Statistical analysis of climate change over Hanumangarh district

In recent years, the earth's climate change is a major issue because it is not only increasing the mean temperature of a particular region; it also disturbs the occurrence and severity of various hydrological events such as floods and drought (Tirkey et al. 2021). An increase in greenhouse gases (GHGs), owing to anthropological activities, has exacerbated the earth's climate system and led to significant climate change over the world (Solomon et al. 2007). Due to global system variation, climatic variables are continuously changing and may have a notable impact on regional water resources (Nistor et al. 2020; Azamathulla 2013). Currently, the rising climate is resulting in higher mean temperatures and less mean precipitation (Chattopadhyay & Hulme 1997; Parry et al. 2004). Climate change hurts human beings, regional agriculture, water resources, and food security (Goyal & Surampalli 2018). As per Shaver et al. (2000), due to climate change earth's weather is changed. As a result, the regional climate is also changed, particularly in terms of temperature and precipitation (Shan et al. 2015).

Climate change also affects the long-term trend of precipitation which has led to various hazards, i.e., flood, drought, etc. (Surendran et al. 2017; Azamathulla et al. 2011). Precipitation is a vital meteorological variable because its intensity and amount can affect the frequency of a drought or flood (Shaikh et al. 2022). In this context, arid or semi-arid regions are given higher importance because precipitation occurrences are very irregular, insufficient, and unpredictable (Das et al. 2020). Groundwater and surface waters could be adversely affected by low precipitation and high temperatures (Azamathulla & Jarrett 2013; Azamathulla et al. 2008). According to Aher & Yadav (2021), precipitation is a crucial element of the hydrological cycle, and variations in its intensity across time can affect the agriculture sector. Dhar et al. (1979) noted that there is a significant fluctuation in precipitation distribution in India from one region to another, from time to time, as well as year to year in the same location.

Precipitation data analysis provides valuable information that can be utilized to enhance water management practices, protect the environment, schedule agricultural production, or effects on a region's economic development (Jain et al. 2013). In India and East Asia, extremely wet or dry events throughout the monsoon season have increased (Nistor et al. 2020). Therefore, the assessment of precipitation trends is crucial for better planning, development, and management of water resources. The procedure of analyzing previous and present trends to forecast future trends is known as trend analysis (Jain et al. 2013). Precipitation trend analysis helps in a clearer understanding of issues related to floods, droughts, and the availability of water for future climate scenarios. Precipitation trend analysis assists in the sustainable management of water resources. Hence, to take necessary precautions to prevent future drought and flood-like conditions, it is crucial to understand the precipitation characteristics, the magnitude of change, the nature of the trend, and variations in monthly, seasonal, and annual precipitation. Additionally, it is significant from a socioeconomic perspective (Aher & Yadav 2021).

Various researchers from different disciplines have studied the effects of precipitation changes (Umar et al. 2022; Wang et al. 2020; Mehta et al. 2022). Numerous scientists throughout the world have studied and examined the precipitation trends in recent years (Thomas & Prasannakumar 2016; Caloiero 2020; Aher & Yadav 2021; Bouklikha et al. 2021; Mehta & Yadav 2021a, 2021b, 2022a, 2022b; Seenu & Jayakumar 2021; Pastagia & Mehta 2022). Thomas & Prasannakumar (2016), Meshram et al. (2018), Das et al. (2020), Aher & Yadav (2021), Mehta & Yadav (2021a, 2021b, 2022a, 2022b), and Halder et al. (2022), Zakwan & Ara (2019), and Perera et al. (2020) used various statistical methods to detect trends in precipitation time series. Broadly, the MK test (Mann 1945) is adopted for trend detection along with the SS estimator (Sen 1968). Kumar et al. (2010) assessed the long-term trends of precipitation in India by applying statistical methods (MK test and SS estimator). Yildirim & Rahman (2022) applied statistical methods (MK test and SS estimator) to detect precipitation trends over 45 stations located in Australia between Victoria and New South Wales states. Jain & Kumar (2012) also carried out trend analysis of precipitation and temperature using statistical methods (MK test and SS estimator) across India. However, the MK test required that the precipitation time series should be free from the effects of autocorrelation.

