1 Introduction

Globally, climate change has recently exacerbated drought risks and associated crop failure (Zhang et al. 2017; Lesk et al. 2016). In India, this risk is more significant due to various factors like the variability of monsoon rains (Sharma and Mujumdar 2017; Thomas and Prasannakumar 2016), dominantly rainfed agriculture (Vikramarjun et al. 2019; Ward and Makhija 2018), and livelihood of 75% people depend on agriculture (Rao et al. 2018) and the growing demand for food population (Kumar and Joshi 2016). In India, the occurrence of meteorological drought is reported to be linked to the shortage of monsoon rainfall, while rising temperatures can potentially increase the drought (Pandey and Srivastava 2019; Bhunia et al. 2020). Due to the erratic and uneven distribution of monsoon rainfall with increasing temperature, drought conditions are prevalent even in the sub-humid regions of India (Pathak and Dodamani 2019). Various studies (Zhai et al. 2020; Sam et al. 2020; Nath et al. 2017; Pai et al. 2017) on drought analysis in India indicated an increasing trend in drought severity and frequency over the agriculturally important sub-humid eastern part of the country in recent decades. Frequent droughts can threaten food security in this region (Sharma and Goyal 2020). A slight departure of average monthly rainfall from normal in the red and laterite zone (RLZ) significantly affects rainfed agricultural production considerably (Brahmachari et al. 2018; Mandal et al. 2018). Subash and Ram Mohan (2011) also found that the variability of Kharif rice yield has influenced the monthly monsoon rainfall. Hence, an intensive study of the drought of various categories in sub-humid RLZ is essential.

Even though the RLZ understudy receives an annual average rainfall of 1200 mm, it became more erratic and uneven during the 1990–2000 period compared to the early decades (Mukherjee and Huda 2018). An increasing temperature trend and uncertain rainfall projection over sub-humid RLZ emphasized high sensitivity to climate change (Shukla et al. 2017). In RLZ, monsoon rain has decreased (Ghosh 2018), while an increase in minimum temperature (0.5–1 °C) was recorded in recent decades (Mukherjee and Huda 2018). As a result, recurrent drought has affected RLZ (Bhunia et al. 2020; Patra 2020). Due to rainfed agriculture, low water holding capacity, excessive drainage, surface runoff, and high soil erosion make such red lateritic zones more vulnerable to drought (Roy et al. 2020; Asutosh 2019; Mukherjee and Banerjee 2009). The changed evapotranspiration pattern due to increased temperature and absence of rainfall is expected to increase incidences of drought during recent periods in RLZ (Mandal and Chakrabarty 2013). Recent studies (Bhunia et al. 2020; Banik et al. 2020) indicated that the increased drought frequency severely impacts crop production in the RLZ. In the context of climate change, frequent droughts are considered limiting factors for the growth of agriculture and socio-economic development in this area (Goswami et al. 2019). In recent times, with 43% draught frequency (Nath et al. 2017) over the sub-humid region in India, agricultural production is progressively getting affected (Kumar et al. 2019). A detailed study is thus necessary to identify changing spatiotemporal characteristics of drought occurrence at the regional level in the context of climate variability for better mitigation planning and preparedness.

Various indices for meteorological drought analysis have been formulated based on various parameters (Guhathakurta et al. 2017). A detailed description of the drought indices can be found in Heim (2002), Mishra and Singh (2011), and Sivakumar (2011). The Standardized Precipitation Index (SPI, Mckee et al. 1993) is one of the most widely used among various drought indices. SPI can interpret rainfall data in different time scales (1, 2, 3, 6, 9, 12, 24, … months) to assess drought duration, intensity, severity, magnitude, and frequency (Bhunia et al. 2020). Compared to other indices, the SPI has simple calculations and is decisively performed space independently (Bhunia et al. 2020). As a result, SPI has already been widely used to detect and characterize drought conditions in many countries and regions, such as Turkey (Dabanlı et al. 2017), Iran (Awchi and Kalyana 2017), China (Li et al. 2020; Zhang et al. 2019, 2017; Xia et al. 2018), Bangladesh (Rahman and Lateh 2016), Italy (Marini et al. 2019), and Ethiopia (Belayneh et al. 2016).

