as a full-fledged El Niño (La Niña), the anomalies must exceed +0.5 °C (−0.5 °C) for at least five consecutive months. Similarly, positive, negative, and dormant phases of the IOD were identified on the basis of the Indian Ocean Dipole Mode Index (DMI), which is the difference between the SST anomalies of Western (10°S–10°N and 50°E–70°E) and Eastern (10°S–0°N and 90°–110°E) Equatorial Indian Ocean regions. Positive (negative) IOD events occur when the DMI exceeds 0.4 °C (−0.4 °C) for at least 3 months. The monthly time series data for the DMI and SST anomaly in the Niño 3.4 region were sourced from the National Oceanic and Atmospheric Administration (NOAA) ESRL Physical Sciences Laboratory site, available at https://psl.noaa.gov/gcos_wgsp/Timeseries/. In this study, the 44-year period was divided into seven groups based on the combination of positive, negative, and neutral phases of ENSO and IOD, as outlined in Table 1. Further analysis of rainfall patterns and atmospheric variables was conducted through the creation of composites that were based on these seven groups in order to explore the impact of ENSO and IOD on rainfall distribution at various time scales.

In this investigation, the assessment of vegetation dynamics relied on the analysis of the satellite-derived spectral index, specifically the normalized difference vegetationi (NDVI). The NDVI functions as a proxy for plant biomass and productivity, utilizing reflected light in the visible and near-infrared bands to gauge the vigour and health of vegetation within each pixel of a satellite image (Borowik et al. 2013; Santin-Janin et al. 2009; Garroutte et al. 2016; Szabó et al. 2020). To scrutinize the influence of teleconnections on vegetation dynamics during the monsoon season, we employed the half-monthly composites of the third-generation NDVI for the state of Haryana. These composites, with a spatial resolution of 0.083 × 0.083°, were sourced from the Global Inventory Monitoring and Modeling System (GIMMS) and spanned the period from 1981 to 2015. The NDVI values ranged from −1 to +1, and the percent deviation in the values of the NDVI was calculated for different ENSO–IOD group composites from the long-term mean (1981–2015) during different months.

The analysis of the correlation between ISMR, ENSO, and IOD was conducted at each grid for different months and the entire monsoon season. This investigation, presented in Figure 6, aimed to understand the intricate patterns and behaviour of rainfall across the state of Haryana. The correlation was assessed at four different levels of significance. The critical values of correlation for a two-tailed test at 1, 5, 10, and 20% levels of significance over the 44-year study period (n = 44) are 0.389, 0.301, 0.254, and 0.199, respectively. During the month of June, the correlation coefficient (CC) value between ISMR and ENSO ranges between −0.41 and 0.13, where the parts of the northeastern, central, and southeastern regions of Haryana showed statistically significant and negative correlations, while the CC value between ISMR and IOD ranged from −0.28 to 0.28, which was statistically non-significant for most of the region. July month witnessed statistically significant and negative correlations between ISMR and ENSO over most of the region with CC values ranging between −0.56 and −0.04, while the CC values between ISMR and IOD were comparatively weak and non-significant for most of the locations with values ranging from −0.26 to 0.21. August registered non-significant correlations of ISMR with both ENSO and IOD for most of the locations with CC values ranging from −0.31 to 0.12 and −0.14 to 0.30 for ENSO and IOD, respectively. Statistically significant and negative correlations between ISMR and ENSO were observed at most of the locations during the month of September with CC values ranging from −0.49 to −0.14, whereas it was non-significant for IOD at most of the region with CC values between −0.32 and 0.08. For the entire monsoon season, the CC value between ISMR and ENSO ranges between −0.69 and −0.15, where almost the whole study area witnessed statistically significant and negative correlations, whereas the CC value of ISMR with IOD was statistically non-significant for most of the region, ranging from −0.25 to 0.12. Overall, it can be seen that ENSO has a stronger influence on rainfall distribution over Haryana, while the association of ISMR with IOD was quite diffuse.

