Scaling Hydraulic Models for National Flood Inundation Mapping

University of Kansas | Slawson 198 | Zoom Link - 5995eight4

August 10, 2026 at 9:00 AM

Presentation Navigation Tips

Slide layout : This deck has one axis, use any key to advance.

: This deck has interactive elements

Items linked/bordered in green are cited in the tooltip on hover.

Items linked/bordered in blue are hyperlinked to relevant resources.

Photos are Allowed | Questions are Encouraged

: ~35 minutes | Last updated:08/09/2026 15:30:59

! PLEASE !

Interrupt me and ask questions or clarifications.

I’m here to talk with you now, not to these slides.

Control tips

My preferred (FOSS) flavor of slidedecks, revealjs, has intuitive but none the less unconventional PowerPoint presentation controls:

• Slides dynamically resize to use the entirety of the browser window, but you can still fullscreen with F.
  • This slide has a red border indicating the content extent.
• Slide navigation is mode dependent. If there are vertical slides, press space, N, or the down arrow key, not the right arrow to advance slides
• Press M to open to the menu, Press O for the slide deck overview, Press B to black out the presentation screen, Press S for a speaker view.
• You can use the chalkboard to freemouse/touchpad draw.
• Slides should render as designed¹ but you can press Alt/Opt + click on the slide to zoom in. Increase text size with Alt/Opt + +, Alt/Opt + - to decrease, and Alt/Opt + 0 reset to the default scale.
• Press C to declare victory and head home.²

This presentation

Speaker notes

1. tested in an updated version of google chrome in light mode

2. A favored quote from one of my giants: Dr. David Maidment

If I have seen further than others, it is by standing upon the shoulders of giants.
- Sir Isaac Newton

There are far too many giants to thank.
AI Prompt: Can you draw me a picture of Sir Isaac Newton standing in a river shaded by many apple trees

Scaling Hydraulic Models for National Flood Inundation Mapping

Outline

1. Defining My Question: Elaborating on my Jargon
2. How We Stage & Scale: RRASSLER & Reusing Legacy Data
3. How We Automate: RAS2FIM and Inundation Libraries
4. How We Compare: Model Bakeoff
5. What We Learned: Results & Insights
6. Where We Go: Healing & National FIM

Takeaways

• A better understanding of the tools and techniques used to create FIM
• A better grasp on the impact different modeling choices have across our scales
• A better appreciation for how YOUR efforts can integrate into a shared national flood map
• A picture of something worth your attention.

Thank you all for being here.

Like Bob Ross, I now have PTSD from my work, but I also really admire the artfully effortless style in which he sculpts the landscape under his brush. That’s exactly the tension at the heart of my dissertation. We have amazing, hand-sculpted local models, but national Flood Inundation Mapping (FIM) requires us to do this everywhere, instantly, and cohesively.

Today, we’re going to talk about how we scale that brush. We’ll start by setting the stage with the history and our current technical debt. Then, I’ll introduce RRASSLER, which is how we rescue legacy data. We’ll put that data to the test in a model bakeoff, and finally, look at what this means for the future of national FIM.

Hopefully, by the end of this, I’ll have painted a picture worth your attention.

Terminology levelset

• FIM: Flood Inundation Modeling/Mapping — Delineating water extents/depths.
• HAND / REM: Height Above Nearest Drainage / Relative Elevation Model — Topographic normalized elevation surfaces used in geospatial FIM.
• Inundation Library: A structured database of pre-computed depth rasters, rating curves, or hydraulic predictions.
• Volume: Parsed, modular database subsets categorized by reach resolution, flow step, or spatial boundary.
• “Operational” Library: Constructed at a resolution sufficient for operational decision-making—encompassing multi-resolution ensembles rather than a handful of static point predictions.
• FAIR: guidelines to improve the Findability, Accessibility, Interoperability, and Reuse of digital assets

AI Overview: Jargon is the specialized, technical language or vocabulary used by a particular profession, group, or activity…

My goal is to communicate the state of the art. If you don’t follow these terms you’re missing some critical context/concept handles.