To avoid these problems in precipitation time series, Şen (2012) proposed a new graphical (ITA method). Using the ITA method, trend analysis of precipitation is carried out by various researchers (Caloiero et al. 2018; Caloiero 2020; Güçlü et al. 2020; Marak et al. 2020; Aher & Yadav 2021; Bouklikha et al. 2021; Pastagia & Mehta 2022; Zakwan 2021; Zakwan & Niazkar 2022). Pastagia & Mehta (2022) used the graphical (ITA method) to find seasonal and annual trends of precipitation over the Rajsamand district of Rajasthan and revealed that the ITA method is simple, effective, as well as very useful for finding trends in precipitation time series. Zakwan et al. (2022) employed the MK test, the ITA method, and the Revised Innovative Trend Analysis (RITA) method to identify the trend in discharge and sediment load at two gauging sites of the lower Drava River Basin and noticed that the ITA method provided greater visualization of the trend of the different magnitude of discharges and sediment load as compared to the MK test. Achite et al. (2022) used and compared the results of the statistical method (Theil-Sen (TS) estimator and MK test) with the results of the graphical method (ITA) and found that the results of the statistical method are slightly different from the results of the graphical method. Zena Besha et al. (2022) adopted the MK test and ITA method to find long-term trends of precipitation. They also compared the results of both methods and observed that the ITA method allows some advantages as compared to the MK test. Sonali & Kumar (2013) used 11 methods to estimate changes in temperature in India and found that, in comparison to results obtained from other methods, the result of the ITA method matched the trends well.

Thus, the objective of the research is given as follows:

• To determine the long-term trends of precipitation for monthly, annual, and seasonal times series using statistical methods (MK test) and graphical methods (ITA).

• To estimate the magnitude of the trend using the SS estimator test.

• To compare precipitation trends between statistical and graphical methods.

The comparative trend analysis of precipitation results attained from the statistical method with the results of the graphical method is an innovation in this research work.

In this research, the Hanumangarh district of North-East Rajasthan is selected as the study area. The district is situated among north latitudes of 28°46′25.0″ to 29°57′26.90″ and east longitudes of 73°47′41.74″ to 75°31′58.70″. The district's geographical area is 9,929.3 km² which is around 2.90% of the state. According to the census 2011, the district's population is 1,774,692. However, the district's population density is 184 persons/km². It was split from Ganganagar district on 12 July 1994, and before that it was one of the Tehsil of Sri Ganganagar districts.

For effective trend analysis, adequate and reliable periods of precipitation data need to be considered. The precipitation record of a shorter period would not give reliable results. The longer the period of record, the trend analysis will give better, more reliable, and more accurate results. According to the Indian Meteorological Department (IMD), India is divided into 36 sub-divisions. Precipitation data for the Hanumangarh district (4 rain gauge stations) was collected through IWRIS-Indian Water Resource Information (indiawris.gov.in). The mean monthly data of precipitation is collected from 1901 to 2022 (122 years).

For the precipitation trend analysis, four seasons were taken into consideration according to IMD. i.e., winter from January to February, pre-monsoon from March to May, SW monsoon from June to September, and post-monsoon from October to December. Analysis of monthly, seasonal, and annual precipitation time series were carried out for the years 1901–2022 (122 years).

The data are considered serially independent when r₁ is concentrated within the confidence interval; otherwise, a pre-whitening approach is used to eliminate autocorrelation when there is a significant correlation (Yue et al. 2002).

In this research, the MK test is adopted to determine trends over precipitation time series values at a 5% significant level. It is one of the consistently used statistical (non-parametric rank-based) tests (Mann 1945). It is usually applied to discover trends in the precipitation time series, whether it is linear or not (Hamed 2008). XLSTAT 2022 was used to conduct the MK test.

To determine whether a significant trend is present or not, the predicted Z_MK value is compared to a typical normal distribution using two-tailed significant levels. The null hypothesis (H₀) is accepted at a 5% significant level. There is an increasing or positive trend in the data, if the value of Z_MK is more than zero (Z_MK > 0). Similarly, when the Z_MK value is less than zero (Z_MK < 0), it shows a decreasing or negative trend exists in the data. However, if Z_MK is equal to zero (Z_MK = 0), the trend is interpreted as no trend.

The ITA method (graphical method), used to find trends in hydro-meteorological variables, was developed by Şen (2012). The main benefit of the ITA method compared to other trend analysis methods is that it is free from all assumptions like autocorrelation, normality, and number of observations. To assess trends in the precipitation data, firstly, the data of a precipitation time series are separated into two equivalent parts and then they are independently ranked in ascending (or descending) order (Wu & Qian 2017). Further, by plotting the 1:1 (45°) axis of no trend, the diagram is divided into two different triangles. The area above the 1:1 line is called an upper triangle and the area below the 1:1 line is called a lower triangle. When the data are found in the upper triangle portion, they reveal an increasing trend (Şen 2012). Whereas it shows a decreasing trend when the data are found in a lower triangle (Aher & Yadav 2021). The scatter diagram is split into three parts, i.e., low, medium, and high, to make a detailed interpretation (Pastagia & Mehta 2022).

