Understanding climate variability and trends is crucial for managing a host of sectors. Everything from water availability to agricultural productivity is affected by variability and trends in temperature, rainfall, evapotranspiration, and solar radiation. Nevertheless, their dynamics have seldom been explored together, especially in India. To address this gap, the present study investigates the variability, trend, and magnitude of those parameters individually and concurrently using fractal dimension and non-parametric statistics over the Indian state of West Bengal from 1951 to 2020. The results show a south–north gradient in overall climate variability. The Gangetic West Bengal (GWB) is experiencing higher variability, along with a rising minimum temperature (≥0.008 °C year−1) and declining rainfall (≥− 1 mm year−1). Though the Sub-Himalayan West Bengal as a whole shows less variability, its foothills reveal modest variation coupled with increasing maximum temperature (≥0.005 °C year−1), reference evapotranspiration (≥0.4 mm year−1), and decreasing rainfall in the post-monsoon and winter seasons. Based on the results, we identified the western GWB, the Sundarbans, and the sub-Himalayan foothills as the most vulnerable areas and recommended proactive crop and water management strategies. Finally, we underline the need to analyze climate dynamics holistically to manage climate-sensitive sectors efficiently and sustainably.
Analyzed the variability, trend, and magnitudes of five critical climatic parameters.
The variability of climate dynamics shows a south–north gradient.
The western, coastal, and foothill regions exhibit a complex and uncertain climate.
The results are discussed in terms of their possible impacts on agricultural and water resources.
In general, scientists agree that anthropogenic activities substantially raise the concentration of greenhouse gases (GHGs) and trigger large-scale variations in atmospheric processes (Ramanathan et al. 2007). This enhanced rise in GHGs alters the Earth's overall climatic condition through changes in different climatic parameters. Thus, assessing the nature, variability, and trend of critical climatic parameters holds significant importance not only for understanding the environmental state but also for managing climate-induced sectors such as water, agriculture, biodiversity, and ecosystem services (Rao et al. 2015; Srivastava et al. 2016; Xu et al. 2017). Such information also serves as a basis for formulating sector-specific climate-resilient strategies.
In order to combat climatic adversities, researchers around the world have urged a holistic evaluation of climate change (Kim & Ivanov 2015; Boyd 2017), which necessitates a more in-depth and holistic approach. Even after the growing call, emphasis has mostly been given to maximum temperature (Tmax), minimum temperature (Tmin), and rainfall (Prc). Consequently, other parameters, like solar radiation (Srd) and reference evapotranspiration (ET0), which are crucial and have significant contributions in agriculture to human health, remain less explored. Furthermore, most research focused on monotonic trends (for instance, Dash et al. 2007; Bandyopadhyay et al. 2009; Jhajharia et al. 2012; Sonali & Kumar 2013; Rao et al. 2014; Rao et al. 2015; Mukhopadhyay et al. 2016; Sharma et al. 2016; Basha et al. 2017; Datta & Das 2019a, 2019b; Das et al. 2022; Mann & Gupta 2022) while trivializing overall or holistic climate variability. Since the climate is chaotic with high variabilities (Rind 1999; Xu et al. 2016), such ignorance may have a detrimental impact on many sectors, especially in agriculture and water resources.
Besides other parameters, the primary driving factors of field-crop production (such as water availability, soil moisture content, and the duration and intensity of sunlight) are explicitly controlled by Tmax, Tmin, Prc, ET0, and Srd variability (Cline 2008; Rao et al. 2014; Jabal et al. 2022). Undeniably, climate change has substantially affected almost all sectors that drive human life and livelihoods. However, the impact on agriculture is widespread and alarming (Das & Goswami 2021), where studies have shown that it would reduce crop yield to a great extent, even after accounting for the benefits of CO2 fertilization on crop growth (Parry et al. 2004; Rao et al. 2015). A rapid and unprecedented alteration in climatic patterns may have cascading effects, beginning with a dip in agricultural production and escalating through food security, employment, and, eventually, the economy of the nation. Several climate-induced hazardous events have occurred in India during the last two decades, wreaking havoc on the agricultural economy. The estimated revenue loss from such impact ranges between 15 and 18% (Government of India 2018a). Therefore, it is essential to analyze climate dynamics holistically in a country like India, where agriculture contributes to 16.38% of the country's gross domestic product (MOSPI 2021), employs over 60% of the population (Srivastava et al. 2016), and feeds more than 17% of the world's population (Singh et al. 2014).