Why FIM?

Although I was neither there nor the victor, I still get to write the history.

Why Now?

How?

FIModeling: Flood Inundation Modeling, or the act of representing the spatial extent, depth, and movement of floodwaters across a landscape. This can be accomplished using either geospatial or Hydrodynamic methods.

Geospatial: utilizes spatial relationships (local, focal, and zonal operations) and more simplistic calculations to approximate floodwaters across a landscape.

How?

FIModeling: Flood Inundation Modeling, or the act of representing the spatial extent, depth, and movement of floodwaters across a landscape. This can be accomplished using either geospatial or Hydrodynamic methods.

Hydrodynamic: Solves variations of the shallow water equations to more exhaustively simulate the flow dynamics of floodwaters across a landscape.

Some New Developments on Two Separate Topics: S…

Jude’s Thesis

A Modern HEC-RAS

USACE History

• first engineer officers of the Army on June 16, 1775
• permanent branch on March 16, 1802

HEC History

• HEC-2: WIZ by Bill S. Eichert in 1964. This early version was developed on a GE 225 system at the Corps of Engineers Tulsa District office. In 1966 the first FORTRAN version of HEC-2 was released by the Hydrologic Engineering Center (HEC) under the name “Backwater Any Cross Section.”
• 1995: first GUI software
• Now we’re here (V4 included enhancements for visualization, unsteady flow, and sediment transport)

St Louis Dredge potter (1932)

The First 2D HEC-RAS

HEC2

HEC-RAS 2025

HEC-alpha

AHPS libraries

AHPS

The Problem: Friction at Scale

Flood Modeling’s Fragmentation Problem

While FIM methodologies and datasets have rapidly expanded, these efforts remain isolated and rarely communicated clearly or combined effectively.

• The Scale Disconnect: Transitioning from reach-scale hydraulic precision to national-scale mapping creates massive technical friction.

It’s me, hi… I’m the problem, it’s me.

But actually, the problem is the sheer technical friction in our field. We have amazing tools and models, but they are completely siloed. We are trapped behind specialized formats and a lack of interoperability.

Whenever we try to scale up reach-level models, or chop down national models to fit regional forecasts, we either lose data fidelity or hydraulic accuracy. We have all these disparate libraries, but no unified operating picture. My work exists because we need a FAIR framework to stop fighting the data pipeline and start actually communicating map generation for emergency responders.

The Problem: Friction at Scale

Flood Modeling’s Fragmentation Problem

While FIM methodologies and datasets have rapidly expanded, these efforts remain isolated and rarely communicated clearly or combined effectively.

• The Scale Disconnect: Transitioning from reach-scale hydraulic precision to national-scale mapping creates massive technical friction.

The Problem: Interoperating modeling frameworks

Flood Modeling’s Fragmentation Problem

While FIM methodologies and datasets have rapidly expanded, these efforts remain isolated and rarely communicated clearly or combined effectively.

• The Scale Disconnect: Transitioning from reach-scale hydraulic precision to national-scale mapping creates massive technical friction.
• Data & Language Barriers: Specialized formats, less-than FAIR data, and model-specific jargon trap insights in silos.

The Result: We spend our time fighting the mechanics of modeling rather than communicating actionable emergency response maps.

How do we bridge the orders of magnitude that separate FIM? How can we more cleanly link otherwise independent libraries?

The HEC-RAS model format

The Theoretical Standard

Files to Track

File grep Pattern (# denotes single numeric wildcard)	HEC-RAS Model Use
.g##	Geometry definitions
.prj	Projection (can be non-standard proj4 string defined file)
.prj	Project (same extension, defines how RAS models are wired)
.p##	Plan file, used to drive the model
.f##	Steady Flow file. Profile information, flow data and boundary conditions
.h##	Hydraulic Design data file
.v##	Velocity file
.o##	Output file
.r##	Run file for steady flow
.u##	unsteady Flow file. Profile information, flow data and boundary conditions
.x##	Run file for unsteady flow
.dss	Data files
.rasmap	Output plan

These files, followed by a .hdf, are transformations of their counterparts needed for newer versions of HEC-RAS and are automatically created as needed.

