RMC-BestFit v2.0.0

RMC-BestFit v2.0.0 is the official release of the next generation of BestFit.

Version 1.0 gave users a focused set of tools for Input Data, Distribution Fitting, and Univariate Analysis. Version 2.0.0 transforms BestFit into a much broader Bayesian hydrologic analysis platform, with new data workflows, new model families, modern diagnostics, expanded uncertainty analysis, and the first public RMC.BestFit .NET model-library package.

Major New Capabilities

• Time Series Elements and Download APIs

  • Create, manage, plot, and transform time series directly in BestFit.
  • Download hydrologic and precipitation time series from supported public data sources.
  • Build block-maximum and peaks-over-threshold datasets from time series workflows.

• Peaks-Over-Threshold Workflows

  • Extract independent threshold exceedance events from time series.
  • Use diagnostic plots and threshold controls to support POT model setup.
  • Move from raw time series to frequency-analysis-ready input data inside one project.

• Nonstationary Frequency Analysis

  • Fit models where distribution parameters vary with time or covariates.
  • Evaluate changing flood-frequency behavior with Bayesian uncertainty propagation.
  • Support engineering workflows where stationarity is no longer a defensible assumption.

• Mixture Models

  • Represent multiple flood-generating populations in a single analysis.
  • Estimate mixture weights and component distributions with uncertainty.
  • Support hydrologic settings where one distribution cannot represent all processes.

• Composite Analysis

  • Combine alternatives through competing-risk, mixture, and model-averaging workflows.
  • Use information criteria such as WAIC, LOO-CV, DIC, and AIC to support model comparison and weighting.
  • Carry model uncertainty into frequency estimates instead of forcing a single-model decision too early.

• Point Process Analysis

  • Fit point process / peaks-over-threshold models for extreme-value analysis.
  • Support workflows that use more information than annual maxima alone.
  • Connect threshold exceedance data directly to Bayesian frequency estimates.

• Bivariate Analysis

  • Model dependence between paired hydrologic variables using copula-based workflows.
  • Estimate joint behavior for variables that cannot be treated independently.
  • Support applications such as flow-stage, inflow-volume, and other paired-risk problems.

• Coincident Frequency Analysis

  • Evaluate response-frequency relationships driven by dependent variables.
  • Support joint exceedance and conditional frequency workflows.
  • Bring bivariate dependence into practical risk and design calculations.

• Rating Curve Analysis

  • Fit stage-discharge relationships with uncertainty.
  • Support multi-segment rating curves for more complex hydraulic controls.
  • Carry rating-curve uncertainty into downstream hydrologic assessment workflows.

• Time Series Analysis

  • Fit time series models, including AR/MA/ARIMA/ARIMAX-style workflows.
  • Include covariates and forecasting workflows where appropriate.
  • Extend BestFit beyond frequency curves into temporal modeling and prediction.

Platform and Developer Updates

• RMC.BestFit.dll is now separated from the desktop UI and available as a reusable model library.
• The desktop application, UI/project layer, and API source remain in the same repository.
• The project is modernized for .NET 10.
• The release uses published RMC.Numerics and RMC.Wpf.Framework NuGet dependencies.
• The software is released under the permissive 0BSD license.

Install the Model Library

dotnet add package RMC.BestFit --version 2.0.0

Thank You

Thank you to everyone who tested the 2.0 beta releases, opened projects, reported issues, reviewed outputs, and pushed the software through real flood-risk workflows. Version 2.0.0 is a major step forward for Bayesian hydrologic analysis in BestFit.

Desktop Application Download Verification

SHA256: 95c61f4b538636c47d5578118bebd56c56d7118462a22093359164c75c5a0bb9

Version 2.0 Beta-5

RMC-BestFit 2.0-beta.5 is a maintenance-focused beta release with more than 40 bug fixes, stability updates, and efficiency improvements across the application.

This release improves performance for large projects, strengthens validation and project-loading behavior, and includes general UI, packaging, and reliability refinements.

Version 2.0 Beta-4

We're excited to announce Version 2.0 Beta-4 of RMC-BestFit, the largest update yet. This release modernizes the platform, separates the model library from the UI for programmatic access, and introduces a substantial collection of new analysis capabilities, diagnostics, and user-experience improvements.

Platform and Architecture

• Migrated to .NET 10 for improved performance, security, and long-term support.
• Separated the model library (RMC.BestFit.dll) from the UI (RMC.BestFit.UI.dll) and App (RMC-BestFit.exe) layers. All distributions, estimators, and analyses are now accessible programmatically without a UI dependency.
• Modernized the WPF-Framework dependencies, including theme system and dialog infrastructure.
• Relicensed under the Zero-Clause BSD (0BSD) for maximum permissiveness in research, government, and commercial use.