While a large-scale assessment of climate dynamics is essential to comprehend the larger picture, regional-level studies may be more significant for policy implementation. In light of this, we choose the Indian state of West Bengal as a study site. West Bengal is India's only state that extends from the mighty Himalayas in the north to the Bay of Bengal in the south, with diverse physiographic, climatic, and agricultural conditions. Such diversities make this state suitable for analyzing spatio-temporal variability and trends in climatic parameters and could adequately reflect the national situation. Also, the state's economy relies on the climate-sensitive agricultural sector, and issues related to inconsistent rainfall, rising temperature, floods, drought, cyclones, and pests and diseases (Datta & Das 2019a) have become a common phenomenon. For instance, in 2008–09, the yield of Boro1 rice dropped by 14% due to the scorching summer, resulting in an 11.5% fall in overall Boro rice production (Government of West Bengal 2010).
Given this context, the present study intends to explore (i) the spatial variability of climate dynamics at seasonal and annual scales by using fractal dimension theory and (ii) the spatio-temporal trend and magnitudes in key meteorological parameters (Tmax, Tmin, Prc, ET0, and Srd) throughout West Bengal by employing multiple non-parametric statistics. Moreover, this is a first-of-its-kind variability and trend assessment in the Indian context by integrating five crucial climatic variables: Tmax, Tmin, Prc, ET0, and Srd. Such a combined study, comprising both variability and trend, is expected to provide a more holistic understanding of the changing climate dynamics and their potential implications for agriculture and water resources in West Bengal. Moreover, this approach could unravel more detailed features of climate change than an independent assessment of a single or pair of climatic factors. The resultant variability, trend, and magnitude of those critical climatic parameters would provide a basis for policymakers to craft suitable adaptation strategies for effectively managing agricultural and water resources to attain sustainable food security.
MATERIALS AND METHODS
The study area
The India Meteorological Department (IMD) has bifurcated the entire state into two distinct sub-divisions: Gangetic West Bengal (GWB) and Sub-Himalayan West Bengal (SHWB). Generally, the latter receives more rainfall than the former due to the northward shift of the monsoon trough (IMD 2008) and its vicinity to the Himalayan range (Figure 1(b)). During the pre-monsoon season (March–May), sporadic rainfall occurs throughout the state due to Nor'wester thunderstorms (Sadhukhan et al. 2000). In contrast, the post-monsoon season (October–November) frequently witnesses tropical cyclones, particularly along the coast (Datta & Das 2019a). The winter, on the other hand, is rather dry and cold. During summer, the western parts of GWB frequently encounter heatwaves, with Tmax exceeding 45 °C (IMD 2008; Datta & Das 2019a). The northernmost districts rarely experience extreme temperatures, while the hilly areas of the Darjeeling district get occasional winter snowfalls. The Bauxa region (in the foothills of SHWB) receives the highest rainfall, while Purulia (in western GWB) the least in West Bengal.
Data type and source
The daily gridded time series of rainfall and temperature (both the Tmax and Tmin) data covering a period of 70 years (from 1951 to 2020) were acquired from the IMD's official web portal (https://www.imdpune.gov.in/Clim_Pred_LRF_New/Grided_Data_Download.html). The high-resolution gridded rainfall (0.25° × 0.25°) data were developed using observations from over 6,995 gauge stations spread across India (Pai et al. 2014), whereas gridded temperature (1° × 1°) data were created using observations from 395 quality-controlled stations (Srivastava et al. 2009). On the other hand, the high-resolution (0.25° × 0.25°) wind speed (at 10 m height) and dew point temperature data covering the same period (1951–2020) were obtained from the ERA5 atmospheric reanalysis products (Hersbach et al. 2020) developed by the European Centre for Medium-Range Weather Forecasts (ECMWF). The ERA5 products were chosen in this study because of their improved horizontal and vertical resolutions, bias correction, and analysis based on a 10-member ensemble 4D-Var (Mahto & Mishra 2019).