Sidecar files are fun.

The HEC-RAS model format

Model Standards

Files to Track

File grep Pattern (# denotes single numeric wildcard)	HEC-RAS Model Use
.g##	Geometry definitions
.prj	Projection (can be non-standard proj4 string defined file)
.prj	Project (same extension, defines how RAS models are wired)
.p##	Plan file, used to drive the model
.f##	Steady Flow file. Profile information, flow data and boundary conditions
.h##	Hydraulic Design data file
.v##	Velocity file
.o##	Output file
.r##	Run file for steady flow
.u##	unsteady Flow file. Profile information, flow data and boundary conditions
.x##	Run file for unsteady flow
.dss	Data files
.rasmap	Output plan

These files, followed by a .hdf, are transformations of their counterparts needed for newer versions of HEC-RAS and are automatically created as needed.

Sidecar files are fun.

The HEC-RAS model format

A “real-world” example

Files to Track

File grep Pattern (# denotes single numeric wildcard)	HEC-RAS Model Use
.g##	Geometry definitions
.prj	Projection (can be non-standard proj4 string defined file)
.prj	Project (same extension, defines how RAS models are wired)
.p##	Plan file, used to drive the model
.f##	Steady Flow file. Profile information, flow data and boundary conditions
.h##	Hydraulic Design data file
.v##	Velocity file
.o##	Output file
.r##	Run file for steady flow
.u##	unsteady Flow file. Profile information, flow data and boundary conditions
.x##	Run file for unsteady flow
.dss	Data files
.rasmap	Output plan

These files, followed by a .hdf, are transformations of their counterparts needed for newer versions of HEC-RAS and are automatically created as needed.

Data from the wild can be a bit unruly.

Even After Overcoming the Pain of Reading in a HEC-RAS model…

Other Potential Readers

• Point and click through the HEC-RAS interface to export
• Con: Unaccountable and unsustainable
• Pro: I can do it
• MCAT-RAS
  • I am unfamiliar with go, it’s implemented language.
• pyHMT2D is primarily oriented towards 2D.

Can you answer these questions?

• Where are models?
• What cross sections do I have for my network?
• Is this the same model?

This pain is not trivial, I weep weekly.

Other archival, tangentially related, draft, or circuitous readers. See related utils

• RAS2FIM has readers, which we will extend as part of RRASSLER
• https://github.com/solomonvimal/pyras
• https://github.com/larflows/PyRASFile
• https://github.com/mikebannis/parserasgeo
• https://hatarilabs.com/ih-en/open-modify-run-and-read-output-from-hec-ras-models-with-python-tutorial

Models from Different Providers?

Texas Water Development Board

FEMA Region 6

asynchronous update cycles

• https://www.twdb.texas.gov/flood/science/ble.asp
• https://www.arcgis.com/apps/dashboards/45d3284db7e64694bdcece58c6de39fc

Although the download url is the same, the status of the maps is not the same and makes determining what model an end user is querying for difficult.

Models from Broad Geographies

Clashing spatial units

• Need to pull the entire watershed to get any data
• Take a minute to acknowledge the geographic pain we all just experienced. Am I worried AI will take my job? A boy can dream…
• Helper function

Models at street scale: Example Austin FloodPro

Manual subsetting

• You have to know your street address (easy), or know your watershed name (harder)
• either way are restricted to the watershed scale download
• https://maps.austintexas.gov/FloodPro/

RRASSLER Can!

“Where are models?” – See the mapping vignette!

Review: RAS2FIM V1

RAS2FIM V1

Review: RAS2FIM V2

RAS2FIM V2

Model chopping impacts

Version 1

Version 2

Why do I do this to myself?

