Leveraging HEC-RAS Models for the Real Time Hydraulic Modeling and Flood Forecasting for the Johnson County Stormwater Management Program, Kansas

The Stormwater Management Program (SMP) is a Johnson County, Kansas program which partners with the 20 cities in the County to manage stormwater and is funded by a 1/10th of one percent, county-wide sales tax. It administers these funds on behalf of the Cities, historically by providing matching funds to Cities for eligible projects, including study, design, and construction projects. SMP has invested significantly on StormWatch Alert 2 systems (www.stormwatch.com) over the past 30 years to implement real-time early flood warning systems. SMP has funded the installation and maintenance of many of the sites throughout the region and has utilized the data generated from the system for SMP-sponsored studies and projects. SMP also funds a portion of the cost to maintain and improve the website. Johnson County owns 68 of the 108 sites in the system. In addition, City of Overland Park, Kansas Department of Transportation, and City of Kansas City, MO owns several sensors. Currently, SMP with the help of City of Overland Park uses manual process of forecasting flooding conditions and implements emergency management procedures within the county based on existing stream level and National Weather Service's forecasted weather information. However, this process is very tedious, time consuming and not reliable to accurately forecast future flood conditions. Given these challenges, SMP engaged with NEER for the pilot study to utilize its cloud-based Machine Learning (ML) solution to automatically forecast early flood warning system (up to 24 hours) for the Watershed Organization 1 using existing hydraulic models and StormWatch real time datasets. NEER obtained existing HEC-1 and HEC-RAS models from FEMA for Watershed Organization 1. After verifying the integrity of the models, NEER converted existing HEC-1 and HEC-RAS models into EPA-SWMM model. During the conversion process, NEER followed the standard engineering practices to update the existing hydrology (subbasin and storage data) and hydraulic (open channel geometry, culverts, and bridge data) characteristics. After updating all the parameters, the hydraulic model was calibrated using StormWatch gage data collected during 2016 to 2020. There are a variety of statistical measures used to measure the goodness-of-fit between a long term continuous measured and a modeled hydrograph. For this study, statistical measures Integral Square Error (ISE) and Nash'Sutcliffe efficiency (NSE) were used as a single, non-subjective, statistical measure of model calibration (https://www.chijournal.org/C414). Generally, calibration results showed very good to excellent NSE and ISE range for all the gage locations. After the calibration, NEER set up a continuous real time and forecasting stormwater simulation model. In this step, StormWatch gage data (rainfall and stage collected every 5 minute) was obtained and stored in a Time Series Database. In addition, the 24-hour forecasted rainfall data obtained from the National Weather Service was also used in forecasting the stage and water surface elevation along the Brush and Turkey Creek. This real time and forecasting model that is scheduled to run every 6 hours, provides a predicted and forecasted floodplain boundary, depth grid, water surface elevation grid, velocity grid, and flood severity grid. that can be used for operational decisions. The total number of buildings and roads flooded, and operational recommendations (such as closing of roads, and evacuation of buildings) for every 6 hours is stored and displayed in the dashboard.