Estimation of reference evapotranspiration (ET0)
Formulation of seasonal and annual data
Only the grids falling inside the state of West Bengal were considered to analyze the variability, trend, and magnitude. The gridded daily data were then transformed into annual and seasonal scales using the conventional summation (for Prc and ET0) and average (for Tmax, Tmin, and Srd) techniques, with four dominant seasons as defined by Rathore et al. (2013): pre-monsoon (March–May), monsoon (June–September), post-monsoon (October–November), and winter (December–February). The study includes annual and seasonal series as the variability and trends, manifesting not only on an annual scale but also in its distinct seasons. Furthermore, a statistically significant (upward) trend on an annual scale may reflect a non-significant (downward) pattern on seasonal scales.
Computation of fractal dimensions
The D value varies from 1 to 2. When D is 1.5, there is no correlation between the magnitude changes corresponding to the two successive timeframes, indicating complete randomness. As a result, neutral or no pattern can be discerned from that particular data, which leads to a highly uncertain process (Rangarajan & Sant 2004). However, a positive correlation can be observed if the D value is between 1 and 1.5, suggesting that a rise in the magnitude of the preceding data is likely to persist in the following data. As the fractal dimension decreases to 1, the process becomes more and more predictable as it exhibits a persistent character (Cui et al. 2022). A negative correlation, on the other hand, may be inferred when D increases from 1.5 to 2, indicating that the variable under consideration is most likely to decline in the event of an increase in the amplitude of the previous data. Therefore, the time series displays anti-persistency when the fractal dimension value is close to 2 (Harrouni & Guessoum 2009). The higher the value of D of a climatic parameter the greater its variability, signifying that the climatic variables under investigation change with more frequency and uncertainty and vice versa (Xu et al. 2017). In this study, following Xu et al. (2017), we defined climate dynamics as a multiplication of the D values of those five meteorological parameters (since the considered variables interact conjointly at the same time) and computed it at seasonal and annual scales.
Identification of trends and their magnitudes
The non-parametric Mann–Kendall (MK) test (Mann 1945; Kendall 1975) was applied to examine trends in the long-term climatic parameters. The MK test is appropriate as the data need not pursue a specific distribution. However, significant positive autocorrelation in the data can augment the likelihood of trends while there is no trend, and vice versa (Hamed & Rao 1998). Therefore, the variance correction approach, namely the Modified Mann–Kendall (MMK) test suggested by Hamed & Rao (1998), was applied to eliminate the effect of autocorrelation from the time series. The magnitude of the obtained trend was computed by non-parametric Sen's slope estimator (SSE) (Sen 1968). All the statistical significance was measured at a 95% confidence interval, i.e., α = 0.05. The mathematical formulas of MK/MMK and Sen's slope were presented in Supplementary material, Methods S1–S3.
Spatio-temporal variability in climate dynamics
Considering the Tmax (Supplementary Figure S1), higher variabilities with a fractal dimension ranging from 1.472 to 1.548 were primarily observed in the northern SHWB and southern GWB during pre-monsoon and winter and western and southwestern GWB during monsoon and post-monsoon seasons. The southern and southeastern GWB witnessed a neutral pattern (as the D value reaches 1.500) on the annual scale. Regarding Tmin (Supplementary Figure S2), most of the central and western parts of the GWB showed a high level of variability (1.288–1.519), whereas the entire SHWB showed lesser variability (1.210–1.306) irrespective of seasonal or annual timescales. It is noticed that areas with higher variability gradually diminished with the passage of seasons (pre-monsoon to winter).
