1 Introduction

Drought affects the people and the economy of northwestern India more than any other climatic extreme (Wilhite 1993a, b; Wilhite and Vanyarkho 2000; Pandey et al. 2010). Among all the natural hazards, droughts are the most severe around the globe, occurring in every region and spreads over lager area than earthquake and floods (Wilhite and Glantz 1985; Tallaksen and Van Lanen 2004; Wilhite et al. 2007; Zhang et al. 2012; Jain et al. 2015). The recent studies on hydrologic extremes indicate that even after advancement in technology every region in the world is susceptible to ill-effects of droughts. For instance, the European heat waves in 2003 reduced the agricultural production (Parry 2007). The severe drought events during 2011–2012 over half of the USA, Korean Peninsula and Eastern Africa have put the region into the disaster zone category (Dutra et al. 2013; Karl et al. 2012; Mosley 2012; USDA 2012). Droughts are the most complex climatic hazard, as it is quite difficult to identify their onset and termination (Wilhite 1993a, b). Meteorological droughts occur due to a deficiency of rainfall during a season or year at given place compared to average seasonal or annual rainfall of that place. It causes reduction in the water table, lowering of reservoir storage, and crop failure. In recent decades, the water demand has grown due to rapid increase in population, irrigation, and industrial development (Shiklomanov 2000; Beniston 2003; Rijsberman 2006; Mishra and Singh 2010; Maccioni et al. 2014). The variability of rainfall and over use of water resources are responsible for water scarcity and reduction in level of groundwater in major portion of India (Kumar et al. 2005; Sakthivadivel 2007). In literature, many studies have been done using the modeling approach, the characteristics of drought were extracted from rainfall data (Shiau 2006; Modarres 2007; Mirakbari et al. 2010; Song and Singh 2010; Madadgar and Moradkhani 2011; Ganguli and Reddy 2012; Mirabbasi et al. 2012; Janga Reddy and Ganguli 2012; Chen et al. 2013; Lee et al. 2013; Ma et al. 2013; Yoo et al. 2013; Yusof et al. 2013; Zin et al. 2013; Rauf and Zeephongsekul 2014; Tosunoglu and Can 2016).

Haan (1977) defined return period of drought, as the average time between the events of certain magnitude or less. In literature, several methods were discussed for the computation of return period of droughts (Clausen and Pearson 1995; Chung and Salas 2000; Pandey and Ramasastri 2001a, b; Kim et al. 2003; Fernández and Salas 1999a, b; Mishra et al. 2009; Amrit et al. 2017). The severity of the drought event is one of its important feature, since it is directly associated to hazardous effects of drought (Hayes et al. 2010). The monthly precipitation data of 46 years (1951–1996) duration has been used for analyzing spatiotemporal relationship of occurrence of droughts and its severity over Korea and East Asia. The study indicated that in Korea during 1980s the frequency of drought get increased, and it ranges between 2–3 and 5–8 years (Min et al. 2003). Edossa et al. (2010) analyzed the characteristics of drought using the hydro-meteorological parameters in Awash River basin of Ethiopia. The spatiotemporal analysis of meteorological drought indicated that in Middle and Lower Awash River Basin droughts were very frequent and severe. The severity of drought events can be estimated in many ways. The magnitude of deficit and duration of hydrological drought are linked, as the deficit accumulates during the period for which drought event lasts (Shiau and Shen 2001; Hisdal et al. 2004; Wong et al. 2013; Loon and Laaha 2015). The several studies have been done on the spatiotemporal variation in drought severity across the globe (Mihajlovic 2006; Vicente-Serrano 2006; Blenkinsop and Fowler 2007; Elagib 2009; Santos et al. 2010; Yan-jun et al. 2012; Gallant et al. 2013; Masud et al. 2015; Rahmat et al. 2015; Li et al. 2016; Khadr 2017). Kleppe et al. (2011) analyzed the severity of drought events during medieval times in Lake Tahoe Basin. The estimation of severity (i.e. magnitude of deficit) and duration of drought events northern Sierra Nevada carried out by using hydroclimatic conditions in Fallen Leaf Lake, California. The water balance calculations indicated that the annual rainfall departed by more than 60% of normal leads to lowering of shoreline of lake ranging from 40 to 60 m from its modern elevation.

