Impact of climate change on design rainfall

Abstract

Quantification of rainfall is needed for planning, designing and operation of water engineering projects such as bridges, culverts, flood control levees, open channels, roof gutters, dykes and dams. The design rainfall in the form of intensity frequency duration (IFD) data is widely used in practice. IFD data is generally derived by applying a regional frequency analysis technique to a rainfall data set consisting of a large number of stations in the region. In Australia, new IFD curves have been developed in 2013 as a part of Australian Rainfall and Runoff (ARR) by the Australian Bureau of Meteorology (BOM). The BOM 2013 IFD data were derived without considering the impacts of climate change. This research focuses on the assessment of the impacts of climate change and variability on design rainfall using data from New South Wales (NSW), Australia. A total of 60 pluviograph stations were used from NSW in the analysis of trends in the extreme rainfall events. A FORTRAN program was developed to extract annual maximum (AM) rainfall events of six sub-hourly durations (6, 12, 18, 24, 30 and 48-minute), six sub-daily durations (1, 2, 3, 6, 8 and 12-hour), and three daily durations (1, 2 and 3-day) from each of the selected pluviograph stations. Mann-Kendall (MK) and Spearman's Rho (SR) tests were applied to assess trends at local stations. For regional trend analysis, the regional MK test was employed. The impacts of climatic variability modes (SAM, SOI and PDO) on the observed trends in the AM and seasonal maximum rainfall events were investigated. For assessing changes in daily rainfall, a total of 200 daily rainfall stations were selected from NSW. The MK test was applied to identify trends in the selected rainfall indices, while the Pettitt change point test was employed to determine the direction and timing of a change point. Van Bell and Hughes homogeneity test was applied to examine homogeneity of the observed trends. Using data from ten pluviography stations in NSW and adopting a non-stationary approach, the IFD curves generated by the two most commonly adopted probability distributions, Generalized Extreme Value (GEV) and Log Pearson Type 3 (LP3) distributions were compared. Three goodness-of-fit tests (i.e. Kolmogrove-Smirnov, Anderson-Darling and Chi-Square tests) were adopted to assess the goodness-of-fit of the GEV and LP3 distributions. Empirical and polynomial regression methods in smoothing the IFD curves were also compared. The latest IFD curves in Australia as a part of the new Australian Rainfall and Runoff (ARR) were also compared with the at-site IFD curves (derived by stationary approach) to examine the expected degree of variation between the at-site and regional IFD curves. It has been found that when the MK and SR tests are applied at individual stations, the number of statistically significant positive trends in the annual maximum (AM) rainfall intensity data is greater than the statistically significant negative trends, especially in the case of rainfall intensity data of shorter durations. The regional MK test results show that there is no significant positive or negative trend in the AM rainfall intensity data when NSW State is considered as a single region. The number of stations exhibiting statistically significant trends in rainfall intensity is decreased when the impact of climate indices (SAM, SOI and PDO) is accounted for through the use of partial MK test, suggesting that much of the observed trends in AM rainfall intensity and seasonal maximum rainfall data are associated with these climate indices. This indicates that variability described by these climate indices is able to explain a significant proportion of the observed trends in the AM rainfall intensity data in NSW State. It has been found that for the annual total rainfall (ATR) and annual maximum daily rainfall (AMDR), a negative trend dominates NSW rainfall regime, while for annual total number of rainy days (ATRD), there is an overall positive trend. For ATR and AMDR, stations showing negative trends are concentrated to south-eastern NSW. However, for ATRD, north-eastern NSW is dominated by a positive trend, and south-eastern and central NSW are dominated by a negative trend. Based on the Pettitt test it is found that AMDR data in NSW is dominated by a negative shift. Furthermore, Van Belle and Hughes method shows that NSW is dominated by non-homogeneous trends in monthly maximum daily rainfall data. Based on the goodness-of-fit tests, it has been found that both the GEV and LP3 distributions fit the AM rainfall data (at 1% significance level) at the selected NSW stations. The developed IFD curves based on the second degree polynomial represent better fitting than the empirical method. The ARR87 and ARR13 IFD curves are generally higher than the at-site IFD curves derived in this study. The median difference between the at-site and regional ARR recommended IFD curves are in the range of 13-19%. The comparison of the new IFD curves based on the stationary and non-stationary approaches with ARR IFD curves (ARR87 and ARR13) illustrates that there is a better match between the ARR IFD curves and the new stationary IFD curves compared to the non-stationary IFD curves. Both SAM and SOI climate indices produce nearly similar effects on non-stationary IFD curves.

Date of Award	2018
Original language	English