The main shortcoming of SPI is that it uses only rainfall data and does not consider other variables, especially temperature. Although rainfall is the leading causative factor for drought in sub-humid regions, recent studies have demonstrated the significance of temperature in affecting the recent trends in the region’s water resources (Gupta et al. 2020; Nath et al. 2017, 2018). The proximal drought index must account for changes in moisture demand caused by increased surface temperature (Pathak and Dodamani 2019). Recently, a new drought index, the standardized precipitation–evapotranspiration index (SPEI), has become popular (Begueria et al. 2010; Vicente-Serrano et al. 2010a, b) to quantify the drought condition in the context of increasing surface warming (Monish and Rehana 2020). The SPEI considers both temperature and precipitation; hence, considered to be a better approach for studying the effects of climate change on drought occurrence (Shaik et al. 2020).

Every region has individual climatic characteristics that interact differently with the influence of anthropogenic activity and climate change. Thus, several recent studies (Li et al. 2020; Pei et al. 2020; Wang et al. 2019; Liu et al. 2018; Tirivarombo et al. 2018; Labudova et al. 2017; Xu et al. 2015; Vicente-Serrano et al. 2015) compared SPEI and SPI to identify drought occurrence at the regional level from various climatic regions.

Several studies analyzed the occurrence and distribution of meteorological droughts using SPI in India (Kundu et al. 2020; Panday et al. 2020; Adarsh et al. 2019; Aadhar and Mishra 2018; Guhathakurta et al. 2017; Joshi et al. 2016). However, in recent times, various researchers used both SPI and SPEI to evaluate the influence of climate change on drought occurrences in India (Singh et al. 2020; Singh and Shukla 2020; Singh et al. 2019; Pathak and Dodamani 2019; Gupta and Jain 2018; Aadhar and Mishra 2017; Alam et al. 2017; Mallya et al. 2016).

The Purulia, RLZ of West Bengal is a sub-humid drought-prone region; many researchers studied meteorological droughts only with the aid of SPI on a long-term time scale (Keskin et al. 2011; Mishra and Desai 2005). Mishra and Singh (2009) prognosticated that Purulia is likely to experience severe drought with more areal extent during 2001–2050. Analysis using SPI in annual and seasonal time scales also have been used to analyze meteorological drought incidences in Purulia for the period 1901–2017 (Patra 2020; Bhunia et al. 2020; Asutosh 2019).

Previous studies in Purulia only used a rainfall-based single index (SPI) to assess meteorological drought and mainly focused on its long-term (annual and seasonal) behavior. In the climate change context, however, temperature and evapotranspiration appear to have a greater influence (Shah et al. 2015) on meteorological drought, especially for sub-humid regions. Thus, evapotranspiration-based SPEI is considered a more suitable index for analyzing the effects of climate change on drought occurrence by various recent researchers (Monish and Rehana 2020; Alam et al. 2017; Das et al. 2016). Due to increasing temperature and monsoon rainfall variability, India’s droughts are becoming more regional, and a spatial shift has been observed toward sub-humid regions (Mallya et al. 2016). RLZ has frequently witnessed short-term droughts in the previous decades (Jha et al. 2013). A comprehensive study of these changing characteristics and trends of meteorological drought is still relatively unavailable in Purulia of RLZ. The effect of the changing behavior of drought on crop production is also lacking in the previous studies in the RLZ of Purulia.

To address the aforementioned research gap in Purulia, the present study aimed to exemplify the changing characteristics of meteorological drought in terms of its duration, intensity, frequency, and spatial–temporal evolution using SPI and SPEI during the 1930–2019 period and their influences on Kharif rice yield in recent times. The present study also intends to analyze the meteorological drought trend and suggest an appropriate meteorological drought index for Purulia. Additionally, for the first time, results have been validated by rigorous field surveys. Such an integrated approach would help us to have a refined view of the region’s spatiotemporal variability of drought, which is necessary for decision-makers and planners to formulate appropriate strategies to minimize the impacts of drought on the overall development of RLZ.

2 Study area

Purulia district, as representatives of RLZ (Roy et al. 2020), lies between latitude 22° 42′ 16.38″ and 23° 41′ 54.36″ N, and longitude 85° 49′ 10.89″ and 86° 54′ 20.64″ E (Fig. 1). The geographical area of the district is 6259 km2. Purulia is registered under Drought Prone Areas Programme (DPAP) by the Department of Land Reforms, Ministry of Rural Development (Bhunia et al. 2020).