Drought is most hazardous climatic phenomenon whose impacts become more severe as it lasts for a longer duration. The persistence of drought event creates worst condition for the society. A persistent drought event is defined as the occurrence of two or more consecutive drought years, such as 2-, 3- and 4-consecutive years. The worst impacts draw the attention of many researchers toward the persistent drought events (Karl 1983; Stahle and Cleaveland 1992; Zaidman et al. 2002; Fye et al. 2003; Huang et al. 2005; Van der Schrier et al. 2006; Marsh et al. 2007; Manuel 2008; Demuth 2009; Seager et al. 2009). Karl (1983) analyzed the spatial characteristics of drought in USA employing Palmer Drought Severity Index (PDSI), and it has been observed that compared to the coastal region in east and west the interior of the USA have longer drought persistency.

The northwestern, central and peninsular regions are more susceptible to drought compared to the other parts of India. Pandey and Ramasastri (2001a, b) show that the average return period of drought decreases gradually from the arid to the semiarid and sub-humid regions. Smakhtin and Hughes (2007) used monthly rainfall data to analyze drought characteristics using the Decile index (Gibbs and Maher 1967), the Effective Drought Index (EDI) (Byun and Wilhite 1999), the Standardized Precipitation Index (SPI) (McKee et al. 1993) and departure from long-term average rainfall. They found that the arid and semiarid regions faced more frequent and severe droughts than other regions.

The northwestern India comes under arid and semiarid region susceptible to frequent and severe drought events. The high variability of rainfall in these regions leads to water shortages, reduced crop production, and impacts the local population. It is very important to improve the level of understanding of drought characteristics to deal with the severe impacts of drought events.

Therefore, the primary objective of this study is to assess the drought events in terms of their return period, severity (i.e. magnitude of deficit), and persistence, in various districts of northwestern India. It is believed that the study would be very helpful in rapid response to drought events and mitigation plans based on long-term regional pattern.

2 Methodology

2.1 Study area and data used

The study area covers four states in northwestern India, Punjab, Haryana, Rajasthan and Gujarat (Fig. 1). The average annual rainfall in India is about 1190 mm (Subramanya 2005) most of which (about 85%) falls during the monsoon period which starts from June and ends in September in central parts, and from June to October/November in the peninsula and northeastern regions. The mean annual rainfall of the study region is 700 mm, below the national average. The rainy season starts in early July continues till September. Rainfall in the area is erratic with high inter-annual variability.

Fig. 1
figure 1

Location map of study area

Full size image

The state of Punjab with population of 27,704,236 covers an area of 50,362 km2 of northwestern part of India. Rainfall ranges from 900 mm in the mountain foothills to 400 mm in plains. The fertile lands of Punjab process about two-third of the total gains produced in India annually. The major crops grown in the state are wheat, paddy and cotton. Haryana state has an area of 44,212 km2 having population of 25,353,081. Rainfall ranges from 300 to 800 mm and is unevenly distributed. The vegetation of Haryana comprises the major growing crops of wheat, paddy, sugarcane, cotton and oilseeds. Rajasthan with an area of 342,239 km2 becomes the largest state of India having human population of 68,548,437. The mean annual precipitation in the state ranges from 100 mm in Jaisalmer to 900 mm in the Jhalawar district. The state of Rajasthan has diversity in vegetation across the different parts. For instance, the vegetation comprises of only thorny bush and cactus in the desert part of the state due to less availability of water, while the maize, guar, sesamum, wheat, mustard, gram, fruits are the major crops in the irrigated northwestern part of Rajasthan and major cash crops like cotton and sugarcane are grown in the black soil region of Udaipur, Chittorgarh, and Sirohi districts.

The state of Gujarat has the population of 60,383,628 covering an area of 196,204 km2. The mean annual precipitation in the state typically ranges from 400 to 800 mm, except in the Dangs district where the maximum mean annual rainfall is 1800 mm, higher than in any other district of the State. The major crops of the state include bajra, groundnut, cotton, rice, maize, wheat, mustard, sesame, pigeon pea, green gram, sugarcane, mango, banana, sapota, lime, guava, tomato, potato, onion, cumin, garlic, isabgol, and fennel. Figure 1 represents the index map of study area.

A code number is assigned to the each district of the study region and name of the district corresponding to each code number is given in Table 1.

