The effects of future extreme precipitation events on stream hydrology and hydraulics of stream crossing structures

Precipitation;stream hydrology;peak flood;climate change;riverine flooding;low-water crossings

Gambill, Daniel R.

关键词: Precipitation;    stream hydrology;    peak flood;    climate change;    riverine flooding;    low-water crossings;   

Others  :  https://www.ideals.illinois.edu/bitstream/handle/2142/95330/GAMBILL-DISSERTATION-2016.pdf?sequence=1&isAllowed=y

美国|英语

来源: The Illinois Digital Environment for Access to Learning and Scholarship

【 摘 要 】

Army field training exercises are conducted to replicate real-world combat requirements and are inherently subject to the effects of the prevailing climate and weather conditions. Adapting to variability of climate is part of realistic training and extreme storm events and associated flooding risks can temporarily limit access to training lands and other training features such as water crossings. If an increase in intensity and incidence of extreme precipitation events is realized, installations associated with river systems are likely to have increased flood risk and associated prevention and mitigation costs. Low water crossings (LWCs) are especially susceptible to changes in intensity and incidence of extreme precipitation events with regards to infrastructure usability, resilience, and safety.The June 2, 2016 incident at Fort Hood, TX where five soldiers from the 3rd Battalion, 16th Field Artillery Regiment, 2nd Armored Brigade Combat Team, 1st Cavalry Division drowned when their troop carrier overturned at a flooded LWC, shows the safety risks associated with LWCs.The first objective of this study is to examine the stationary assumption of total annual precipitation, total annual wet days and wet hours, and regional frequency analysis for a range of design storms and extreme events.Both parametric and non-parametric statistical tests were used to detect trends.In addition, precipitation frequency estimates were calculated for the 0.5-, 1-, 2-, 5-, 10-, 25-, 50-, and 100-year Annual Return Interval (ARIs) and durations of 1-, 2-, 3-, 6-, 12-, and 24-hr; and 2-, 5-, and 10-days for three different time periods 1970-1989, 1980-1999, and 1990-2009. The results suggest the assumption that precipitation is stationary over time with regards to total annual rainfall and storm frequency is not valid for the three Midwest regions, Central Indiana, East Central Illinois, and Central Michigan.The second objective is to analyze projected precipitation data from four climate models with regards to total annual precipitation, total annual wet days and wet hours, and regional frequency analysis for a range of design storms and extreme events and compare findings to analyses of historic precipitation data.Four dynamically downscaled CMIP5 global climate models were bias corrected using quantile mapping.The average projected (2080-2099) design storm of the four climate models across all locations, durations, and return periods is larger than the corresponding observed design storm and usually significantly larger based on the standard deviation of the model results, especially for return periods greater than 1 year. The four models provided bias-corrected precipitation output that agreed in general trends, however, the precise levels of precipitation increases varied substantially, especially for the 2040-2059 projected timeframe.The third objective is to apply projected climate model precipitation to hydrologic models to determine projected stream flow characteristics and compare to current stream flow characteristics.The result indicate increases in peak flood events across all return periods for the projected timeframes compared to observed conditions.In Indiana and Michigan the projected (2080-2099) peak flow events are larger than the projected (2040-2059) peak flows while in Illinois the projected (2080-2099) peak flow events are larger than or equal to the former events.Analysis of the number of days projected stream flow exceeds the 0.5-, 1-, and 10-yr design flows during the 20-yr simulation showed a wide range of results between the two models representing the upper and lower bounds of maximum precipitation estimates for a majority of return periods for each projected timeframe at each location. Generally there was an increase in exceedances during the projected timeframes compared to current conditions, but on a yearly average basis the increase were minimal.The fourth objective is to route projected design flow and continuous stream flow hydrographs through hydraulic models to determine usability and sustainability of current structures and feasibility of alternative designs for projected flow regimes. The results indicate the riverine crossing structures considered for this study are projected to see an increase in the magnitude and frequency of high flow events by the end of this century. The projected (2080-2099) 10-yr event is on the order of the present 50-yr (sometimes 25-yr or 100-yr) event for many of the studied streams, suggesting possible future conditions should be considered when designing new infrastructure. Unfortunately the uncertainty inherent in the climate modeling makes it difficult to develop specific recommendations on how to revise current LWC design criteria with regards to climate change in the study regions. The continuous model simulations and projections proved to underestimate average yearly flow durations for small flow events with very frequent return periods. Special care must be taken when using and applying frequent events from dynamically downscaled climate model precipitation data.An investigation into the timing and intensity of very frequent observed and simulated precipitation events could be needed before applying similar climate model data to hydrologic and hydraulic models.Overall, they study showed that the riverine crossing structures considered for this study are projected to see an increase in the magnitude and frequency of high flow events by the end of this century. The changes in hydrologic flows were constant with changes in projected precipitation. The four climate models provided bias-corrected precipitation output that agreed in general trends but the precise levels of precipitation increases varied substantially, especially for the 2040-2059 projected timeframe.In addition, watershed specific variables, such as those found in Michigan, can add a great deal of uncertainty to modeling results.The projected increases in precipitation and subsequent changes in peak flood events are large enough that corresponding impacts on stormwater infrastructure design should be considered, however, the uncertainty of the future projections makes it difficult to develop specific design recommendations.