Fig. 1
figure 1

Location map of the study area

Full size image

A subtropical monsoon climate dominates the area. The principal source of rainfall in the district is the southwest monsoon. The average annual and monsoon rainfall is 1300 and 950 mm, respectively. Due to its undulated topography and red and laterite soil, nearly 50% of the rain flows away as runoff. Temperature varies over a wide range from 7 °C in winter to 46.8 °C in the summer, causing a high evaporation rate. Purulia district primarily depends on surface water, which is highly susceptible to drought. Increasing drought frequency has been recorded in Purulia during the recent decade (Patra 2020).

Agriculture is the primary livelihood source for more than 75% of the district population. Paddy is the primary crop of the district. A total of 77% of the net-cropped area is under rain-fed Aman rice cultivation (Roy et al. 2022a, b, c). High dependency on monsoon rains for recharging most of its water resources directs towards severe water deficiency and drought in the state if monsoon failures occur. Also, high variability in crop production exists because of the uncertain climate in the state (Roy et al. 2022a, b, c).

3 Materials and methods

The Indian Meteorological Department (IMD) generated gridded daily rainfall (0.25° latitude × 0.25° longitude) (Pai et al. 2014) and daily temperature data (1° latitude × 1° longitude) (Srivastava et al. 2009) from 1930 to 2019 (90 years) were used as the input data in this study. Temperature data were re-gridded to the scale of rainfall using bilinear interpolation. The details of bilinear interpolation by the National Centre for Atmospheric Research (NCAR) are available at https://climatedataguide.ucar.edu/climate-data-tools-and-analysis/regridding-overview. District-level rice production data (from 1991 to 2015) was collected from the Directorate of Agriculture, West Bengal.

Monthly rainfall and mean temperature were used to examine monthly and seasonal drought intensity, duration, frequency, and spatial extent in Purulia during 1930–2019 by calculating the SPI and SPEI. Trends in the SPI/SPEI were detected by the nonparametric Mann–Kendall (M–K) test. Three equal climatic periods have been classified (1930–1959, 1960–1999, and 2000–2019) to determine the changing behavior of recent drought events compared to the past based on drought intensity, duration, and frequency. Based on the rainfall and temperature grid points (30 points), interpolation was employed at a 250 m resolution through inverse distance weighted (IDW) interpolation technique to generate the spatial drought distribution maps and proceed with the homogeneity test.

3.1 Computation of SPI

The SPI (McKee et al. 1993) calculation for a specific time scale and location requires a long-term monthly rainfall series. The first step is to find the probability density function that best describes the distribution of the rainfall data for a selected time scale. Here, the rainfall data (1930–2019) was fitted to a gamma distribution function (Liu et al. 2021). Detailed calculation procedures for gamma distribution and SPI are in the supplementary material. EasyFit 5.5 was employed in the current investigation for gamma distribution. It is a programmer for data analysis and simulation that enables users to fit probability distributions to data samples, choose the best model, and use the analysis’s findings to improve their decision-making.

SPI is an index that was developed to quantify rainfall deficit at different time scales, and thus it enables the assessment of drought severity. It was computed using the monthly rainfall data as defined by (Abramowitz and Stegun 1965; Zarch et al. 2015). Positive SPI values signify greater than average rainfall (wet condition), and negative values imply less than average rainfall (dry condition). The current study evaluated drought using the SPI program (http://drought.unl.edu/MonitoringTools/DownloadableSPIProgram.aspx) to compute SPI at various time scales from monthly rainfall data. The present research calculates SPI at 1-, 3-, and 12-month time scales representing monthly, seasonal, and annual droughts, respectively. Table 1 shows the different categories of drought intensity according to the SPI values given by Mckee et al. (1993).

