Climate change is a well debated issue in recent times and possibly one of the gravest global challenges in the present century. It has been found that the Earth's climate has been changing notably, at a fast pace since the last century, and the changes are expected to continue. Evidences of changes are prominent in the environment through the increase in global and regional temperatures and perceptible changes in the hydrologic cycles in many parts of the world including India (Goswami et al., 2006). Manifestation of such global-scale climate variations can be observed as fluctuations in the normal display of any local climatic features. Rainfall is one such important feature, understanding the trends and changes of which will help to resolve uncertainties (Singh and Sontakke, 1999) and provide the knowledge base for decision making on a broad series of local issues related with agriculture, industry, irrigation, generation of hydroelectricity, and other human activities (Singh 1998). Spatio-temporal trend analysis of rainfall has a vital role in different engineering and managerial strategies such as watershed/river basin development and management (Burguenõ et al., 2004).
Annual trend in rainfall pattern has been examined extensively by different researchers from various parts of the world. Odekunle et al., (2005); and Eltahir (1992) in Africa, Burguenõ et al., (2004) in Spain, Smith (2004) in Australia, Bidin and Chappel (2006) in Malaysia, Shen et al., (2007) in China, Ashley et al., (2003) and Haylock et al., (2006) in USA, are a few among such studies. Similarly, monsoonal rainfall in India has been explored in its general nature (Rajeevan et al., 2006, Gadgil and Joseph, 2003), its predictability (Venketesan et al.1997, Pai and Rajeevan 2006, Ratnam et al., 2007, Raju et al., 2007, Goswami and Gauda 2007), and its regional occurrences and variations (Mohapatra and Mohanty, 2007, Simon and Mohankumar 2004, Arora et al., 2006, Ghosh and Mujumdar, 2006, Singh et al., 2007, Murata et al., 2008). Recent studies by Ramesh and Goswami (2007) reported the shrinking of the Indian summer monsoon in terms of total rain days as well as in total area of rainfall.
Kerala is almost the entry point of the summer monsoon rainfall in the Indian subcontinent. The principal rainy seasons in Kerala are the southwest monsoon (June–September) and the northeast monsoon (October–November). The pre-monsoon months (March–May) are characterised by major thunderstorm activity in the state, and the winter months (December–February) are marked by low clouding and low rainfall (Ananthakrishnan et al., 1979). Several studies evaluated the monsoonal rainfall of Kerala; Parameswaran (2001), Fasullo and Webster (2003), Joseph et al., (2004), and Guhathakurtha (2005), documented general characteristics and the pattern of monsoon in Kerala. The study conducted by Simon and Mohankumar (2004) using multivariate statistics revealed the spatial variability in the occurrence of rainfall in the state.
Recent studies (Soman et al., 1998, Kumar et al., 2004 and Krishnakumar et al., 2009) reported spatial and temporal variations in Kerala's annual rainfall. Nevertheless, little is known about the local-scale rainfall trends in the state (Raj and Azeez 2010), though it is vital for an agriculture-based economy, with rainfed irrigation systems. The present study focuses on the general trend analysis of rainfall in the Bharathapuzha River basin of Kerala state.
2. Study area, data, and methodology
River Bharathapuzha (10°25′ to 11°15′N and 75°50′ to 76°55′ E) is the second longest (209 Km) among the west-flowing perennial rivers in Kerala. It originates from the Western Ghats as brooks and rivulets that confluence later to form four major tributaries namely Kalpathipuzha, Gayatrhipuzha, Thootha, and Chitturpuzha. The main river formed by these tributaries finally discharges to the Arabian Sea at Ponnani on the west coast (Figure 1). The river has a total basin area of 6186 Km2 of which 4400 Km2 falls in the State of Kerala and the rest in Tamil Nadu State of India. The river basin covers 1/9 of the total geographical area of the state that has 41 west-flowing rivers within its boundary. The flow regime of the river covers highlands (>76 m), midlands (8–76 m) and the low lands (<8 m).
The river is the lifeline water resource for a population residing in four administrative districts, namely Malappuram, Trissur, and Palakkad districts of Kerala and Coimbatore, and Thiruppur district of Tamil Nadu. There are ten irrigation projects and several sub-surface dams in the river basin catering to 493 064 ha cultivations (CWRDM, 2004 and Ravi et al., 2004). In recent years, the river basin was found going through severe dearth of water and drought conditions.
To estimate the general trend in annual rainfall, monthly rainfall data of 34 years (1968–2002) from 29 rain gauge stations located in the basin were collected from the Department of Irrigation, Government of Kerala, Thrissur, Kerala Engineering Research Institute (KERI, Peechi) and Regional Agricultural Research Station (RARS, Pattambi). The pooled data were then analysed for various fundamental statistics such as mean (Normal), standard deviation (STDEV) and Coefficient of variation (CV) (Table I). Time series of data was also examined for different temporal changes. Trend line and its significance were checked using t-test statistics. The data is categorised according to Ananthakrishnan et al., (1979) into four principal rain-giving seasons; the southwest monsoon (June–September), the northeast monsoon (October–November), the pre-monsoon months (March–May), and the winter months (December–February). Temporal changes in the seasonal and annual rainfall were also analysed by Man-Kendall rank correlation statistics (Tt), since it is the most suitable statistical test for a long period of data (Basistha et al., 2007, Krishnakumar et al., 2009). Moreover, it is a robust statistical tool for analysing non-parametric climate datasets. The value of t can be used as the basis of a significant test by comparing it with
Table I. Rainfall statistics in Bharathapuzha River basin; the average rainfall (Normal), standard deviation (STDEV), coefficient of variation (CV), and % of contribution towards annual rainfall is also given
% to annual rainfall
Where, tg is the desired probability point of the Gaussian normal distribution. We considered tg at 0.01 and 0.05 as the points for significance. The trend line fitted to the data was analysed using Student t-test to verify the results obtained from Man-Kendall statistics. In order to visualise the occurrence of events in a broad time scale, wavelet analysis was performed on the time series of rainfall data since the analysis is well known for its multi-resolution analytical capabilities and for a broad insight to the periodic occurrence of climate processes (Jianhua et al., 2009, Nicolay et al., 2008, Sonechkin and Datsenko, 2000). A series of time scale analysis was conducted by taking symmlet as the basic wavelet. We used ‘Sym 8’ as the operational wavelet function, and decomposed and reconstructed the average annual rainfall time series at three time scales, i.e. 16 years, 8 years, and 4 years. SPSS 10 Software package and MATLAB 7.5 were used for carrying out all the statistical analysis.
