Biological Systems Engineering


Date of this Version



Khan, M., F. Munoz-Arriola, R. Shaik, and P. Greer (2019). Spatial heterogeneity of temporal shifts in extreme precipitation across India. Journal of Climate Change. DOI: 10.3233/JCC190003.


Published by IOS Press


When analyzing trends and patterns of extreme precipitation, one can easily generalize the shifts caused by external climate forcings and map a single temporal shift of annual precipitation. However, the complexity of a changing environment evidences spatially distributed shifts particularly those of extreme precipitation which are essential in planning and designing enterprises and, ultimately, in managing infrastructure across scales. The goal of this study is to characterize the spatial heterogeneity of shifts in the increasing pace of extreme events over India. The study has a two-part hypothesis: (1) the number of grid cells with significant trends in annual precipitation (P), extreme precipitation (R95) and very extreme precipitation (R99) will reflect the extent of geophysically vulnerable areas subject to increasing or decreasing annual precipitation and (2) the dispersion of cells with significant shifting points (which has occurred at different historical periods) will evidence the heterogeneity of the changes in P, R95 and R99 regimes.

To test this hypothesis, we used the Mann-Kendall and Pettitt’s tests to estimate the significance of the increasing and decreasing trends and shifting points, respectively, in P, R95 and R99 from 1901-2015 for mainland India. Based on a gridded dataset of 0.25° resolution, results showed significant temporal trends for spatially averaged R95 and R99, whereas non-significant inclining temporal trend was found for P. Trend analyses applied to the precipitation gridded product of the Indian Meteorological Department revealed statistically significant trends for almost 38%, 36% and 31% of India’s territory for P, R95 and R99, respectively. Further, the magnitude of these trends proved higher for R95 (i.e., 0.42 mm year-1) compared with R99 (i.e., 0.31 mm year-1), supporting the idea of an increasing liability for flash floods. Results also showed that most of the temporal shifts in the time series of P, R95 and R99 occurred between 1941-1980, at 34%, 31% and 22% of the grids, respectively. In addition, the opposite trends before and after the inflection point were found for locations showing significant temporal shifts in R95 and R99