Table 1 Classification of drought category
Full size table

3.2 Computation of SPEI

SPEI considers both rainfall and temperature components and presents the influence of evaporation variations, which is more responsive to drought occurrences caused by rising temperatures (Liu et al. 2021). Under the warming condition, SPEI can better exhibit dry and wet conditions owing to its combination of rainfall and temperature (Zhao et al. 2017). The Log-logistic distribution provided a better fit for SPEI (Hernandez and Uddameri 2014). The SPEI is obtained by normalizing the water balance into the Log-logistic probability distribution. In the present study, the procedure for calculating SPEI adopted by Vicente-Serrano et al. (2010a) has been used. The measures of the drought intensity classifications according to SPEI are the same as the SPI value (Table 1). The supplementary material includes step-by-step computation instructions for SPEI.

3.3 Analysis of drought characteristics

Drought characteristics, which include drought intensity, frequency, and duration, have been defined using the runs theory proposed by Yevjevich (1967). The following formula calculated the drought intensity:

$$S=\frac{\sum_{n=1}^{T}\left[{S}_{\mathrm{SPI}/\mathrm{SPEI}}-K\right]}{T}$$

where S refers to drought intensity, SSPI/SPEI is SPI or SPEI value below the threshold, K is a drought threshold value, set to be less than or equal to − 0.99 in the present study, means the drought level is mild, T refers the duration of the drought. The drought frequency calculation formula is as follows (Liu et al. 2021):

$$P=\frac{n}{N}*100\mathrm{\%}$$

where N denotes the time and n signifies the number of droughts during the time.

3.4 Method of trend analysis

The rank-based Mann–Kendall (M–K) trend test, a nonparametric statistical test, is used frequently to evaluate significance in a monotonic increasing or decreasing trend in hydro-meteorological time series (Sicard et al. 2010; Kumar et al. 2009) including a series of drought indices (Damberg and AghaKouchak 2014). The M–K test had high computational efficiency and was not sensitive to measurement error, missing values, and outlier data. Therefore, in this study, M–K test has been used to detect the significant trends in SPI and SPEI time series. In this study, the M–K trend test was performed in R at a significance level of 0.05 and was used to determine the drought characteristic trend on different time scales (1 month, 3 months, and 12 months). The supplemental material contains comprehensive computation methods for the M–K trend.

3.5 Sampling method and data collection for field verification

Fieldwork followed in September 2015 and 2019 to develop a more comprehensive understanding of short-term droughts and their impacts on crop production in the study site. A multiple-stage sampling approach was applied. Kashipur block was chosen because the block experiences irregular monthly rainfall and is highly vulnerable to drought (Roy et al. 2022a, b, c). Ten villages of Kashipur (Fig. 2) were recognized and selected based on the percentage of the tribal population, who are primarily small and marginal farmers. The farmers were selected for the study using the probability proportional to size sampling method among the ten villages, and individual farmers were selected using the stratified random sampling method (Delfiyan et al. 2021). The sample size of the farmer unit was determined by the following equation (Sathaiah and Chandrasekaran 2020):

$$n=\frac{N{z}^{2}p(1-p)}{N{d}^{2}+{z}^{2}p(1-p)}$$

where n is the sample size, N is the total number of small and marginal farmers (21,245), z denotes confidence level (at 95% level z is 1.96), d signifies error limit (0.05), and p refers to estimated population proportion (0.5 maximizes the sample size). A total of 375 small and marginal farmers were selected for the study and interviewed to quantify the impact of the drought on crop production in 2015 and 2019 (Tables 2 and 4).

Fig. 2
figure 2

The villages in the Kashipur block used for field Survey: a: Pabra Pahari b: Kashidi c: Sura d: Seja e: Jagannathdi f: Jibanpur g: Bhatin h: Ranjandi i: Bodma j: Lari

Full size image
Table 2 Basic socio-economic outlines of the respondents’ farmers in villages of Kashipur
Full size table

A pre-tested questionnaire consisting of open-ended questions, multiple choice, and focus group discussions (FGD) was used to explore farmers’ experiences with droughts. A structured questionnaire was designed for the present study (Mare et al. 2018; Keshavarz et al. 2013; Udmale et al. 2014) to quantify the impact of the drought in 2015 and 2019. Face-to-face interviews were conducted in the local language, mostly with heads of households but sometimes with other family members. Interest in drought-related issues was high in the area, and the response rate was almost 100%. The checklist-guided discussions with focus group members, both men and women, concentrated on the effects of drought on Kharif crop production.

