Overview

The Optimization Trial (also referred to as Optimization Analysis) is a simulation type in HEC-HMS designed to automate the estimation of model parameters. Rather than manually adjusting parameters through trial-and-error, the Optimization Trial uses a mathematical search algorithm to iteratively adjust user-selected parameters until a quantitative measure of model performance — the Objective Function — is optimized (minimized or maximized). The result is an objective, reproducible parameter set that best fits observed data or maximizes a hydraulic outcome of interest.

An Optimization Trial is composed of a Basin Model, Meteorologic Model, Time Window, Search Method, Objective Function, and the Parameters to be estimated. Results including optimized parameter values, objective function progression, and computed-vs-observed comparisons are all viewable in the Watershed Explorer after compute.

Important note: Optimization is an aid to calibration — not a replacement for engineering judgment. Results should always be reviewed for physical reasonableness.

Use Cases

1. Single-Event Model Calibration

The most common use case. Configure the trial to minimize the difference between simulated and observed streamflow at a gage location during a historical flood event. The trial automatically adjusts parameters such as loss rates, transform coefficients, and baseflow constants to improve hydrograph fit.

2. Continuous Simulation Calibration

Apply Differential Evolution or Simplex search methods over a long-period simulation to calibrate parameters across multiple seasons or years of observed data. Scale Factors allow simultaneous adjustment of the same parameter type across all subbasins upstream of a target gage.

3. Flood Maximization / Hazard Analysis

Use the Maximization Goal to find the parameter set that produces the highest peak discharge or reservoir pool elevation. This is particularly valuable for dam safety studies, where the goal is to determine the most critical storm configuration rather than match observed data.

4. HMR52 Storm Optimization (PMP Studies)

A widely-used application for Probable Maximum Flood (PMF) estimation. Configure the trial to automatically adjust HMR52 Storm parameters — storm center location (X/Y coordinates), orientation, and area — to maximize inflow into a reservoir. This replaces the need for dozens of manual simulation runs.

5. Gridded Precipitation Optimization

Optimize storm center placement for gridded precipitation datasets to maximize peak discharge or discharge volume at a target location.

6. Multi-Location and Multi-Window Calibration (v4.14+)

Specify multiple objective target locations and/or multiple time windows within a single trial. Useful when calibrating to multi-peaked events or to observed data at more than one gage simultaneously. Sequential Searches allow upstream parameters to be locked before optimizing the next downstream location.

7. Forecast Optimization (v4.14+)

Link Optimization Trials directly to Forecast Alternatives via the new Zonal Config Optimization workflow. The trial automatically calibrates scale-factor parameters by zone against observed data from the forecast lookback period, then pushes optimized values to the Forecast Parameter Override Tables.

Best Practices and Gotchas

Choose an appropriate Search method

MethodTypeBest For
Simplex (Nelder-Mead)DeterministicOptimizing well-defined parameters with good initial guesses, much faster than Differential Evolution
Differential EvolutionStochasticOptimizing parameters with unknown ranges and uncertain initial guesses, slower than Simplex

A combined approach can work well; start with Differential Evolution to narrow in on appropriate parameter bounds and then switch to Simplex.

Choose a relevant goodness-of-fit Objective Function

  • For most general use cases, Peak-weighted RMSE is a good choice that attempts to best match all of the observed data, with extra emphasis on the higher values (i.e. the peak portions of the hydrograph).
  • If all you care about is matching the observed peak value, consider Peak-weighted variable power.

Be selective when adding Parameters

  • This is where hydrologic expertise is critical.
  • Try to limit the parameters to those that matter the most (i.e. have the most impact on computed results).
    • 3 to 6 parameters per Search is generally a reasonable amount to start with.
  • For each parameter, assign minimum and maximum limits that are reasonable to your watershed.
    • The tighter the bounds, the better
  • Optimization algorithms have no notion of hydrology. An optimization routine might give you combinations of parameter values that agree well with observed data but individually are non-sensical.

Assess the computed results, and iterate

  • It is rare to get useful results from the first Optimization Trial compute; iteration is key.
  • Adjust the Search method tolerance and iteration constraints.
    • Start with relaxed tolerances and lower Max Iteration constraints to make sure that the Optimization is trending in a reasonable direction.
    • Tighten up for subsequent runs when the included parameters and their initial guesses and limits are dialed-in.
  • Adjust based on Parameter results
    • Are the parameters converging on reasonable values that make hydrologic sense?
    • Is a parameter optimizing towards minimum or maximum values?
      • Adjust the minimum or maximum values only if it makes sense.
      • This could be a sign that a critical, sensitive parameter is missing from the Optimization Trial and should be added.
    • Is a parameter not converging?
      • This could be a sign that the Optimization Trial is reaching convergence too early.
      • It could also be a sign that the parameter does not significantly impact the computed results and can be removed from the Optimization Trial.

Documentation & Resources

📖 User's Manual

Step-by-step guidance on creating, configuring, and interpreting Optimization Trials — including search methods, objective functions, parameter setup, and results.

📐 Technical Reference Manual

Theoretical background on the optimization algorithms, objective functions, search method mathematics, and applicability/limitations.

🛠️ Tutorials & Guides

Hands-on worked examples with downloadable project files.

This document was drafted from the HEC-HMS User's Manual (HMSUM), Technical Reference Manual (HMSTRM), and Tutorials & Guides (HMSGUIDES) Confluence spaces.