In this research, the monthly, seasonal, and annual trends of precipitation are calculated using statistical methods (MK test and SS estimator). The graphical method (ITA) is also applied for annual and seasonal precipitation for the period 1901–2022 (122 years). The statistical method (MK test and SS estimator) results are compared to those obtained from the graphical (ITA method).

The seasonal and annual precipitation time series of the Hanumangarh district were checked by applying a lag-1 autocorrelation coefficient (r₁) at a 5% significance level. By performing the test, −0.06981, 0.201715, 0.147105, 0.029664, and 0.14013 values of autocorrelation coefficient were obtained for the winter, pre-monsoon, SW monsoon, post-monsoon, and annual season, respectively. The results revealed that neither seasonal nor annual precipitation series showed any significant autocorrelation since their value was found within the allowable limit (−1.17818 < r₁ > 1.17818). Therefore, the MK test is applied directly to the precipitation time series without any pre-whitening.

In this research, the statistical and graphical methods are used to assess the precipitation trend of different seasons. In the statistical method, the MK test and SS estimator were used. While for the graphical method, the ITA method is used. Table 3 represents the comparison of results between these three methods. During the annual precipitation time series, increasing trends are detected equally for the MK test, SS estimator, as well as ITA method. In the winter season, using the MK test and SS estimator method, a decreasing trend is detected. The ITA method also detected a decreasing trend during the winter season. Similarly, for the post-monsoon season, a decreasing trend is observed using all three (MK test, SS estimator, and ITA) methods.

During the pre-monsoon season, the increasing trend is noticed for all seasons under the MK test and SS estimator. Correspondingly, an increasing trend is observed by applying the ITA method. For the SW monsoon season, the MK test and SS estimator revealed a decreasing trend, which is highlighted in bold in Table 3. Meanwhile, the ITA method represents an increasing trend for the SW monsoon season. So, from the above results, it is concluded that the graphical method (ITA) showed a tendency to detect the same trends compared to the statistical method (MK test and SS estimator) without any assumptions.

In this study, statistical (MK test & SS estimator) as well as graphical (ITA) methods are adopted to find trends in monthly, seasonal, and annual precipitation time series for the period of 122 years (1901–2022). The graphical (ITA method) has been used and compared with statistical methods (MK test & SS estimator) for the Hanumangarh district of N-E Rajasthan. For the pre-monsoon and annual precipitation time series, the increasing trend is detected using both statistical (MK test & SS estimator) as well as graphical (ITA method). However, for the winter season and post-monsoon season, using statistical (MK test & SS estimator) and graphical (ITA method) decreasing trend is detected. On the other hand, the SW monsoon season represents the decreasing trend for the statistical method (MK test & SS estimator). Meanwhile, the graphical (ITA method) detected an increasing trend. During the SW monsoon, decreasing precipitation can affect the availability of water in the Hanumangarh district. It can also lead to short-term meteorological droughts, which have a severe effect on the agriculture sector. To reduce this risk, it is recommended to adopt the rainwater harvesting system and proper management of water in the future.

Except for the SW monsoon season, the results of the graphical method (ITA) agrees with the result of the statistical method (MK test & SS estimator). Also, it was found that the graphical method (ITA method), which compares two ascending (descending) ordered time series, is easier to use than the statistical method (MK test & SS estimator). Because, like the MK test, the ITA method does not take into account any prior conditions. Therefore, the graphical (ITA method) is more suitable for identifying trends in precipitation time series. This study of precipitation trend analysis and its comparison along two methods (statistical and graphical) will help water resources engineers and local people in sustainable management, planning as well as the development of water resources in the study region. The outcomes of this study will help to determine the suitable methodology for the assessment of drought in future studies. It will also help to decide the future strategies for drought mitigation.

This research received no external funding.

S.P. has carried out the work using various techniques and prepared the manuscript and D.M. has verified the analysis work and compiled the manuscript with proper corrections.

All relevant data are available from https://drive.google.com/file/d/124iVw9cfObGMKkXcvKlR5AajvttjAwyr/view?usp=sharing.

The authors declare there is no conflict.