4 Results

The 1-month SPI provides a monthly estimation of rainfall (Ji and Peters 2003), while the 3-month SPI reflects medium-term (seasonal) trends in rainfall patterns (Potop et al. 2012), and the 12-month estimate provides an annual estimation of water condition. Therefore, this study used the SPI/SPEI values at 1-, 3- and 12-month scales to explore the drought variability and their duration, intensity, and frequency at short and long-term (monthly, seasonal, and annual time scales), respectively. The time series of all drought indicators were examined from 1930 to 2019, which were split into three intervals of 30-year (1930 − 1959, 1960–1989, and 1990–2019) duration to identify recent changes in meteorological drought events. The present study also focused on the impact of drought on rainfed agriculture of such low permeability RLZs.

4.1 Temporal variation in the SPI and SPEI

Usually, the SPI/SPEI values at the 12-month time scale indicate the status of annual drought (Wang et al. 2014), and the SPI/SPEI values at a 3-month scale are appropriate indicators of the status of seasonal drought. The pre-monsoon, monsoon, and post-monsoon values were denoted by the May, September, and January SPI/SPEI values, respectively. Figure 3 shows the annual and seasonal SPI/SPEI series of the entire Purulia during 1930–2019.

Fig. 3
figure 3

Variation in annual a, pre-monsoon b, monsoon c, and post-monsoon (d) SPI and SPEI values at the 3-month time scale in Purulia during the years 1930–2019

Full size image

Out of 90 years (1930 to 2019), an annual drought occurred in 47 years (52%). Severe to extreme annual drought (− 1.5 to >  − 2) occurred in 1962, 1966, 1979, 1992, 2000, 2001, 2010, and 2015. The maximum number of severe to extreme annual drought years (63%) have been identified by the SPI/SPEI in the recent period (1990–2019) compared to the first (1930 − 1959) and second (1960–1989) periods.

Pre-monsoon, monsoon, and post-monsoon drought events occurred 44, 49, and 41 times respectively from 1930 to 2019. In pre-monsoon and post-monsoon seasons, the maximum number of moderate to severe droughts (− 1.0 to − 1.5) have been identified by the SPI/SPEI in the first (1930 − 1959) and second (1960–1989) periods. But in monsoon season, the maximum number of severe to extreme droughts (− 1.5 to − 2.0) has been found by the SPI/SPEI in the recent period (1990–2019) only.

The monthly meteorological drought assessment found that 56% of the monthly drought was concentrated from June to August, while other months experienced 42% of monthly droughts. Extreme droughts were prevalent from May to October, whereas severe droughts were found between February and April. Above 50% of the years, extreme to severe meteorological droughts occurred from June to October. Around 80% of the year, mild droughts are concentrated in November and December.

4.2 Drought duration, intensity, and frequency

The drought duration identified by the SPI is generally consistent with that by the SPEI. Annual drought duration identified by both SPI and SPEI was 6–8 months during the years 1930–1990, but after 1990, drought duration was 5–7 months. Both SPI and SPEI identified a maximum annual drought duration (37 months continuous) from Sep 2000 to Aug 2006. For seasonal drought analysis, a 5–7-month duration has been found in the recent period (1990–2019), and a 3–5-month duration has been identified in 1930–1990 years.

Drought intensity evaluated from the SPI for most years was greater than those from the SPEI in Purulia. For the 12-month time scale, extreme drought intensity (> − 2) was found to occur for times during the years 1990–2019 (1992, 2000, and 2010). In the 12-month timescale, mild drought (i.e., − 0.99 to 0) intensity was observed in 73% of the years, while 20% of the years were under moderate to severe drought conditions (i.e., − 1.49 to − 1.0) and 7% of the years exhibited extreme condition (> − 2), 80% of them occurring after 1990.

For seasonal drought conditions, 65%, 24%, and 11% years belonged to mild, moderate to severe, and extreme intensity, respectively, in the post-monsoon. While 58% and 41% years under moderate to severe and mild intensity in the pre-monsoon and 53% and 45% years under moderate to severe and mild intensity in monsoon.

Most extreme to severe droughts have been found in the monsoon season, 85% of them occurring during recent years (1990–2019). Moderate drought years occurred mainly in the monsoon and pre-monsoon seasons, concentrating in the first (1930 − 1959) and second (1960–1989) part of the century. In the second part, around 60% of the years experienced mild droughts in all seasons.

