Characterization of Extreme precipitation over India

This study characterizes gridded precipitation data over India in terms of presence of potential temporal trends and their upper tail properties. Daily gridded precipitation data having resolution of 0.25° prepared by Indian Meteorological Data (IMD) for a record period of 110 years (1901–2010) over entire India is used for the analysis. The objectives of this study are (i) to assess presence of potential trends in annual maximum daily precipitation series by using a variety of non-parametric methods and (ii) to investigate the upper tail behavior of daily precipitation series. Detailed trend detection analysis is carried out to find abrupt change/step or monotonic trend using a variety of non-parametric and graphical methods. Detection of abrupt change/step in the data is accomplished through Modified Pettitt’s test whereas the monotonic trends are examined by applying different tests, such as original and modified Mann Kendall (MK) tests, Spearman rank correlation (SRC), Block bootstrap (BBS) with MK and SRC and innovative trend analysis (ITA). Quantitative assessment of monotonic trend is performed based on Sen’s slope method. The implication of change magnitude is studied in terms of percentage change over mean.  Further, the upper tail behaviour of annual maximum daily precipitation series is tested based on the framework of generalized extreme value (GEV) theory. Subsequently, the behaviour of extremes in the precipitation data is diagnosed in terms of their frequency of occurrence by using a state-of-the-art algorithmic procedure which is a graphical method, famously known as Mean Excess Function (MEF). Finally, Multi-criteria decision – making (MCDM) techniques are used for identification of critical regions in terms of behaviour of extremes (i.e., increasing or decreasing trend, change magnitude, upper tail properties) over India.