The atmospheric meteorological variables like vertically integrated moisture transport (VIMT; kg·m⁻¹·s⁻¹) and convective available potential energy (CAPE; J·kg⁻¹) are good indicators for assessing the state of the atmosphere and understanding the dynamics of the ISM. In this study, we utilized the ERA5 datasets to estimate mean seasonal composites of VIMT and CAPE for six different combinations of ENSO and IOD phases from 1980 to 2023. The aim was to evaluate the deviation of these variables from the mean seasonal neutral years composite observed during the study period of 44 years from 1980 to 2023, in order to determine whether the atmospheric meteorological variables had increased or decreased compared to neutral conditions. CAPE measures the amount of energy available for convection in the atmosphere. Higher CAPE values indicate a more unstable atmosphere, which is conducive to the formation of strong updrafts and thunderstorms. This instability is crucial for the development of convective clouds and subsequent rainfall. Higher CAPE values are generally associated with more intense and severe convective storms. During the monsoon season, increased CAPE can lead to more frequent and intense rainfall events, contributing to the overall monsoon rainfall. CAPE works in conjunction with moisture availability, often represented by variables like VIMT. Even with abundant moisture, the lack of sufficient CAPE can limit convective activity and rainfall. Therefore, both high CAPE and effective moisture transport are necessary for significant monsoon rainfall. CAPE values are influenced by teleconnections like ENSO and IOD. For instance, during La Niña years, higher CAPE values often lead to enhanced monsoon rainfall. Conversely, during El Niño years, lower CAPE values can reduce the ISMR. Positive IOD conditions can help mitigate the negative effects of El Niño on CAPE and rainfall. CAPE affects the spatial distribution of rainfall. Regions with higher CAPE are more likely to experience localized heavy rainfall events, which can impact water resources, agriculture, and flood risks.

In July (Figure 12(b)), a significant decrease in NDVI values compared to neutral years was observed throughout the state during El Niño conditions, particularly in the western to southwestern parts of Haryana when IOD was neutral. However, an increase in NDVI was observed when IOD was positive. During La Niña conditions, an increase in NDVI from normal was observed in different parts of the state, with a higher magnitude of increase when IOD was positive. The patterns observed in August (Figure 12(c)) and September (Figure 12(d)) were similar to those of July, but the magnitude of deviation in the NDVI values decreased as the monsoon season ended. Nonetheless, the NDVI increased during La Niña conditions when it occurred with either positive or negative IOD during both August and September. Overall, these results suggest that teleconnections have a significant impact on the NDVI over Haryana, and positive IOD conditions have a more pronounced effect on the NDVI than negative or neutral IOD conditions. These findings could be useful for agricultural planning and management in the region.

Figure 14(b) presents the time series of crop production for rice, bajra, maize, and jowar in the state during the kharif season. The mean production for these crops was 2,741.89, 708.96, 40.2, and 30.35 thousand tonnes, respectively. Rice and bajra have experienced an upward trend in production, increasing at the rate of 92.19 and 16.49 thousand tonnes/year, respectively. However, maize and jowar have experienced a decline in crop production, decreasing at the rate of 1.39 and 0.24 thousand tonnes/year, respectively.

Finally, Figure 14(c) shows the time series of crop productivity of rice, bajra, maize, and jowar in the state during the kharif season. The mean crop productivity for rice, bajra, maize, and jowar was 2,816.05, 1254.07, 2,002.46, and 344.85 kg/ha, respectively. The productivity of all crops has shown a positive trend over the years, with a mean increase of 20.55, 43.72, 52.56, and 8.35 kg/ha/year for rice, bajra, maize, and jowar, respectively. These results are important for understanding the agricultural productivity in Haryana and could be helpful in making informed decisions for sustainable crop management in the future.

Table 2 provides the results of a multiple linear regression analysis conducted to estimate the changes in the yield of four kharif crops viz., rice, bajra, maize, and jowar based on predictor variables of temperature and rainfall. The coefficients and associated p-values for each predictor variable are presented, along with the intercept and R² value. The coefficients indicate the impact of rainfall, maximum temperature, and minimum temperature on crop yield. The intercept represents the value of the crop yield when all predictor variables are zero. The R² value represents the proportion of variation in the crop yield that is explained by the predictor variables.