The frequency of drought events in the area has been estimated in 1-, 3-, and 12-month time scales by SPI and SPEI. Annual and monsoon extreme drought frequency increased in Purulia for the 1990–2019 period compared to the period of 1930–1990. For pre-monsoon, monsoon, post-monsoon, and annual time scales, moderate and severe drought frequency gradually increased for 1990–2019 compared to the period of 1930–1959 and 1960–1989. But mild drought frequency gradually decreased from 1990–2019 compared to 1960–1989.

4.3 Spatial variability of drought

The spatial distribution of drought maps has been generated by Inverse Distance Weighting (IDW) interpolation method in the ArcGIS platform. A spatial extent with the intensity of seasonal drought was analyzed in 3 recent drought years (2010, 2015, and 2019) by SPI and SPEI to understand the spatial variation during the current decade (Fig. 4). Drought in the area seems to affect the area first in the west and progresses further to the east and south.

Fig. 4
figure 4

Spatial variation of drought indices SPEI (A) and SPI (B) for the pre-monsoon, monsoon, and post-monsoon seasons in the drought years 2010, 2015, and 2019

Full size image

In the pre-monsoon of 2010, the district was under mild to no drought conditions. Due to the deficit of monsoon rainfall, the scenario changed considerably in the monsoon, with 91.45% of the area found to be under extreme to severe drought as identified by SPI. In comparison, 55% area could be identified under severe drought by SPEI. However, post-monsoon rainfall drought conditions recovered significantly with the prevalence of mild-moderate drought over 85% of the area in 2010.

In 2015, during the monsoon season, 85% of the area came under severe drought and the other 10% was under moderate drought conditions. In pre-monsoon and post-monsoon, the area experienced mild drought to no drought at all.

In 2019, a severe drought was observed in the monsoon months in 80% of the district. Drought beginning in western areas extended to east and southwards. In pre-monsoon, mild drought conditions prevailed in 90% area of the district. But in the post-monsoon, wet condition prevailed all over the area due to good post-monsoon rainfall.

Spatial analysis of propensity of drought incidence identified the western, eastern, and southern blocks (Jaldhi, Joypur, Baghmundi, Arsa, Purulia II, Balarampur, Kashipur, HuraBandoyan, Barabazar, etc.) to be prone to severe drought. The blocks of the northern part like Para, Raghunathpur, Nituria, and Santuri of the study area were prone to mild or no drought.

4.4 Effect on crop

Rainfed paddy cultivation is a major livelihood for this region. The rain-dependent nature of Kharif agricultural practices in the district makes it vulnerable to soil moisture fluctuations, which is a function of precipitation and temperature during the monsoon season. The Pearson correlation coefficient of monsoon-SPI and SPEI with Kharif rice production anomaly was performed to understand whether the recent meteorological drought affected the agricultural production in the area (Madadgar et al. 2017). The analysis of correlation coefficients displays that the monsoon SPEI exhibits a stronger association with Kharif rice production (r2 = 0.6542) than does SPI (r2 = 0.6144; Fig. 5). Figure 5 demonstrates that the negative SPI and SPEI indices (< − 1) indicating meteorological droughts recorded in the years 1998, 2000, 2005, 2010, and 2015 were associated with Kharif crop failures in Purulia district leading to agricultural drought.

Fig. 5
figure 5

Pearson’s correlation coefficient functions of the Kharif rice production with monsoon a SPI and b SPEI from 1991 to 2015

Full size image

4.5 Trend analysis

M–K trends and Sen’s slope were calculated from the SPI/SPEI values at the 1-, 3-, and 12-month time scale during the years 1930–2019 in Purulia. On a monthly scale, droughts identified by the SPI increased in August and decreased in May and no trends were found in other months. While, those calculated by the SPEI increased in February, July, and August (Table 3). At the seasonal scale, drought analyzed from the SPEI has increasing trends in monsoon seasons, but for that calculated from the SPI, no trends were detected in all seasons. At the annual scale, the increasing annual droughts were identified by the SPEI while the SPI revealed no trend.