Differences in automation

Model Comparison Metrics
Analysis of WSE Residuals (V2 - V1) across interpolated flow range
River Stationing	MAE (ft)	RMSE (ft)	Mean Bias (ft)	Max Absolute Diff (ft)
704	0.024	0.057	0.008	0.478
534	0.055	0.129	0.022	0.549
250	0.026	0.057	−0.014	0.351
149	0.252	0.631	0.117	2.618
Metrics calculated over the shared discharge domain.

The FIM Bake-off: Mosaics of CatFIM

Stage Based Libraries

There are lots of ways we map FIM. Let’s take a look at some of the most popular:

• A gage based FIM library.
• A HAND library from OWP FIM.
• A categorical FLDPLN library.

Judging the Bake-off: Common Metrics

Flood Model Performance Metrics

To compare these models against our AHPS ground truth, we’ll deploy common validation metrics:

Metric	Range	Ideal	Formula	Hydrologic Context
CSI	0 to 1	1	\(\frac{TP}{TP + FP + FN}\)	Benchmarks for flash flood guidance often peak around 0.20 to 0.44; global models frequently score between 0.4 and 0.6.
F1	0 to 1	1	\(\frac{2TP}{2TP + FP + FN}\)	Usually higher than CSI. High-performing damage prediction models often achieve 0.76 or higher.

Setting the stage for our model validation. We have harmonized these rasters to the same CRS and extent so we can do an apples-to-apples cell comparison.

Optimizing Access Patterns

Conclusions and Contributions

Conclusions

• FAIR principles unlock scale: Data transformations are required to facilitate complex process integration across domains.
• Legacy models have a second life: Existing models contain highly usable information when properly spatialized and harmonized.
• Monolithic > Segmented: Processing large hydraulic models as units significantly transforms hydraulic predictions, physical subsetting improves computational speed.
• Decoupling is necessary: Separating hydrologic, hydraulic, and geospatial drivers from map creation allows disparate libraries to form a clearer operating picture.

Contributions

• RRASSLER: Developed a novel utility to spatialize legacy HEC-RAS models, aligning them with FAIR data principles.
• RAS2FIM: Created an automated framework to generate inundation libraries that integrate directly with National Water Model (NWM) forecasts.
• Architectural Clarity: Oriented towards the separation of concerns between Modeling (the underlying mechanics) and Mapping (the communication of those results).
• Towards Operational implementation: Moved the conceptual bottleneck from mundging towards the communication of actionable maps.

Why do I do this to myself? Illiterate, incompetent, Imbecile

Thank you

Outcomes:

• A better understanding of the tools and techniques used to create FIM.
• A better grasp on the impact different modeling choices have across our scales.
• A better appreciation for how YOUR efforts can integrate into a shared national flood map.
• A picture of something worth your attention.

Next Steps:

• Scale and extension: Both RRASSLER archives and RAS2FIM methodologies have room for enhancements and extensions.
• Mapping and Modeling surface adaptation - healing and impacts.
• Optimization: Accessibility across user base and compute environment.
• Write: One of these day’s I’ll figure that out.

This presentation

See notes (s) and frames for image sources.

ASFPM Mapping & Engineering Standards Committe, David Knipe, and Steve Story. 2017. Guidance for Flood Risk Analysis and Mapping. FEMA.

Dewberry. 2011. Guidelines for the Development of Advanced Hydrologic Prediction Service Flood Inundation Mapping. NOAA.

FEMA. 2021. Base Level Engineering (BLE) Analysis and Mapping. FEMA.

Kastens, Jude H. 2008. “Some New Developments on Two Separate Topics: Statistical Cross Validation and Floodplain Mapping.” PhD thesis, University of Kansas.

Liu, Yan Y., David R. Maidment, David G. Tarboton, Xing Zheng, and Shaowen Wang. 2018. “A CyberGIS Integration and Computation Framework for High-Resolution Continental-Scale Flood Inundation Mapping.” JAWRA Journal of the American Water Resources Association 54 (4): 770–84. https://doi.org/10.1111/1752-1688.12660.

Wallace, David S. 2022. Flood Decision Support Toolbox Executive Summary and Submittal Guidance.

Thank you for allowing me to ramble at you.