2D Storm Water Modeling: Case studies and lessons learned

Abstract
Two-dimensional (2D) storm models are crucial for assessing current and future conditions of various storm system components. These models help identify system deficiencies and understand flooding causes. Two common methods for developing 2D stormwater models are the Catchment-Based Hydrology method, which calculates runoff at the subcatchment level, and the rain-on-mesh approach, where rainfall applies directly to the 2D mesh. Each has its own strengths and weaknesses, with the choice dependent on available data and the required detail.

This presentation discusses insights and findings from two case studies: Houston, TX's stormwater master plan using the rain-on-mesh approach, and Fayetteville, NC's watershed master plan using Catchment-Based Hydrology for secondary systems and HECRAS for primary streams and channels.

Objective
Flooding in U.S. communities, exacerbated by riverine impacts and outdated stormwater systems, can be mitigated through 2D stormwater modeling, which improves flood risk prediction and system deficiency identification. The presentation highlights lessons and challenges from case studies in Fayetteville, NC, utilizing a Catchment-Based Hydrology approach known for its lower computational requirements and reliance on precise subcatchment delineation and runoff parameters; and Houston, TX, applying a rain-on-mesh approach, which captures detailed surface flow routes but demands higher computational resources.

The presentation details the necessity of 2D stormwater modeling, the advantages and drawbacks of different methods, and critical steps in data gap analysis involving storm network connectivity, sewer data, and LiDAR data review.

Additionally, this presentation elaborates on model validation steps and concludes with a discussion on addressing concern areas and selecting solutions from one of the case studies.

Model Development Methodology
Model development typically begins with a data gap analysis for inputs such as impervious cover, building footprints, first-floor elevations, pavement edges, storm systems, channel systems, and roadside ditches. Field surveys fill in missing GIS and as-built drawing data. LiDAR data updates are crucial in redeveloped areas or where LiDAR cross-sections are shallower than as-built drawings.

Modeling approaches impact input details, such as detailed breaklines for rain-on-mesh models or runoff catchment delineation for Catchment-Based Hydrology models. The modeling approach selection, guided by project goals and terrain data quality, is critical.

In Catchment-Based Hydrology, runoff calculations at the catchment level, applied to model nodes and 2D meshes, require lower computational needs since most of the 2D mesh remains dry until flooding occurs. This method may oversimplify surface flow routing and depends on accurate catchment delineation and runoff parameters. Conversely, the rain-on-mesh approach directly applies rainfall to the mesh, relying on detailed terrain and land-use data, leading to comprehensive surface flow routes but higher computational demands.

Next, input layers for both 1D and 2D model components are prepared, including storm pipes, 1D open channels, main channels where no crossing structures are needed, and roadside ditches. LiDAR data is crucial for terrain representation and elevational changes. Accurate boundary conditions, including inflows from offsite areas and downstream water levels, are essential for model stability.

Model Validation
Model validation ensures accuracy. When flow data is unavailable, existing flood claims can serve as validation sources. For Houston's stormwater model, three storm events per watershed (selected by flood claims data availability) were used for validation. Hurricane Harvey was universally applied due to its significance. One event was used to validate the closed storm system without receiving channel water elevation impacts.

Validation criteria included a 50% match between flooded structures based on flood claims and model results and a 75% match between flooded parcels based on flood claims and model results. Models showed good alignment with claims data, and flood extents were consistent with claims locations and previous models.

Case Study Results
After validation, models ran seven storm frequency simulations to determine the service level for system components and pinpoint deficiencies. Results identified locations with hydraulic deficiencies due to storm system capacity or receiving channel water elevation, leading to better future solutions.

Conclusion and Lessons Learned
- The 2D model assessed system performance during various design storms.
- High-quality inputs reduced stabilization and validation time.
- Applied initial conditions or baseflow where necessary.
- The 2D model was validated using several historical rain events' flood claims.
- Adequate troubleshooting time is essential.
- Field visits during and after storms confirmed stormwater flooding conditions as part of the validation process.
- Resident input and social media posts during large storms were secondary validation sources.