Table 3 M–K trends calculated from the SPI/SPEI at the1-, 3-, 12-month time scale during years 1930–2019 in Purulia
Full size table

5 Discussions

Analysis of monthly and seasonal drought characteristics over the last 90 years in the RLZ indicates that the most extreme to severe droughts were found in monsoon seasons in the recent century (1990–2019) while maximum moderate droughts were identified in monsoon and pre-monsoon seasons during the first (1930–1960) and second parts (1960–1990). The area was affected by long-term (8–10 month duration) droughts during the years 1960–1990, which were less frequent. But after 1990, the area experienced more frequent short-term (4–5-month duration) droughts. Since 1990, frequent occurrences of extreme to severe droughts for a short duration in monsoon months over the sub-humid district could be attributed to rainfall variability and rising temperature due to climate change at a regional scale. Climatic variability and occurrence of extreme events during monsoon season in the sub-humid region of India have been attributed to climatic change by various researchers (Sam et al. 2020; Parida and Oinam 2015; Udmale et al. 2014). An increase in evapotranspiration due to global warming would likely play a major role in affecting drought dynamics in the sub-humid region of India (Gupta and Jain 2018) in the future.

Trend analysis of monthly drought also denotes the increasing intensity and frequency of drought in July and August when maximum monsoon rainfall is expected. A statistically significant rising trend in monsoon and annual drought at a 95% confidence level have been observed by the analysis of the SPEI in Purulia. Kumar et al. (2013) had similar findings from Central India where drought frequency increased during the period 1990–2010. Similarly, an increasing trend of frequency and magnitude of the monsoonal drought was observed in the Eastern regions of India (Das et al. 2016). Drought events have increased also in South Asia in recent decades (Aadhar and Mishra 2017).

From monthly and seasonal SPI/SPEI analysis, non-uniform spatiotemporal characteristics (duration, intensity, and frequency) were observed in drought occurrences. In 1998, 2005, and 2015, severe monsoon drought leading to crop failure in the Kharif season could be detected but annually no drought condition was shown because annual rainfall was the same as the long-term annual mean. On the other hand, in 1993, 1996, 2004, 2007, 2011, and 2012, annual rainfall was greater than the long-term annual mean, yet drought conditions prevailed for pre and post-monsoon periods. The analysis indicates that the onset and monthly quantum of monsoon rainfall is the driver of drought notwithstanding whether the annual rainfall is greater or less than the normal rainfall. Thus, monthly or seasonal SPI/SPEI estimation can be taken up as a more appropriate indicator of drought assessment and preparedness planning. Similarly, Guhathakurta et al. (2017) and Joshi et al. (2016) concluded that the Eastern regions of India showed an increasing incidence of monthly drought during the second half of the twentieth century.

Results show that SPI fared equally well as SPEI and successfully detected the most significant drought events before 2000. SPEI could identify a larger number of drought years after 2000 than SPI in Purulia. Moderate drought conditions could be placed in 5 specific years (1993, 2003, 2004, 2005, and 2012) in recent times (1990–2019) by SPEI, while SPI could pick up no such drought conditions. Similar findings were obtained in another investigation (Tefera et al. 2019; Labudova et al. 2017; Wang et al. 2015). According to the literature, both indices could detect major drought incidences in the first half of the twentieth century; however, SPI was unable to capture the degree of drought severity suggested by SPEI in the later part of the twentieth and early twenty-first centuries. Additionally, the SPI values seem higher than the corresponding SPEI values. Other investigations (Tefera et al. 2019; Labudova et al. 2017; Wang et al. 2015) have shown the similar differences between SPI and SPEI. When temperature changes are combined with precipitation, the findings may differ from when precipitation alone is used. SPI solely uses rainfall data as an input, but SPEI uses rainfall and temperature data as input, allowing it to account for the effects of temperature variation on drought assessment. However, variances between SPI and SPEI values do not imply that they provide wholly different results. SPI can be equally effective as SPEI in locations with low-temperature changes. The correlation of Kharif rice production with SPEI is much more significant than the SPI’s. The above differences between the SPEI and the SPI are mainly attributed to the rising temperature in Purulia. Several other researchers have found similar results (Pathak and Dodamani 2019; Aadhar and Mishra 2018; Tirivarombo et al. 2018). SPEI performs well in capturing drought occurrences in sub-humid RLZ in a climate change context.

