Data Used and Methodology
India Meteorological Department (IMD) has recently generated new monthly rainfall data for more than 640 districts (Guhathakurta et al. 2011a, b) from the station rain gauge data. This rainfall time series was prepared using 5894 rain gauge stations spread over the country. However, there is a variation of total number of stations on every year. On an average, more than 3100 stations are available during the period of analysis. The average number of stations during the period 1951-2010 is around 3800. In addition, the IITM all India rainfall time series (http://www. tropmet.res.in) for the same period are also used in the present analysis. This data set is based on fixed network of 306 stations spread over the country (Parthasarathy et al. 1993). In this time series, the hilly regions of the country are not considered for the analysis.
Existence of monotonic trend in the rainfall time series has been examined using the least square linear fit, and its significance is tested using the Student’s t-distribution. In addition, we have also used the well-known nonparametric Mann- Kendal trend test to detect the significance trend in the rainfall time series. The statistics used here for change point detection is a nonparametric test, developed by Pettitt (1979). This test is very useful for evaluating the presence of abrupt or sudden changes in climate records (Smadi and Zghoul 2006; Sneyers 1990; Tarhule and Woo 1998; Wijngaard et al. 2003) than the other tests used by Guhathakurta and Srivastava (2004), Leith (1978), Yamamoto et al. (1985).