Rainfall dynamics observed over India during last century (1901–2020) using innovative trend methodology

Climate variability and climate change have changed rainfall dynamics all throughout the world (Thornton et al. 2014; Dilling et al. 2015). Determining the presence and magnitude of climate change is one of the most difficult climatological challenges. Scientists from all over the world have studied climate change and endeavoured to comprehend the trajectory of numerous climate indices (Ferreira et al. 2021; Teixeira et al. 2021). Exploring current trends in climate conditions using long-term meteorological data is critical for climate change research. A significant portion of identified trends have concentrated on temperature and precipitation indices, according to historical records of meteorological stations. Droughts and diminishing rainfall are likely to worsen as a result of the rising global average temperature over the last century (Dai et al. 2018). Extreme climatic events such as floods, droughts, heat, and cold waves have severe consequences on human and animal health, the environment, and the economy (Mall et al. 2011). One of the main implications of climate change is variation in rainfall quantities and distribution (Trenberth 2011), which demands immediate attention and rigorous research. Because rainfall is such an essential part of the water cycle, it's crucial to investigate noteworthy climate changes, especially fluctuations in rainfall intensity and distribution patterns, in different climatic, hydrological, meteorological, industrial, and agricultural studies around the world for sustainable planning and management of water resources, which necessitates a thorough understanding of long-term rainfall dynamics. The variability of the seasonal rainfall over India has been examined by many researchers in recent decades (Pant & Rupa Kumar 1997; Kripalani & Kulkarni 2001; Guhathakurta et al. 2014; Maurya & Singh 2016; Sahany et al. 2018; Fukushima et al. 2019). Several investigations have also been conducted in different geographical areas of the globe by many researchers to discuss the rainfall dynamics (Marumbwa et al. 2019; Sa'adi et al. 2019; Wang et al. 2020; Jonah et al. 2021). Numerous studies have examined the spatio-temporal dynamics of rainfall and its magnitude in hydrological and meteorological time series datasets at regional as well as country levels concerning climate change and extreme rainfall events in India (Kumar et al. 2010; Joshi et al. 2020; Malik & Kumar 2020; Sahoo et al. 2020). Patra et al. (2012) found a long-term non-significant reduction in monsoonal and annual rainfall, but an increase in post-monsoonal rainfall for Odisha state. Krishnakumar et al. (2009) analyzed Kerala's long-term rainfall data and reported a significant reduction in southwest monsoonal season rainfall and an increase in post-monsoonal season rainfall. Goswami et al. (2006) recorded a substantial rise in the magnitude and frequency of heavy monsoonal rainfall in Central India. Findings of Rajeevan et al. (2008) were consistent with those of Goswami et al. (2006) as they recorded a decline in moderate rainfall events, with significant variability in frequencies of severe rainfall events at inter-decadal and inter-annual time scale, that lead to massive risks of heavy floods in central India. Vittal et al. (2013) discovered a rise in spatially aggregated extreme rainfall events over India in the second half of the twentieth century; however, substantial variations in the pattern of extreme events were also observed during the pre- and post-1950 eras.

Rainfall affects agricultural production by influencing soil moisture and its availability to the crops, and influencing soil health by inducing soil erosion, land degradation, desertification, along with the power generation and industrial output of the region, thereby affecting the overall economy of the nation. Rainfall studies are of paramount importance, as they help to foresee the challenges which affect natural endowments and economic activities of that region. The temporal trends in rainfall are a climate change proxy. The spatio-temporal trends of rainfall must be identified and adopted as signs of climate change. Floods are exacerbated by heavy rainfall, while droughts are due to inadequate rainfall, resulting in decreased agricultural production. Several numerical modelling studies suggested that the upsurge in the frequency of extreme events with the inter-annual rainfall variability is extensively associated with the rising concentration of CO₂. Trend analysis of historical rainfall data allows policymakers to gain insight into local rainfall characteristics at the regional level, assisting them in developing effective hydrological strategies to combat drought and reduce flood risk through proper water resources management. The economy, as well as food security of nations like India, are reliant on the amount and distribution of rainfall. So, the trend analysis of long-term rainfall is essential for the sustainable utilization and management of available water resources (Pingale et al. 2014; Datta & Das 2019). To do so, we used the innovative trend analysis (ITA) method, which is a new and dependable approach of trend detection that has been successfully applied to compute trends in climatic variables such as rainfall (Ay & Kisi 2015; Ahmad et al. 2018; Malik et al. 2019; Chauhan et al. 2022a, 2022b), streamflow (Diop et al. 2018; Malik et al. 2020a, 2020b), temperature (Alemu & Dioha 2020), drought variables (Mahajan & Dodamani 2015; Shiru et al. 2018), evaporation (Goroshi et al. 2017; Ghalami et al. 2021), and water quality parameters (Abaurrea et al. 2011; Al-Taani 2014) all over the world. However, the ITA approach has only been utilised in a few prior studies to find trends in a century-long time-series data of rainfall in India.