New Analysis Capabilities

• Bulletin 17C Analysis with the LinkedMVN method, supporting systematic, uncertain, and censored records as well as prior information on parameters and quantiles.
• Coincident Frequency Analysis for joint exceedance probabilities of dependent variables.
• Time Series Analysis with ARIMAX models including covariate forecasting and Bayesian uncertainty.
• Composite Analysis for weighted model averaging across competing distributions, with WAIC, LOO-CV, DIC, and AIC weighting options.

Bayesian Diagnostics and Model Selection

• WAIC and LOO-CV with Pareto-smoothed importance sampling, enabling principled model selection and model averaging.
• Influence Diagnostics with Pareto-k shape parameters to identify high-leverage observations and to assess the influence of priors on posterior estimates.

User Experience

• Undo and Redo across the entire application, including data edits, plot customizations, distribution and parameter changes, and analysis settings.
• Dark Mode theming for reduced eye strain.
• International numbering corrections for users in locales using comma decimal separators.
• POT diagnostics in the Input Data element to support threshold selection for peaks-over-threshold studies.
• Enhanced time series download options including USGS surface water site retrieval, water-year and calendar-year time blocks, and custom block windows.

A Note on Bugs

With this many changes, expect some bugs. If you find one, please send a bug report so we can fix it before the final release. Your feedback during the beta is what will make the final release solid. Thank you for testing.

Version 2.0 Beta-3

RMC-BestFit Version 2.0 Beta-3 introduces a critical fix to the composite distribution module, addressing a defect that could impact certain flood frequency analyses. All users are encouraged to upgrade immediately to ensure accurate results.

Version 2.0 Beta-2

RMC-BestFit Version 2.0 Beta-2 introduces incremental improvements aimed at enhancing performance, flexibility, and ease of use. This update includes new features, Bayesian analysis enhancements, and minor bug fixes.

Highlights

• Bug Fixes

  • Several minor bugs resolved to improve software stability and user experience.

• New Features

  • HEC-DSS Importer: Import time series data directly from HEC-DSS files to streamline flood frequency analysis workflows.
  • Point Process Analysis: Fit a Generalized Extreme Value (GEV) distribution using peaks-over-threshold (POT) data.
  • Rating Curve Analysis: Fit multi-segment stage-discharge rating curves to support more detailed hydraulic modeling.

• Bayesian Analysis Enhancements

  • Posterior Point Estimator: Users can now select either the posterior mean or posterior mode as a point estimate for Bayesian analyses.
  • DIC Output and Model Averaging: All Bayesian models now report the Deviance Information Criterion (DIC). For univariate analyses, model averaging can be performed using DIC-based weights.

• Improved Defaults

  • Simulation Settings: Default simulation parameters have been refined to promote better convergence across a broad range of analyses.
  • Flat Priors for Nonstationary Models: Default flat priors have been improved to work reliably across a wide range of distributions and datasets.

Version 2.0 Beta

We are thrilled to announce the release of RMC-BestFit Version 2.0 Beta! This major update brings powerful new features designed to enhance time series analysis and modeling capabilities. Key enhancements include:

• Expanded Data Import: Import time series data directly from USGS and GHCN platforms, streamlining the creation of block maximum or peaks-over-threshold series for comprehensive analysis.
• Advanced Hypothesis Testing: Conduct rigorous hypothesis tests on time series data to identify trends, nonstationarity, and other statistical patterns.
• Measurement Error Integration: Incorporate measurement error into analyses for enhanced accuracy and reliability in results.
• Additional Probability Distributions: Access a wider selection of probability distributions to better fit diverse datasets and hydrological conditions.
• Nonstationary Flood Frequency Analysis: Model evolving flood frequency patterns over time to gain valuable insights for risk assessment and management.
• Improved Prior Distributions: Set parameter and quantile priors from 10 distribution options. Users can also apply Jeffreys’ Rule for Scale parameters, providing a superior noninformative prior and enabling compatibility with other Bayesian software, such as FLIKE®.
• Complex Modeling Techniques: Utilize mixture models, competing risks, and model averaging to address complex hydrological phenomena and improve model performance.
• Dependency Modeling with Bivariate Copulas: Capture the dependence structure between hydrological variables for a deeper understanding of your data.

With these powerful enhancements, RMC-BestFit 2.0 significantly broadens its capabilities, equipping users to tackle a wider array of hydrological challenges with greater precision and confidence.

Version 1.0

RMC-BestFit is a state-of-the-art Bayesian estimation and fitting software developed collaboratively by the U.S. Army Corps of Engineers' Risk Management Center (RMC) and Engineer Research and Development Center's Coastal and Hydraulics Laboratory (CHL). Tailored to expedite flood hazard assessments for the Flood Risk Management, Planning, and Dam and Levee Safety communities, the software employs a Bayesian framework to integrate a variety of data sources, including historical records, paleoflood evidence, regional data, rainfall-runoff models, and expert judgment.

This intuitive, menu-driven application provides a comprehensive environment for distribution fitting and Bayesian estimation. Its modern graphical interface, combined with robust data input, analysis, and reporting functionalities, enables users to effectively conduct flood frequency analyses and produce high-quality visualizations.

Version 1.0 was released in September 2020.