Effects of Global Change on Extreme Precipitation and Flooding: New Approaches to IDF and Regional Flood Frequency Estimation
Dr. Dennis Lettenmaier | University of California, Los Angeles
The overall objectives of this project are to develop: (1) protocols for estimation of precipitation intensity-duration-frequency (IDF) and intensity-duration-area-frequency (IDAF) curves in a changing climate that incorporate both historical observations and predicted future manifestations of climate change and (2) approaches to incorporate ongoing and future projections of the effects of climate warming on flood frequency estimates for watersheds whose frequency distributions are affected at least in part by snowmelt and/or rain-on-snow events. The overarching science/applications question this research will address is how can engineering design criteria associated with extreme precipitation and flooding be adjusted to reflect observed and projected effects of climate change in a manner that is consistent with standards of engineering practice?
Nearly all Department of Defense (DoD) facilities are exposed to weather and climate-related risks at some level and most have substantial impervious areas that are protected by drainage systems sized based on standard engineering design criteria related to precipitation extremes. These criteria are now based on stationary statistical methods. There is, therefore, an urgent need for new protocols that will reflect the evolving effects of climate change on precipitation and flood extremes. Such protocol development is a component of this work.
The technical approach includes two elements that will address the first overall objective and a third element that addresses the second overall objective.
The first element will incorporate non-stationarity as observed in historical records of extreme precipitation into protocols for estimating IDF and IDAF relationships via incorporation of a time-varying parameter in the Generalized Extreme Value distribution. It is intended for application to projects for which the planning time horizon is relatively short (e.g., several decades), and locales where it is reasonable to design for the “now” climate.
For estimation of non-stationary IDF and IDAF relationships from historic observations, a key element will be adaptation of regional L-moments estimation, or regional (constrained) maximum likelihood-based estimation of the widely used Generalized Extreme Value (GEV) distribution, with a time-varying parameter. Using Monte Carlo methods, Monte Carlo simulation will be used to test alternate approaches to accomplishing the preceding that exploit observations at multiple sites across a region.
The second element will address design criteria for projects with planning horizons and economic life longer than a few decades or projects that involve irreversibility considerations. It will incorporate regional climate models (RCMs) into estimates of IDF and IDAF. This element will include aspects both of historical flood extremes and projections of future climate. For this element, methods will be developed that extend IDF and IDAF analysis using regional GEV estimates beyond the period of historic observations using the Weather Research and Forecast Regional Arctic System Model (WRF/RASM) regional climate model (RCM) for future extreme precipitation simulations. For these cases, statistical downscaling and bias correction methods will be applied, in conjunction with areal reduction factors, to impose a consistency constraint between estimates based on RCM and point historical observations.
The third element is specific to flood frequency estimation in watersheds with flood regimes that are currently at least partially affected by snowmelt. This project will extend previous work in the western United States to the entire portion of the continental United States that has experienced or could potentially experience snowmelt-related floods. The work will use the Variable Infiltration Capacity (VIC) hydrologic model and perturbations of historic temperatures to reflect current climate conditions as well as RCM simulations of future climate.
This project will leverage the best current generation (stationary) extreme precipitation and flood frequency estimation to incorporate time-varying parameters into the GEV probability distribution using regional parameter estimation. This work is expected to have direct benefits to risk-based flood design not only at DoD facilities, but also across the broader realm of hydrologic engineering design. The project will work with other SERDP-funded projects in the area of hydrologic extremes to produce a cross-project “Best Practices” manual. Results of the research are anticipated to be incorporated into future updates to the National Oceanic and Atmospheric Administration’s Atlas 14. Although this is the most likely way in which DoD will implement project results, the Best Practices manual will identify interim strategies by which DoD could implement project results in conjunction with current volumes of Atlas 14. (Anticipated Project Completion - 2019)