Spatial analysis of recent droughts could identify community blocks (Jaldhi, Baghmundi, Balarampur, Bandoyan, Manbazar-II, Barabazar, Para, Raghunathpur, Nituria, Kashipur, Hura, etc.) of the districts to be prone to severe to extreme drought. The spatial analysis identified such blocks in need of special assistance at the time of drought and helps in preparedness planning by augmenting water and agricultural resources.

6 Field validation

The changing characteristics of meteorological drought and associated crop failure have been validated using field surveys in the recent 2 consecutive drought years (2015 and 2019) to assess the accuracy of the assessment. Ten villages of predominantly rural and tribal populations located in the Kashipur community development block of the Purulia district were surveyed. These communities have also experienced extreme to moderate droughts in the past years 2005 and 2010 and hence, make good targets for comparative case studies.

A total of 375 farmers’ households in the selected villages were surveyed in 2015 and 2019 (Table 4). The interviews investigated their experience of drought occurrences in 2015 and 2019, the time and intensity of such droughts, and their impacts on crop production. The villagers informed us that they faced drought in June, from August to October 2015, and from June to August 2019. Farmers also mentioned the intensity of drought in 2015 was higher than in 2019. More than 95% of surveyed households reported that Kharif rice production was significantly reduced in these 2 drought years, however, lesser in 2019 (Table 4). The field validation thus could establish the relationship between the indices derived existence of short-term monsoon droughts with human experience and crop failure.

Table 4 Village wise sample size and Kharif rice production failure in two drought years 2015 and 2019
Full size table

7 Conclusion

In the present study, multi-scalar meteorological drought indices, SPI, and SPEI have been used to identify the changing characteristics of drought conditions for 1-, 3-, 6-, and 12-month time scale in sub-humid RLZ (Purulia) over 90 years (1930 to 2019). The Mann-Kendell and Sen’s slope tests were applied to evaluate trends in monthly drought and spatiotemporal drought characteristics obtained from SPI and SPEI during the recent decade. Further, the relationship between drought and Kharif (monsoon) rice production was analyzed to identify crop failures during identified droughts. Results indicate a minor change in the long-term droughts (12-monthly) condition while a major change is observed in short-term (1-, 3-monthly) droughts on a monthly time scale, extreme droughts were found from May to October, while severe droughts were found in the lean periods of February to April. Mild droughts could be found in the months of concentrated in November and December. Among the cases of drought incidences in the region, 50% were found to be of extreme to severe (− 2 to − 1.5) nature, which occurred from June to October between 1990 and 2019. A majority (53%) of mild droughts could be identified to have occurred before the 1960s. The long-duration drought of 8 to 10 months, more prevalent during 1930–1960, appeared to have been replaced in recent times by frequent short-duration droughts in the region. Both drought frequency and intensity appeared to have increased in recent decades. At a seasonal scale, extreme to severe droughts is observed to be more prevalent in the monsoon season of the post-1990 period. Frequent meteorological droughts in monsoon months consequently imparted the loss of the Kharif crop production of the RLZ. From the perspective of the persistent rise of temperature and variability in the monsoon rainfall the changing characteristics of meteorological droughts over the RLZ need immediate intervention to initiate drought preparedness and management planning. The relation between drought and Kharif rice production indicates that SPEI functioned as a better index compared to SPI, especially in recent years, when the temperature is significantly higher, promoting more evapotranspiration and dryness. Trend analysis of meteorological drought through the SPEI index revealed a significant increasing trend in July, and August in the monsoon season, which indicates that the sub-humid RLZs may undergo frequent droughts during monsoon months in the future due to increasing temperature. The close correspondence of monsoon meteorological droughts in the Kharif season in the lateritic terrain of poor percolation capability is significant for the region’s water resource and agriculture management. The findings of our study can be helpful for policymakers, planners, and all the other stakeholders to obtain an initial idea into better preparedness for droughts, to be monitored monthly for identification in higher resolution. The study also suggested suitable drought indices for RLZ in the context of rising temperatures having a more pronounced effect on drought incidences of recent times. This understanding of the changing nature of meteorological droughts and increased incidences of short-term droughts in the monsoon season is expected to enhance our ability to plan, manage, and implement sustainable water use and agricultural practices in the face of climate change.