Keywords
Markov Chain Monte Carlo; MCMC; HEC-HMS; Hydrologic Modeling System
Location
Colorado State University
Start Date
26-6-2018 5:00 PM
End Date
26-6-2018 7:00 PM
Abstract
The Hydrologic Engineering Center Hydrologic Modeling System (HEC-HMS), developed by the United States Army Corps of Engineers (USACE), has now served for several decades as a key tool for hydrologic modeling planning analysis within the United States. The software’s popularity stems from its ease of use and widespread applicability to a variety of hydrologic conditions. The current risk-informed planning analysis approach within the USACE requires comprehensive assessments of uncertainty, including HEC-HMS model uncertainty. To meet this need, the USACE has leveraged related research from the academic community, such that forthcoming versions of HEC-HMS will now include a Markov Chain Monte Carlo (MCMC) sampler to support optimization and inference. Application of the new MCMC feature converts HEC-HMS from a deterministic to a probabilistic flow simulation tool. It is a leap forward for HEC-HMS to substantively support studies that require probabilistic results, such as risk-informed engineering analyses. However, several opportunities exist to improve upon the initial MCMC sampler implementation in HEC-HMS, including treatment of informative priors, the consideration of additional likelihood functions and formulations, and also parameterization schemes among others. In addition, the sampler could be blended with other algorithms in attempts to improve efficiency and/or reliability. Moreover, the sampler implementation could be leveraged to quantify model structural uncertainty. We will profile one or more of these identified opportunities to help inform future HEC-HMS software development.
HEC-HMS Markov Chain Monte Carlo Implementation: Existing Capabilities and Future Opportunities
Colorado State University
The Hydrologic Engineering Center Hydrologic Modeling System (HEC-HMS), developed by the United States Army Corps of Engineers (USACE), has now served for several decades as a key tool for hydrologic modeling planning analysis within the United States. The software’s popularity stems from its ease of use and widespread applicability to a variety of hydrologic conditions. The current risk-informed planning analysis approach within the USACE requires comprehensive assessments of uncertainty, including HEC-HMS model uncertainty. To meet this need, the USACE has leveraged related research from the academic community, such that forthcoming versions of HEC-HMS will now include a Markov Chain Monte Carlo (MCMC) sampler to support optimization and inference. Application of the new MCMC feature converts HEC-HMS from a deterministic to a probabilistic flow simulation tool. It is a leap forward for HEC-HMS to substantively support studies that require probabilistic results, such as risk-informed engineering analyses. However, several opportunities exist to improve upon the initial MCMC sampler implementation in HEC-HMS, including treatment of informative priors, the consideration of additional likelihood functions and formulations, and also parameterization schemes among others. In addition, the sampler could be blended with other algorithms in attempts to improve efficiency and/or reliability. Moreover, the sampler implementation could be leveraged to quantify model structural uncertainty. We will profile one or more of these identified opportunities to help inform future HEC-HMS software development.
Stream and Session
Stream F
F3: Modelling and Decision Making Under Uncertainty