RMC-BestFit is free and open-source software developed by the U.S. Army Corps of Engineers Risk Management Center (USACE-RMC) for Bayesian flood-frequency analysis, distribution fitting, uncertainty quantification, rating curves, bivariate and coincident frequency analysis, time-series modeling, and related hydrologic risk workflows. Version 2.0 exposes the core statistical engine as a reusable .NET model library while retaining the desktop application, UI/project layer, and REST API source in the same repository.
Note
RMC-BestFit 2.0.0 is the official public 2.0 release. Expect ongoing bug fixes, minor enhancements, and documentation improvements as the project continues to evolve.
| Component | Target | Notes |
|---|---|---|
RMC.BestFit |
.NET 10.0 | Portable model library and first public NuGet package |
RMC.BestFit.Tests |
.NET 10.0 | Public model-library unit tests run by CI |
RMC.BestFit.Api |
.NET 10.0 | REST API and MCP server source over the model library |
RMC-BestFit desktop app |
.NET 10.0 Windows | Windows 10+ WPF application; HEC-DSS still resolves through explicit local binary/native references during migration |
Install the model library from NuGet:
dotnet add package RMC.BestFit --version 2.0.0Or search for RMC.BestFit in the NuGet Package Manager. The package depends on RMC.Numerics 2.x for probability distributions, optimization, MCMC sampling, and numerical methods.
Source builds include the model library, unit tests, UI/project layer, desktop application, and API projects. Public CI initially restores, builds, and tests the portable model-library path while public packaging for the broader desktop stack and HEC-DSS dependency is finalized.
Note
Documentation for RMC-BestFit 2.0 is published with this release and will continue to expand with additional examples, verification materials, and API coverage. The Version 1.0 User's Guide and Verification Report linked below remain published references for the previous major release.
| Document | Description |
|---|---|
| Getting Started | Minimal namespaces and first model-library workflows |
| Technical Reference | Model, data, distribution, estimation, analysis, and diagnostic documentation |
| REST API + MCP Server | Headless API and MCP server over RMC.BestFit.dll |
| References | Consolidated bibliography for the public documentation |
| Version 1.0 User's Guide | Published desktop user guide for the previous major release |
| Version 1.0 Verification Report | Published verification report for the previous major release |
| Area | Path | Description |
|---|---|---|
| Model Library | src/RMC.BestFit/ |
Core statistical models, analyses, diagnostics, data frames, trend functions, and link functions |
| Model Tests | src/RMC.BestFit.Tests/ |
Public unit tests for data, estimation, distributions, analyses, diagnostics, rating curves, time series, bivariate analysis, and spatial extremes |
| UI Layer | src/RMC.BestFit.UI/ |
Project serialization and UI wrapper layer used by the desktop application |
| Desktop App | src/RMC.BestFit.App/ |
WPF application shell and RMC-BestFit desktop interface |
| API | src/RMC.BestFit.Api/ |
REST API and MCP server source for programmatic workflows |
| Documentation | docs/ |
Public technical documentation and references |
| Examples | examples/ |
Tutorial .bestfit projects, input data, spreadsheets, and walkthrough markdown |
The long-running verification project, verification datasets, and additional validation reports are intentionally excluded from the initial public source release. Those materials are being prepared for follow-on publication.
Restore, build, and test the public model-library path:
dotnet restore src/RMC.BestFit.Tests/RMC.BestFit.Tests.csproj
dotnet build src/RMC.BestFit.Tests/RMC.BestFit.Tests.csproj -c Release --no-restore
dotnet test src/RMC.BestFit.Tests/RMC.BestFit.Tests.csproj -c Release --no-buildCreate the public NuGet package locally:
dotnet pack src/RMC.BestFit/RMC.BestFit.csproj -c Release -o packages /p:Version=2.0.0| Domain | Public API and workflows |
|---|---|
| Input data | DataFrame with ExactSeries, UncertainSeries, IntervalSeries, and ThresholdSeries for exact, uncertain, interval-censored, and threshold observations |
| Estimation | MaximumLikelihood, MaximumAPosteriori, GeneralizedMethodOfMoments, and BayesianAnalysis with information criteria and MCMC diagnostics |
| Flood-frequency models | FittingAnalysis, UnivariateAnalysis, UnivariateDistribution, MixtureAnalysis, CompetingRiskAnalysis, PointProcessAnalysis, and CompositeAnalysis |
| Bulletin 17C | Bulletin17CAnalysis and Bulletin17CDistribution workflows for Log-Pearson Type III frequency analysis, penalties, and uncertainty methods |
| Bivariate and coincident frequency | BivariateDistribution, BivariateAnalysis, and CoincidentFrequencyAnalysis for copula-based joint frequency and response-frequency workflows |
| Rating curves | RatingCurve and RatingCurveAnalysis for Bayesian stage-discharge relationships with one to three segments |
| Time series | AutoRegressive, MovingAverage, ARIMA, ARIMAX, and corresponding AR/MA/ARIMA/ARIMAX analyses |
| Spatial extremes | SpatialGEV, SpatialGEVAnalysis, Gaussian copula support, and spatial correlation models |
| Diagnostics | Prior and posterior predictive checks, observation influence, prior influence, leverage diagnostics, threshold diagnostics, R-hat, ESS, WAIC, LOO-CV, DIC, AIC, BIC, and RMSE outputs |
| Automation | REST API and MCP server endpoints for time-series retrieval, input data creation, analysis execution, workflow chaining, metadata discovery, and JSON results |
The examples folder contains tutorial .bestfit projects and supporting files that can be opened in RMC-BestFit 2.0.
| Chapter | Topic |
|---|---|
| Time Series Data | Importing and downloading raw time series from USGS, GHCN, CHMN, ABOM, HEC-DSS, and manual entry |
| Input Data | Extracting block maxima, peaks-over-threshold, and USGS peak-flow samples |
| Univariate Distribution Analysis | Bayesian, Bulletin 17C, point-process, mixture, and composite frequency analyses |
| Bivariate Distribution Analysis | Copula-based bivariate fitting and coincident frequency analysis |
| Rating Curve Analysis | Bayesian piecewise power-law stage-discharge rating curves |
| Time Series Analysis | ARIMA, ARIMAX, and regression fitting with autocorrelated residuals |
- 2026 - Improving Bulletin 17C using the Generalized Method of Moments
- 2026 - Nonstationary Flood Frequency Analysis for Urban Watersheds Using Open-Source Bayesian Software: Contrasting Case Studies from Texas
- 2024 - Nonstationary Flood Frequency Analysis with RMC-BestFit
- 2023 - Moving Beyond Bulletin 17C with Bayesian Flow Frequency Analysis
- 2021 - Incorporating Regional Rainfall-Frequency into Flood Frequency using RMC-RRFT and RMC-BestFit
- 2019 - Estimating Design Floods with a Specified Return Period Using Bayesian Analysis
USACE-RMC maintains RMC-BestFit 2.0 with regular bug fixes, documentation improvements, and validation updates. Public issues should include the RMC-BestFit version, operating system, .NET SDK version when relevant, the workflow or project file involved, and enough detail to reproduce the behavior.
The repository includes a fast model-library test suite that is intended to serve both as regression coverage and as API usage examples. Longer-running verification materials are being prepared separately for public release.
Bug reports, feature requests, documentation feedback, and independent validation results are welcome. Please read CONTRIBUTING.md before opening a pull request. Review capacity is limited, so proposed code changes should start with an issue discussion.
RMC-BestFit is provided under the Zero-Clause BSD (0BSD) license.