For rice productivity, the regression analysis showed a significant negative effect of rainfall (coefficient = −0.823) on the yield, indicating that a decrease in rainfall leads to a decrease in rice productivity. However, the coefficients for maximum temperature and minimum temperature were not significant. The R² value of 0.209 indicates that the predictor variables explain only 20.9% of the variability in rice productivity. For bajra productivity, the regression analysis showed a significant positive effect of rainfall (coefficient = 0.143) on the yield, indicating that an increase in rainfall leads to an increase in bajra productivity. The coefficient for maximum temperature was significant and negative (coefficient = −181.333), indicating that a higher maximum temperature leads to a decrease in bajra productivity. However, the coefficient for minimum temperature was not significant. The R² value of 0.430 indicates that the predictor variables explain 43.0% of the variability in bajra productivity. For maize productivity, the regression analysis showed a significant negative effect of both rainfall (coefficient = −0.835) and maximum temperature (coefficient = −288.398) on the yield, indicating that a decrease in rainfall and an increase in maximum temperature led to a decrease in maize productivity. However, the coefficient for minimum temperature was significant and positive (coefficient = 209.866), indicating that an increase in minimum temperature leads to an increase in maize productivity. The R² value of 0.162 indicates that the predictor variables explain only 16.2% of the variability in maize productivity. For jowar productivity, the regression analysis showed that none of the predictor variables (rainfall, maximum temperature, and minimum temperature) had a significant effect on the yield. The R² value of 0.232 indicates that the predictor variables explain only 23.2% of the variability in jowar productivity. Overall, the results indicate that the effects of temperature and rainfall on crop productivity are complex and can vary across different crops. While some crops are negatively affected by increased rainfall or temperature, others show a positive response to these variables. Additionally, the models explained only a small percentage of the variability in crop productivity, suggesting that other factors beyond temperature and rainfall may also play a significant role in determining crop yields.

Table 3 depicts the average productivity of rice, bajra, and maize crops observed during different combinations of ENSO and IOD phases to emphasize the variability in crop productivity under different phases. Rice productivity is the highest during neutral ENSO with positive IOD (3,052.0 kg/ha) and the lowest during La Niña with negative IOD (2,239.0 kg/ha). Bajra productivity peaks under neutral ENSO with positive IOD (1,524.2 kg/ha) and is the lowest during El Niño with positive IOD (880.7 kg/ha). Maize productivity follows a similar trend, being the highest during neutral ENSO with positive IOD (2,276.3 kg/ha) and the lowest during El Niño with positive IOD (1,560.0 kg/ ha). The higher crop productivity observed during neutral ENSO with positive IOD for rice, bajra, and maize can be due to the optimal climatic conditions associated with these phases. Neutral ENSO conditions are typically characterized by relatively stable and moderate SSTs in the Pacific Ocean, leading to more predictable and consistent weather patterns. When combined with a positive IOD, which is associated with warmer SSTs in the western Indian Ocean, these climate variability conditions generally enhance the monsoon rainfall over the Indian subcontinent and increase the moisture content over the Indian Ocean, resulting in more abundant and well-distributed rainfall across the agricultural regions like Haryana. This climate variability underscores the significant impact of climatic phases on agricultural productivity, which is critical for understanding and mitigating risks in agricultural planning.

Due to the complexity of conducting precise analyses at finer scales using climate change models, accurately depicting regional-scale phenomena remains challenging. This limitation underscores the critical need for a scientific understanding of long-term rainfall patterns and vegetation dynamics to assess the impacts of teleconnections such as ENSO and IOD on ISMR and vegetation. Long-term observations of seasonal rainfall and temperature are essential for comprehending climate variability and its implications for local ecology, agriculture, and human well-being.

In Haryana, where the mean seasonal rainfall averages 478.63 mm, the distribution varies significantly across the state, with northeastern districts receiving higher rainfall compared to western and southwestern regions. This spatial pattern is consistent with previous studies (Malik & Singh 2019; Abhilash et al. 2021), attributed to orographic effects where northeastern districts benefit from the foothills of the Himalayas, which intercept southwest monsoon winds (Singh et al. 1995; Das & Meher 2019). The influence of teleconnections like ENSO and IOD on ISMR is well-documented (Krishnamurthy & Goswami 2000; Ashok & Saji 2007; Cherchi & Navarra 2013; Behera & Ratnam 2018; Chauhan et al. 2022a, b). During El Niño years, warm waters in the eastern Pacific weaken the Walker circulation and easterly trade winds, reducing moisture transport from the Indian Ocean and delaying the onset of the Indian monsoon. Conversely, La Niña enhances the Walker circulation, increasing moisture transport and resulting in early monsoon onset. The IOD, characterized by SST differences between its western and eastern parts, also significantly affects rainfall. Positive IOD enhances monsoon winds and increases Indian rainfall, while negative IOD weakens monsoon winds and reduces rainfall (Behera & Ratnam 2018; Hrudya et al. 2021).

Our study confirms significant negative correlations between ENSO and ISMR during the monsoon season in Haryana, whereas correlations with IOD are mostly non-significant, indicating a stronger influence of ENSO on regional rainfall patterns. This aligns with findings by Ashok & Saji (2007), highlighting the broader impact of ENSO as compared to IOD on regional rainfall variability. The combined effects of ENSO and IOD result in substantial reductions in rainfall during El Niño years, particularly when IOD is neutral, whereas positive IOD mitigates this reduction to some extent. Atmospheric dynamics, including VIMT and CAPE, further elucidate these patterns, showing enhanced moisture transport and convective activity during La Niña and positive IOD conditions.