We observed quite a homogeneous spatial variability of Prc, i.e., no substantial differences throughout the seasons, including annual scales (Supplementary Figure S3). The variabilities gradually decreased from 1.516 in the western and southwestern GWB (Purulia, East Medinipur, and West Medinipur districts) to 1.210 in the northernmost SHWB (Darjeeling Himalayas).
As for ET0 (Supplementary Figure S4), the western and southwestern parts (Purulia, East Medinipur, and West Medinipur districts) and some patches over coastal GWB (South 24 Parganas district) exhibited consistently higher variability (1.398–1.486). In contrast, the central and southern SHWB (especially in Malda, Dakshin Dinajpur, and Uttar Dinajpur districts) revealed lesser variability (1.238–1.298) throughout the seasonal and annual timescales. Except for the monsoon, several grids in northern and northeastern SHWB (Darjeeling, Jalpaiguri, and Cooch Behar districts) and eastern GWB (Nadia, North 24 Parganas, Howrah, Kolkata, Eastern Burdwan, and some portion of Murshidabad) displayed high and low variability across seasonal and annual scales, respectively.
Concerning the Srd dynamics (Supplementary Figure S5), the whole SHWB showed fractal dimensions lower than 1.250, except for a few grids in the northeastern section, which were only present during the pre-monsoon and winter seasons. However, the western and southwestern areas of the GWB (e.g., the western part of Purulia, East Medinipur, and West Medinipur districts) revealed higher fractal dimensions ranging from 1.384 to 1.434, indicating the strongest variability (Supplementary Figure S5c).
Upon examining the variability of the studied parameters individually (Supplementary Figures S1–S5), it is found that the temperature (both Tmax and Tmin) had the highest variability, followed by ET0, while Prc and Srd had the lowest variability, irrespective of seasonal and annual scales. Barring a few grids from Tmax and Tmin, the fractal dimensions of the considered climatic parameters revealed a mostly persistent nature, indicating that the future trend may likely follow the present pattern. The anti-persistent nature is mainly witnessed in pre-monsoon, winter, and post-monsoon seasons of the Tmax, Tmin, and Prc parameters, translating that a decrease (increase) in the degree of variability is more likely to follow an increasing (decreasing) pattern in the near future. Quite a few grids, on the other hand, particularly from Tmax (during pre-monsoon) and Tmin (during post-monsoon), showed a fractal value of exactly 1.5, i.e., complete randomness in the data, leading to a highly uncertain process and thus unpredictable. The fractal dimension of the studied parameters ranges between 1.210 and 1.548, confirming the self-affine characteristics, i.e., repeating patterns across scales. It was evident in the spatio-temporal variability of climate dynamics (Figure 3), including the individual variability of the studied climatic variables (Supplementary Figures S1–S5). For instance, the variability of climate dynamics (Figure 3) is almost identical during the pre-monsoon and winter seasons. Likewise, the variability of monsoon dynamics exhibited repetitive behavior on the annual scale.
Climate trends and their magnitudes
Overall, the study found a higher variability of climate dynamics in GWB (particularly in the western and southwestern parts) than in the SHWB region. However, several grids in the north and northeastern SHWB showed higher climate dynamics during pre-monsoon and winter seasons. It is due to the pronounced variability of Tmax (more than 1.540), Srd (more than 1.300), and ET0 (more than 1.390). It was further accentuated by the clear-sky conditions during this season because more solar radiation reached the Earth's surface, increasing the energy availability and enhancing the rate of ET0, eventually exacerbating the variability of climate dynamics. Such a finding was also observed in northern China (Xu et al. 2017).