Table 1 Names of the district corresponding to each code
Full size table

This study used monthly rainfall data for 113 years (1901–2013). The data for the years 1901–2000 were supplied by the India Meteorological Department (IMD). Data for Punjab, Haryana and Gujarat, covering the period 2001–2008, were obtained from the India Water Portal. Data for the period 2009–2013 were obtained from the website of IMD. For Rajasthan, data covering 2001–2012 were from Department of Water Resource, Government of Rajasthan. Missing data were estimated computing the mean of the rainfall of the surrounding stations. The rainfall data for three districts (Navasari and Valsad in Gujarat, and Karauli in Rajasthan) were in complete and these stations are not included in the analysis. Documented drought events in the literature were used to compare and verify our results (Pandey and Ramasastri 2001a, b; Dutta et al. 2015).

The percentage deviations from corresponding long-term averages, in annual and seasonal rainfall, were calculated to identify drought seasons or drought years, and to assess the severity of drought events in each district of the study area. Seasonal rainfall was defined as the total rainfall from June to September. Drought events were identified using the IMD definition—a period when the rainfall deficit is more than 25% of the corresponding mean (Ministry of Agriculture and Irrigation 1976). Based on the percentage deviation, the annual and seasonal droughts were categorized as moderate, severe or extreme, using the values given in Table 2. A persistent drought event was defined as two or more consecutive drought years.

Table 2 Drought classes on the basis of rainfall departure from mean.
Full size table

The percentage departure of seasonal rainfall of every year (i.e. 1901–2013) from the corresponding long-term average was estimated for each district of the study area. Based on the rainfall departure, the drought events are classified as moderate, severe and extreme according to the range given in Table 2. Further, the number of various droughts (i.e. moderate, severe and extreme) occurred over the period of 113 years in each districts of the northwestern India has been computed.

The drought return period for each district was estimated as the ratio of the total number of years of data analyzed by the number of years having rainfall deficiency of more than 25%.

$$ T = \frac{N}{n} $$
(1)

where T is Drought return period, N Number of years of rainfall records analyzed, and n number of years having rainfall deficiency of more than 25% out of N. The frequency (F) can be mathematically described as the reciprocal of return period (T), (i.e. \( F = \frac{1}{T} \)).

The return periods of annual and seasonal severe droughts events were similarly derived by dividing the total number of years of analyzed data by number of drought events having magnitude of deficit of more than 45%.

The rainfall records were divided into three eras: 1901–1950, the first half of the twentieth century; 1951–1981, the mid to late twentieth century; and 1981–2013 the most recent three decades. The eras were analyzed to identify changes in drought return period.

The methodology used in the research is presented in Fig. 2.

Fig. 2
figure 2

Flow chart represents the step-wise procedure used in the analysis

Full size image

3 Results and discussion

The northwestern part of India has arid and semiarid climatic conditions, with a rainy period of less than 3 months. There is greater inter-annual and intra-annual variability in rainfall in this region than in other parts of India (Pandey and Ramasastri 2001a, b; Vittal and Kar 2010). During drought period, huge water scarcity and reduced crop production have in observed in this region.

Our analysis shows that there have been maximum annual rainfall occurrences of about 2.0–3.5 times the mean annual rainfall over the period 1901–2013. The minimum annual rainfall varied between 0.2 and 0.4 times the mean annual rainfall across the districts in the study area.

3.1 Return period

The analysis indicated that in Punjab the number of drought years varies from 22 to 36. The frequency of meteorological drought in Punjab is once in 3 years for all the districts except Gurdaspur, Ludhiana and Roopnagar, where the frequency was one in 4 years. Figure 3a shows the distribution of drought return periods in Punjab for the analysis period. As shown in Fig. 3b, almost half of the districts in the Haryana have faced droughts once every 3 years, and the remaining once every 4 years. Rajasthan is drier than the other states in the study area. The western most areas are desert. The state has an average drought frequency of once every 3–4 years. Bharatpur and Chittorgarh are two districts, located in east of the State with a drought frequency of once in 5 years. The spatial variation of drought return periods in the State of Rajasthan is shown in Fig. 3c. The return period of drought in the districts located toward western part of Gujarat is 3 years. The frequency of drought in districts located toward east of the state estimated to be once in 4 years. Bharuch and Dangs have a drought frequency of one in 5 years. The spatial distribution of drought return periods in Gujarat State is shown in Fig. 3d. The results obtained in this study is in agreement with the studies done in past by various researcher in the region. Dutta et al. (2015) assessed the drought in Rajasthan using remote sensing and SPI and estimated the drought frequency of once in 3 years in Jaisalmer, Bikaner, Barmer and Jalore, and once in 4 years in Ajmer, Bundi and Dungarpur districts.

