Analysis of uncertainty and non-stationarity in probable maximum precipitation in Brazos River basin

Abstract

Probable Maximum Precipitation (PMP) is used for designing major hydraulic structures, such as dams and reservoirs, nuclear power plants, and flood protection works. However, the estimated PMP values are associated with uncertainties that have received significant attention in recent years, partly because hydrologic extremes are projected to become more frequent, severe, and uncertain with non-stationarity and natural climate variability. This study compared four methods of estimating PMP, ranging from hydrometeorological to statistical, including up-to-date grid based and site-specific in the Brazos River basin (BRB). BRB is the largest river basin in Texas and it contains a range of climates from subtropical arid to subtropical humid. The objective of this study was to quantify the uncertainty associated with PMP estimation in terms of change in the PMP value and emphasize the necessity of including the effect of non-stationarity on PMP. The uncertainty analysis incorporates the effect of three sources of error: (1) PMP estimation method, (2) topography, and (3) non-stationarity. The contribution of each of the uncertainty sources to the 24-hour PMP estimation was quantified and found to be as 53.5% (selection of method), 31.4% (effect of non-stationarity), and 15.1% (effect of topography). The uncertainty of PMP estimation was more sensitive to the existing observation statistics and the selection of method than to the differences between climate zones. Results showed an overall significant increase in 24-hour record precipitation (+19.5 mm) and PMP (+22.3 mm) in BRB between two historical periods 1940–1976 and 1977–2013. Thus, it is concluded that extreme precipitation in BRB showed non-stationarity which affected PMP shift during the historical period.