Estimation of Extreme Rainfall Thresholds and Their Variability Across Tapi Basin, India

The selection of base period for estimation of threshold for hydro-meteorological extremes is of key importance as inappropriate selection of base period may produce erroneous design estimates. As per World Meteorological Organization (WMO) technical regulations, climatological standard normal (CSN) is considered as an optimal period of 30 years for recognizing the climatic changes in the region. As Indian summer monsoon rainfall (ISMR) exhibits high multi-decadal variability, the selection of extreme rainfall threshold corresponding to CSN period rather than entire time period, for detecting long-term changes, could result in underestimation or overestimation of design estimates. In present study, the daily rainfall data of fifty-four rain gauge stations in the Tapi basin in India for period of 70 years (1944-2013) have been analyzed for estimation of extreme rainfall thresholds, i.e., 95^th (R95) and 99^th (R99) percentile threshold values. The values of R95 and R99 are derived from empirical cumulative distribution function of observed daily station rainfall during monsoon months (June – September) within the basin. Aforesaid thresholds values were estimated for 10 year moving CSN periods (i.e. 1944-1973, 1954-1983 and so on), and their variability vis-à-vis thresholds computed for entire period of 70 years (1944-2013), were assessed. Invariably, R95 and R99 values for different CSN periods exhibited minimal variability, i.e., within ±10% of the values corresponding to those for entire period. Thus, consideration of entire period (1944-2013) for estimation of extreme rainfall threshold would be more appropriate, as it would eliminate the short-term variability induced by natural climatic cycles. Also, the magnitude of rainfall equaling or exceeding the quantum of rainfall on very wet days (R95p) and extremely wet days (R99p), were estimated on annual basis for entire period (1944-2013). The trends in R95p and R99p were further analyzed using non-parametric Modified Mann-Kendall (MMK) and Sen’s slope tests. The results showed increasing trend in extremely wet days (R99p) across Upper and Lower Tapi sub-basins, while decreasing trend were predominant over Middle Tapi sub-basin. The very wet days (R95p) also exhibited similar results, as R99p, in Upper, Lower and Middle Tapi sub-basins. Thus, increasing trend in extremely wet and very wet days across Upper and Lower Tapi sub-basins would escalate the risk of flash floods in the region, and have implications on urban storm water management. The current study would provide region-wise estimates of magnitude and variability in extreme rainfall across Tapi basin which would be useful in hydrologic designs in future.