Fig. 3
figure 3

Spatial variation in drought return period in a Punjab state, b Haryana state. c Rajasthan state, d Gujarat state

Full size image

In the western and northern parts, where the mean annual rainfall is below 600 mm and potential evapotranspiration is about 2200 mm, droughts occurred once every 3 years. Figure 4 shows spatial variation in return periods of severe drought for the whole northwestern India.

Fig. 4
figure 4

Spatial variation in frequency of severe drought in northwestern India

Full size image

In the western districts, the frequency of severe drought events is between once every four to once every six years. Figure 4 clearly represents that the north, northeast and southern part of the study region experienced the severe drought with the frequency of one in 7–10 years. The return period of severe droughts in districts located toward east is one in 10 years or more. The frequency of extreme droughts in Banskantha, Kutch, Porbandar, Jamnagar, Sirsa, Barmer, Ganganagar and Jaisalmer varies from once in 6 to 10 years while in Ferozepur, Muktesar, Moga, Mehsana, Patan, Surendranagar, Junagarh, Fatehabad, Jhajjar, Sirohi and Nagaur it ranges from 11 to 15 years. In Bhatinda, Ahmedabad, Bhavnagar, Kheda, Rajkot, Sonepat, Bikaner, Jalore, Jodhpur, Pali, Sikar and Rajasmand, it ranges between once in 16 and 19 years, while in rest of the districts, it is 20 years or more.

The changes in drought return period in northwestern India for the three analytical eras is presented in Fig. 5. Figure 5a, b shows that the return period of drought decreased in parts of the study region between the early and the mid to late twentieth century. The analysis of data indicates that droughts were more frequent during the period from 1951 to 2013 compared with the period 1901–1950. For almost all districts, the frequency of drought in the most recent period (1981–2013) has been higher than in the preceding periods as shown in Fig. 5c. Exceptions are the Bundi, Ajmer and Chittorgarh districts.

Fig. 5
figure 5

Spatial distribution of drought frequency a 1901–1950, b 1951–2013, c 1981–2013

Full size image

Pingale et al. (2014) estimated the spatiotemporal trends of rainfall and temperature in 33 urban centers of Rajasthan. The significant decreasing trend of seasonal and annual rainfall has been observed, while average temperature shows the significant increasing trend. The increasing temperature may cause increase in drought severity, as the temperature rises the rate of evaporation from water bodies and land surface will rise and causes moisture reduction in soil and subsequently reduces the crop production. The decrease in seasonal or annual rainfall during different periods is the major reason of increasing drought frequency in various parts of study area.

Table 3 identifies the districts where notable changes in drought return period have occurred between three eras and shows the associated changes in the return period.

Table 3 Drought return period in selected districts over the three eras
Full size table

3.2 Drought severity

The analysis of seasonal rainfall deficit for the monsoon season shows extreme severity droughts (see Table 1) occurring in the Ferozepur, Fatehgarh Sahib, Moga, Sangrur and Mukteshar districts of Punjab. In Gujarat, Kutch, Porbandar and Banskantha experienced the greatest rainfall deficit (80–85%), falling in the severe drought class. Severe deficits occurred in 1904, 1905, 1911, 1918, 1931, 1939, 1985, 1986 and 1993. The State of Rajasthan has the highest seasonal rainfall deficits, over 90% in the districts of Bikaner, Jalore, Jaisalmer and Rajasmand. Severe (equal to or more than 60% deficit relative to long-term average) rainfall deficits occurred 1904, 1911, 1915, 1918, 1938, 1968, 1969, 1974, 1987 and 2002. The highest rainfall deficits in Haryana fall in the range of 60–70%. However, it reaches between 80 and 85% in Hisar, Jinnd, Rohtak and Sirsa in 1987. The years of severe rainfall deficits in Haryana are 1905, 1907, 1918, 1938, 1947, 1987, 1999 and 2005. Figure 6 shows the spatial variation in drought severity in study region.