3. Results and discussion
3.1. Annual rainfall
The average annual rainfall in the Bharathapuzha River basin is 1828 mm with a standard deviation of 456.6 mm. Among the months, while July receives the highest (525 mm) which accounts for 29% of the total annual rainfall in the basin, January receives the lowest (3 mm). Among the four principal seasons' rainfall, the highest was during the southwest monsoon season (1318 mm with a standard deviation of 391 mm) followed by northeast and pre-monsoon seasons (Table I). Further, the historical rainfall for five years' class of annual rainfall in the basin also shows the prominence of rainfall during south-west monsoon (Table III). Mann-Kendal test shows a significant decrease in the annual rainfall in the river basin (Table II and Figure 2). Additionally, a significant periodic change in a 16 years time scale (R2 = 1) is observed, compared to 4 and 8 years (R2 = 0.08) in the wavelet analysis (Figure 3). The decrement in rainfall in the region could alter the hydrology of the basin. In the recent decades severe drought and dearth of water was reported from the basin (CWRDM 2004). The decrement in the total rainfall may be a manifestation of the global climate change, perhaps added on to by local changes. The study conducted by Krishnakumar et al., (2009) rainfall has also revealed decrement in the annual rainfall in Kerala.
Table II. Mann-Kendal statistical analysis of rainfall characteristics
Table III. Contribution (%) of the seasonal rainfall to the total annual rainfall
3.2. Seasonal and monthly rainfall
The characteristic of seasonal and monthly rainfall of the basin is studied using Mann-Kendall statistical analysis (Table II). The winter rainfall (Figure 4) does not indicate a statistically significant declining trend (Table II). This trend does not conform to the findings of Krishnakumar et al., (2009) done for the entire state wherein an increasing trend for winter rainfall was seen. The pre-monsoon rainfall (Figure 5) shows a significant decrease during the study period. It is also found that the pre-monsoon rainfall is negatively correlated with the February rain. Study conducted by Kumar et al., (2004) reported that the pre-monsoon peak of rainfall occurs about seven pentads prior to the onset of monsoon over Kerala. The variation in pre-monsoon rainfall may be a factor influencing the annual rainfall decrement in the area.
The southwest monsoon (Figure 6) also shows a significant (p = 0.01 level) negative trend. The decreasing trend in southwest monsoon rainfall over Kerala is also reported by other researchers (Kumar et al., 1992; Guhathakurta and Rajeevan, 2007). Since the southwest monsoon contributes major share of the total annual rainfall the decrement will have huge impacts on the general hydrology of the basin and the agriculture. Joseph and Xavier (1999) observed a strong decreasing trend during last 100 years in the monsoon depression frequency.
The study conducted by Soman et al., (1998) reported the decrement in northeast rainfall in the region. In the present situation although the rainfall during northeast monsoon (Figure 7) tends to decrease, its decline is not statistically significant. Rainfall in any month individually shows any significant change. However, there is a trend of increase in rainfall during the months of January, February, October, and November. On the other hand, the rest of the months show a decreasing trend. In a year, June, July, and August are the principal rain-giving months in the basin and the analysis shows a decrease in trend in the rainfall (Table II) during these months. The t-test statistics (Table IV) indicated that southwest, pre-monsoon and annual rainfall trends in Bharathapuzha basin to be significant at 0.01 levels.
Table IV. t-test significance for linear equations
The variation in the rainfall pattern of the basin has to be viewed in the context of global climate change. An increase in surface air temperature of about 1 °C over the Indian peninsula has been reported by Dash (2007). The area is highly pressure from anthropogenic sources such as deforestation (Mehar-Homji 1991), and urbanisation. In states like Kerala, large-scale landscape changes, including flattening of several hillocks, are happening at a fast pace. Studies conducted by Chattopadhyay (1985) has shown remarkable extent of land use change in the basin.
The present study revealed significant changes in annual, pre-monsoon and southwest monsoon rainfall in the study area. No significant decrement could be seen in the northeast monsoon. The decrement in the annual rainfall could have a negative impact on major water input to the basin and the hydrologic realm of the basin. These changes are expected to seriously affect the irrigation and hydroelectric projects in the basin, which will indirectly affect the agriculture productivity in the basin.
We thank the Department of Irrigation Government of Kerala, Thrissur (Engineers Anitha, Denni, and Solin), the librarian KERI Peechi, and the Director, RARS Pattambi for providing us valuable help. We are thankful to Prof C Haneefa Muhammad (College of Applied Medical Sciences, King Saud University, Riyadh) for the English corrections, and Mr Nithin, University of Kerala, Prof. KP Soman, and Dr Manikandan, Amritha university, Coimbatore, for their help and discussion on wavelet analysis.