The rationale for using MSDs as an administrative region in this study is to acquire the spatio-temporal patterns of seasonal rainfall as in many parts of India; insufficient and anomalous rainfall is one of the major constraints to agricultural and other socio-economic activities. This research is needed to gain a better knowledge of rainfall dynamics, which will aid in assessing the altered precipitation pattern and trends across the study area, as well as detecting hotspots where the frequency of above and below normal rainfall categories is increasing. This could aid policymakers in locating probable drought and flood-prone areas at the micro-level, allowing for more effective resources management and rapid socio-economic decisions throughout the state. Furthermore, the MSDs level rainfall datasets utilised in this analysis span more than a century, which is a significant improvement over previous studies in India. In light of the abovementioned discussion, the aim of the presented work was to analyze the distribution pattern and trends of seasonal rainfall data of 120 years (1901–2020) for all 34 mainland MSDs of India. This research paper is divided into five sections. The data used and the study area are described in Section 2. Sections 3 and 4 deal with methodology, results, and discussion, respectively. Section 5 summarises the conclusions drawn from the study.

The normal seasonal rainfall for each MSD was calculated as mean rainfall (MR) by accounting for the average rainfall of the entire study period from 1901 to 2020. Likewise, MR for each season in all the MSDs was computed for QDT1, QDT2, and QDT3.

IMD have categorized seasonal rainfall over India into five classes depending upon the deviation from normal rainfall (Kothawale & Munot 1998), which are (i) large excess (+60% and above), (ii) excess (+20 to +59%), (iii) normal (−19 to +19%), (iv) deficient (−59 to −20%), and (v) large deficient (−60% and below). This categorization was done for the entire study period of 120 years to depict the spatio-temporal patterns of seasonal rainfall over India.

According to India Meteorological Department (2022), a rainy day has been defined as a day with rainfall of at least 2.5 mm received at any station. The average, maximum and minimum number of rainy days in all MSDs was calculated over the entire study period of 120 years during different seasons. To study the spatio-temporal changes in the amount of rainfall downpoured in a single day, rainfall events based on the daily rainfall intensity (RI) during monsoon seasons were classified as RI10 (2.5–10.0 mm), RI20 (10.1–20.0 mm), RI30 (20.1–30.0 mm), RI40 (30.1–40.0 mm), RI50 (40.1–50.0 mm) and RI50+ (50.1 mm and above). Deviation in the intensity of rainfall events (DRI) throughout the monsoon season at the MSDs level in different QDTs was also calculated for investigating spatio-temporal variation in different RI categories over a 120-year period.

The graphical non-parametric ITA method (Şen 2012) was executed to detect the trends in rainfall time series. The ITA is capable of detecting monotonic and sub-trends in time series and also has the capability to detect the different types of trends in different time series’ periods by scrutiny of the ITA graphical figures (Pour et al. 2020). In ITA, assumptions like serial normality, autocorrelation, outliers, and data length are absent. The rainfall time series dataset was bifurcated into two equal portions from starting to end date, then both divided sub-series were organised in increasing order to conduct ITA. The first half of the series was placed on the X-axis of the Cartesian coordinate system, while the second half was placed on the Y-axis. The data on the 45° line (1:1) indicate that there is no trend in the time series. Data above the 45° line represents a rising trend, while data below the 45° line represents a declining trend. The ITA slope (ITAS) test was proposed by Şen (2017). Positive and negative ITAS values show increasing and decreasing trends in time series, respectively. In this study, the null hypothesis (no trend in the rainfall time series) was compared to the alternate hypothesis (there is a trend in the rainfall time series) at two distinct levels of significance (α), i.e., α=5% and α=1%.

Descriptive statistical parameters of rainfall include mean rainfall (MR), standard deviation (SD), skewness (SK), kurtosis (KU), and maximum seasonal rainfall (MSR) of different seasons for 34 inland MSDs of India throughout the study period of 120 years are briefed in Table 1. Mainland India, on average, received 3.5%, 11.0%, 75.3%, 10.2% rainfall during winter, pre-monsoon, monsoon, and post-monsoon seasons, respectively. Among different MSDs, the highest MR of 173.1 mm, 668.3 mm, 2,910.8 mm, and 490.8 mm was observed in Himachal Pradesh, Arunachal Pradesh, Konkan & Goa, and Kerala, whereas the lowest MR of 2.0 mm, 7.2 mm, 247.0 mm, and 9.2 mm was observed at Konkan & Goa, Saurashtra & Kachh, Jammu & Kashmir, and West Rajasthan during winter, pre-monsoon, monsoon and post-monsoon seasons, respectively. Among different MSDs, the maximum value of SD in seasonal rainfall of 79.4 mm, 202.9 mm, 498.6 mm, and 141.6 mm was observed at Himachal Pradesh, Arunachal Pradesh, Konkan & Goa, and Tamil Nadu & Puducherry, while the minimum value of SD of 4.2 mm, 14.3 mm, 89.2 mm and 13.3 mm was observed at Saurashtra & Kachh, West Madhya Pradesh, Tamil Nadu & Puducherry, and West Rajasthan during winter, pre-monsoon, monsoon and post-monsoon seasons, individually. All MSDs have indicated positively skewed rainfall during winter, pre-monsoon, and post-monsoon seasons, whereas slightly negative skewness was observed in West Uttar Pradesh, East Madhya Pradesh, and Konkan & Goa during monsoon season.