While comparing the results of the present study with those from earlier studies, we found substantial (little) differences considering the variability (trend) of studied climatic parameters, respectively. Except for ET0, the observed variability and trend of the studied parameters are comparable to previous studies (Mandal et al. 2013; Sharma et al. 2016; Datta & Das 2019a, 2022) for the SHWB region (barring a few areas in Cooch Behar and Jalpaiguri districts) while contrasting for the GWB region (particularly in the western, southwestern, and central parts). Broadly, the ET0 trend corresponds to the findings of Bandyopadhyay et al. (2009) for the GWB region (except for monsoon and post-monsoon) and contradicts the outcomes of Jhajharia et al. (2012) for the SHWB region (except for pre-monsoon and winter). Such disparities might be attributed to the differential analysis period and timescales to construct seasonal data (like January–February vs. December–January–February for winter), including the preferred methodologies. Notably, regions with higher variability in previous studies usually exhibited a lower variability in the present study and vice versa. For instance, Sadhukhan et al. (2000) reported a higher rainfall variability in the central and northwestern GWB region during the pre-monsoon season, whereas the present study observed a lower variability. Likewise, Datta & Das (2019b) found a rising Tmax and Tmin variability in the Darjeeling and Malda districts within the SHWB region, but the current study revealed declining variability (barring a few grids in Darjeeling during pre-monsoon and winter seasons) as the fractal dimension of those areas ranges between 1.276 and 1.334 compared to the higher variability areas of 1.472–1.545 (Supplementary Figures S1 and S2). It is likely to use fractal dimension, which provides more detailed characteristics across space and time than the conventional coefficient of variation technique often used in earlier studies. Moreover, regions with higher variability of climate dynamics appeared to have unclear long-term trends, making them highly unpredictable.
Since the GWB is one of India's crucial rice-growing belts, higher variability of climate dynamics across seasonal and annual scales could have far-reaching consequences on agricultural production. Furthermore, temperature rises in this region are expected to impact crop growth by increasing respiration and basal metabolism and modifying the plant response system to various biotic stresses. On the other hand, declining monsoonal rainfall coupled with rising ET0 may result in reduced water and soil moisture availability, followed by drought. Together, these changes will negatively impact agricultural production, especially rice productivity, by lowering the biomass and yield. Since more than 6% of India's food production exclusively comes from West Bengal (Government of India 2018b), such a situation is alarming for the livelihood of the booming population and the country's ability to feed itself. Apart from this, higher variability of climate dynamics means more complexity and uncertainty, which may inhibit farmers' decision-making regarding principal food grain (especially rice) production, as they will not be able to anticipate the changes in the onset and cessation of monsoon, occurrence of consecutive dry and wet spells, and so on. Inevitably, the farmers of this region rely on irrigation facilities to meet the crop water demand even in monsoon months, leading to 31 m of groundwater level depletion, as found by Pradhan et al. (2022). The uncertain climate and plummeting groundwater levels could force farmers to switch to less water-intensive non-food grain production, such as oilseeds, sugarcane, and mesta, despite those crops requiring longer growing days. Such evidence has already been observed in several districts of West Bengal. For instance, the non-food crop acreage has increased (from 2004–05 to 2014–15) in Bankura, Burdwan, Cooch Behar, and Purulia districts, albeit marginally (Government of West Bengal 2016).
The present study also reveals a moderate variability (4.209–4.619) of climate dynamics in the SHWB region, notably in a few isolated areas of Darjeeling, Jalpaiguri, and Cooch Behar districts during the lean period (October–May). A further variation in climate dynamics, including rising temperature, evapotranspiration, and diminishing rainfall, may increase the likelihood of drought, impacting the world-renowned Darjeeling tea production. Surprisingly, this region, which used to be abundant in groundwater, has also shown diminution in recent decades due to over-abstraction followed by decreasing rainfall (Rudra 2017). Therefore, there is a need for efficient crop and water management not only to offset any further deterioration of groundwater but also to reduce crops' vulnerability and risk from climate change and/or variability. In this regard, agroforestry, short-duration cultivars, crop rotation and diversification, and mixed farming could be viable options concerning crop management. For groundwater depletion, using deep farm ponds might lessen the risk while restoring soil moisture and providing irrigation to fulfill crop water needs during dry seasons (Das et al. 2022). Additionally, encouraging maize over water-intensive Boro rice during the Rabi season (October–May) may revive groundwater levels as it only requires one-fifth of the total water consumed by rice and delivers higher returns (Kapuria & Banerjee 2022). On the other hand, the foothills of the SHWB have significant potential for mushroom farming (Datta & Das 2021), which may also be adopted to minimize the additive economic damages.