Fig. 6
figure 6

Spatial variation in drought severity in northwestern India

Full size image

The years 1918 and 1987 are observed to be very severe drought years of the century. The reduction of 7 million tonnes was observed in food production due to severe drought condition in 1987 (CAZRI-ICAR 2006). Drought has both socioeconomic and environmental effects which are not easy to quantify because of its non-structural nature (Jayasree and Venkatesh 2015). The immediate impact of drought can be seen on agriculture of the region. In the period of severe droughts, there is increase in evaporation rates which reduces moisture in the soil leading to reduction in crop production (Murthy et al. 2015). In addition to this the region face, the lack of drinking water for both human and livestock which is another major concern during drought events. Droughts thus have multiplier effect on agricultural production; their impact is felt during the subsequent years due to deficit ground water recharge, land degradation, fall in investment capacity of farmers, rise in prices, reduced grain trade, and power supply.

3.3 Drought persistence

Figure 7 shows the spatial variation in maximum drought persistence. Droughts longer than 4 years occurred in 13 districts in Punjab. Moga and Mukteshar experienced droughts lasting 3 years. Gurdaspur, Ferozepur and Ludhiana had 2-year droughts. In Gujarat, the longest droughts lasted 5 years (1938–1942), in Porbandar and Jamnagar. Droughts of 4 years occurred in Anand, Narmada, Dahod, Surat, Dangs, and Amreli. All the other districts had droughts lasting no more than 3 years, except Junagarh, where the longest droughts lasted only 2 years. In the Ambala, Yamunanagar, and Hisar districts of Haryana droughts persisted for two consecutive years at most, and 3 year droughts occurred in Kaithal, Panipat, Mahendragarh, Rewari, Gurgaon, and Faridabad. In most areas of Rajasthan, droughts lasted between 3 and 4 years. In Bhilwara, Baran, Tonk, Chittorgarh, and Udaipur drought persistence was only 2 years. Droughts of 5 years were experienced in Alwar (1937–1941), Jhunjhunu (2002–2006), Hanumangarh (2000–2004), and Dausa (1937–1941). Six years of consecutive droughts were experienced in Ganganagar (1934–1939) and Dungarpur (1964–1969). Persistent drought can induce severe decrease in vegetation cover and cause changes in composition of the herbage layer.

Fig. 7
figure 7

Spatial variation in maximum drought persistence in the study region

Full size image

Table 4 shows the occurrences and return periods of droughts lasting 2, 3, and 4 years, as well as the return period for the categories of drought severity from Table 2. Over the period analyzed, the number of droughts lasting 2 years ranged from three to ten in Haryana and Punjab, from two to ten in Gujarat, and from two to 12 in Rajasthan. In general, persistent drought events of two consecutive years are quite common in the northwestern parts of India. The occurrence of droughts lasting for three consecutive years is much lower, and even more so for droughts of 4 years or longer.

Table 4 Drought occurrences and return periods
Full size table

The drought occurs more frequently in northwestern India (Vittal and Kar 2010) and causes numerous socioeconomic problems in the region. The drought management should be planned keeping in view low rainfall, unpredictable monsoon, increased human and industrial activities, and degree of drought severity in the region. Depending upon the onset of monsoon, short-term measures for drought management for early, mid and late season drought should be adopted. The effect of drought can be reduced by cropping short duration legumes having low water requirement, conserving soil moisture through crop management (Das 2005), weed management, use of groundwater, utilization of wasteland by planting fodder and conservation of excess rainwater to use during lean period.

4 Conclusion

Meteorological drought events in the districts within Punjab, Haryana, Rajasthan, and Gujarat States of India were analyzed using rainfall data for the period 1901–2013. Typical record length was 112 years; the shortest was 102 years, and the longest 113.

Most of northwestern India experience droughts every 3–4 years. There are only four districts, Chittorgarh and Bharatpur in Rajasthan and Bharuch and Dangs in Gujarat where the drought frequency dropped to one in 5 years. The drought return period was shorter in the western parts of the study area than in eastern parts. The frequency of drought in western part is 3 years, which increases to between 4 and 5 years in eastern parts. Droughts lasting 2 years are widespread and not that unusual. Drought events of 3 or 4 years do occur but are more unusual. The frequency of severe droughts ranges between once every 5 years to once every 15 years. For extreme droughts, it is between one in nine and one in 56 years.

The study enhances our understanding of regional drought frequency, severity and persistence in northwestern India. It is believed that this will be very beneficial in development of effective drought mitigation plans.