The relationship between ENSO–IOD interactions and vegetation dynamics in Haryana was assessed using the NDVI, a robust indicator of vegetation health (Lenney et al. 1996; Fan et al. 2009; Kamble et al. 2010). The NDVI patterns closely followed ISMR variations during different ENSO and IOD phases, reflecting higher vegetation growth with increased rainfall and vice versa (Dubey et al. 2012; Kumar et al. 2013). Crop statistics reveal a shifting cultivation pattern in Haryana, with an increase in rice cultivation at the expense of bajra, maize, and jowar (Gautam & Sangwan 2021). This shift is accompanied by overall increases in crop productivity, attributed to advancements in agricultural technologies such as improved varieties and management practices (Kamble et al. 2010; Sehgal et al. 2011; Dhakar et al. 2013). Analysis of rainfall and temperature trends from 1980 to 2023 shows varying impacts on predominant crops like rice, bajra, maize, and jowar. While rice area and productivity show positive trends, other crops exhibit negative trends, indicating differential responses to climate variability. High temperatures during the monsoon season in Haryana increase evaporation rates, potentially impacting water availability and agricultural productivity (Zhao et al. 2017; Meshram et al. 2020). Temperature variations also affect crop yields, with productivity declining as rainfall decreases and temperatures rise. Overall, crop productivity was found to decrease with decreasing rainfall and increasing maximum temperature. The temporal variability of mean seasonal rainfall, minimum temperature, and maximum temperature observed in this study provides valuable insights into the climate dynamics of the region. These findings can aid in developing sustainable strategies for climate change adaptation and mitigation in Haryana's agricultural sector.

This comprehensive study of climatic and agricultural data in Haryana from 1980 to 2023 provides valuable insights into the spatio-temporal distribution of ISMR, atmospheric variables, and their impact on crop production. The analysis reveals distinct rainfall patterns across Haryana, with the northeast region consistently receiving higher rainfall compared to the western and southwestern parts. EOF and PC analyses illustrate spatial and temporal variability in ISMR, highlighting higher covariance in the northeastern region during EOF1 and a reversal pattern in EOF2. Years characterized by excess or deficient precipitation are identified through normalized PC scores, with notable examples such as 1988, 1990, 1994, 1995, 1996, 1998, 2008, 2010, and 2011 showing excess rainfall, and 1982, 1987, 2002, and 2014 exhibiting deficient rainfall influenced by PC1. Similarly, PC2 influences include excess rainfall in 1993, 2008, and 2021 and deficient rainfall in 1986, 1989, 1999, and 2013. The study demonstrates the significant influence of ENSO and IOD on temperature, wind patterns, and rainfall. ENSO shows a stronger impact on ISMR than IOD, with correlation coefficients ranging from −0.69 to −0.15 for ENSO and −0.25 to 0.12 for IOD. During El Niño years with neutral IOD, rainfall reduces by up to 50%, while reductions are less pronounced during El Niño years coinciding with positive IOD. VIMT and CAPE are influenced by ENSO and IOD, with La Niña conditions contributing to higher values. The NDVI patterns closely match ISMR variations during different ENSO and IOD phases, indicating the impact of rainfall on vegetation growth. The study also highlights shifts in crop cultivation patterns, with increased rice cultivation and varying trends for crops like bajra, maize, and jowar. This research contributes significantly to understanding the complex interplay between climatic factors, teleconnections, and agricultural dynamics in Haryana. The findings provide a robust foundation for informed decision-making in agriculture, climate resilience, and land management in the region. Future research should focus on extending climate projections using advanced models to predict ENSO and IOD impacts on ISMR and agricultural productivity under various climate change scenarios. Detailed crop-specific studies and integrated hydrological, climate, and crop models will enhance our understanding of water availability and irrigation needs. These efforts will guide the development of adaptive strategies, including drought-resistant crop varieties and climate-smart agricultural practices, crucial for enhancing resilience and ensuring food security in the face of changing climatic conditions.

The author(s) would like to thank the India Meteorological Department (IMD), Pune, for providing the daily rainfall time series data for this study.

A. S. C., A. R., S. S., and A. D. were involved in conceptualization of methodology, investigation, data analysis, preparation of figures, and writing original draft. R. K. S M. performed climatological data analysis for ISMR and teleconnections.

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

The authors declare there is no conflict.