During the pre-monsoon season, QDT3 had the highest MR in the country, followed by QDT2 and QDT1, respectively. Arunachal Pradesh, Assam & Meghalaya, Manipur, Mizoram, Nagaland, & Tripura, and Sub-Himalayan West Bengal & Sikkim witnessed the highest MR, while the lowest MR was observed at Saurashtra & Kachh, Gujarat region, West Madhya Pradesh, and West Rajasthan in the pre-monsoon season. The formation of semi-permanent heat lows in northwest Indian and adjoining Pakistan during the summer months, driven by deep convective activity and thunderstorms results in pre-monsoon rainfall (Sadhukhan et al. 2000; Sathiyamoorthy et al. 2010; Sinha et al. 2019).

During the summer monsoon season in India, QDT2 received the highest MR, followed by QDT3 and QDT1, respectively. Among MSDs, Konkan & Goa, Coastal Karnataka, Sub-Himalayan West Bengal, Arunachal Pradesh, Kerala, and Assam & Meghalaya observed the highest MR, however, Jammu & Kashmir and West Rajasthan received the lowest MR in the summer monsoon season. This can be attributed to the increasing distance of the inland MSDs from the source of moisture. The Western Ghats and northeast India are favoured location of the Tropical Convergence Zone (TCZ), with mean maximum precipitation during the monsoon season (Sikka & Gadgil 1980). It is expected that the convection over the equatorial Indian Ocean is critical for the monsoon because of the contribution by the northward propagations of tropical convergence zones that occurred in this district to the monsoon rainfall over India (Sikka & Gadgil 1980; Kumar et al. 1992; Gadgil 2003). Nevertheless, convection over the equatorial Indian Ocean can likewise be unfavourable due to the competition between the continental and oceanic TCZ (Gadgil 2003). The primary reason for high precipitation over the mountainous regions could be attributed to the strong orographic convection, which undergoes a diurnal cycle in which these mesoscale mountains play an important role (Xie et al. 2006).

Similarly, QDT2 had witnessed the highest MR in India during the post-monsoon season, followed by QDT1 and QDT3, respectively. Among MSDs, Kerala, Tamil Nadu & Puducherry, Coastal Andhra Pradesh, and Coastal Karnataka observed the highest MR, however, West Rajasthan, Saurashtra & Kachh, East Rajasthan, Haryana, Chandigarh & Delhi, Gujarat region, and Punjab received the lowest MR in the post-monsoon season. The major rainfall season over south peninsular India is the post-monsoon season (Dhar & Rakhecha 1983; Prasanna & Yasunari 2008), which is beneficial for agricultural production in this district (Kumar et al. 2007). Approximately 50% of the rainfall is annually received by the southeastern tip of the Indian peninsula in the post-monsoon season (Prasanna & Yasunari 2008). Cyclones developed during the post-monsoon season are another peculiar climatic feature over the southern peninsula due to the influence of the Bay of Bengal (Singh et al. 2001; Krishnakumar et al. 2009).

A negative PDR observed during QDT3 in the winter season, particularly in the MSDs lying in northwest India is in conformity with the studies of Pant et al. (1999), Kumar & Jain (2010), and Kumar et al. (2015), showing decreasing trends in winter rainfall over various parts of India which may be due to the significant decreasing frequencies of western disturbances (WDs) over the region. A positive PDR observed during QDT3 in the pre-monsoon season in the MSDs lying in northwest India might be due to extreme heat lows and resultant rainfall (Das et al. 2002; Chandrasekar 2010; Kumar et al. 2010). A negative PDR observed in the MSDs lying in Indo-Gangetic plains during QDT3 in the monsoon season is in confirmation by the results of Malik & Kumar (2020). Results of Kumar & Jain (2010) and Patra et al. (2012) are in conformity with the PDR observed during the post-monsoon season in QDT3. Krishnan et al. (2020) also reported a 6% decline in summer monsoon rainfall over India from 1951 to 2015 which was more prominent over the Indo-Gangetic plains and the Western Ghats. Kulkarni et al. (2017) reported that the rainfall has been reduced by 1–5 mm/day in QDT3 as compared to the QDT1 and QDT2 over Central India, Kerala, and extreme north-eastern India, whereas increased in the Jammu & Kashmir region and some parts of Western India. This can be attributed to climate change.

The ITA method was applied to detect the trends in time series of rainfall in different QDTs as well as CLM120. Traditional studies which usually consider the whole time period give an outlook of overall changes acquainted in any region, but this type of trend analysis is helpful in detecting the significant changes encountered during the specific time period, which can direct the researchers and policymakers in identifying the probable cause of these changes at that particular point of time, thus giving them a narrow study window than considering a broad temporal scale.