CONCLUSION AND THE WAY FORWARD
The present study examined the spatio-temporal variability of climate dynamics by integrating five climatic parameters (Tmax, Tmin, Prc, ET0, and Srd) critical for field-crop production through the lens of fractal dimensions over the Indian state of West Bengal from 1951 to 2020. Furthermore, this study assessed the trend and magnitudes of those climatic parameters to comprehend the changing climate scenarios holistically. The results showed a higher variability of climate dynamics in the western, southern, and southwestern GWB, whereas central, eastern, and southeastern sections showed a modest degree of variability with an increasing trend in temperature and a decreasing trend in rainfall. Contrarily, the SHWB witnessed a lower fractal dimension in all parameters, resulting in low variability. Nevertheless, the immediate foothill areas of SHWB (especially Darjeeling, Jalpaiguri, and Cooch Behar districts) exhibited a modest degree of variation in climate dynamics along with a rising trend in temperatures and reference evapotranspiration and a decreasing trend in rainfall during the post-monsoon and winter seasons, indicating the possibility of drought in the near future. Furthermore, the study observed the presence of persistence, anti-persistence, neutral, and self-similar patterns among the analyzed variables. However, more research is required to untangle such a complex behavior of climatic factors, which would contribute to a more comprehensive understanding of the changing climate dynamics. Considering variability, trend, and magnitudes of the analyzed parameters, the study identified (i) the western and southwestern GWB; (ii) the coastal belts, especially the deltaic areas of the Sundarbans; and (iii) the immediate foothills of the SHWB region as the climatically most vulnerable areas in West Bengal and in need of proactive crop and water management policy measures.
The changing climate has significant repercussions on agricultural productivity, followed by food security, livelihoods, and socio-economic development of the growing human population. A marginal increase in the non-food crop area signifies a short-term adjustment or business-as-usual strategy to offset the economic loss due to the yield reduction in conventional food grain production under highly complex and uncertain climatic conditions. A low to moderate crop diversification rate in the GWB region, specifically in Purulia, Bankura, Birbhum, and Burdwan districts (Birthal & Hazrana 2019), corroborated our standpoint. Indeed, such an adjustment in crop cultivation solely driven by profit maximizations under climate change may have helped the farmers achieve some economic stability; however, it might not be a sustainable path considering the global issue of food security. Sustainable food security requires systemic and transformational change (Campbell et al. 2018; Vermeulen et al. 2018) in a feasible manner, i.e., strategies with values and visions and the way they are implemented in practice (Bentz et al. 2022), which is currently lacking in the state (for instance, see Dey et al. 2016). Therefore, future studies should include crucial socio-politico-economic attributes to examine how climate dynamics, including climate-induced extremes, affect the agricultural systems and render them susceptible, which may aid in formulating appropriate adaptation strategies.
The authors express their gratitude to the Editor-in-Chief, the Associate Editor, and the anonymous reviewers for their critical review and constructive suggestions. The first author received support from the University Grants Commission (UGC), New Delhi, India, in the form of the Junior (Senior) Research Fellowship Award [(Award No. 3291/(SC)(NET-DEC.2015)] to pursue the research. Furthermore, the first author would like to thank Lilu Bhoi for his assistance in coding and Pritha Datta for her encouragement and insightful input on the study design.
A wet variety of paddy sown in winter and harvested in summer.
C: Temperate; w: Dry winter; b: Warm summer.
C: Temperate; w: Dry winter; a: Hot summer.
A: Tropical; w: Savannah.
S.D. worked on conceptualization, formal analysis, methodology, resources, software, visualization, writing – original draft, writing – review and editing. K.G. focused on conceptualization, supervision, validation, writing – review and editing.
The authors declare that the article is not funded by any scientific institution.
DATA AVAILABILITY STATEMENT
All relevant data are included in the paper or its Supplementary Information.
CONFLICT OF INTEREST
The authors declare there is no conflict.