Development of a Real-Time Flood Forecasting System for Railroad Crossings in the Midwest

With associated costs of millions of dollars, floods are one of the major concerns for the railroad transportation system. Floods wash out roadbeds, bury rail lines, destroy bridges, and derail trains, causing delays and expensive operational and maintenance costs. In the U.S. Midwest, the floods of 2008 and 2019 caused infrastructure damages and multiple delays to the railroad system. Considering the floods’ relevance to the railroad industry, we propose to develop a real-time flood forecasting system for the railroad crossings in the Midwest. Our efforts will focus on Iowa, the East region of Nebraska, and the northern area of Missouri. We will build the system on top of the Iowa Flood Center (IFC) developments including the hydrological Hillslope Link Model (HLM) with data assimilation and the representation of processes relevant for the region. Also, depending on the needs, we may install bridge sensors upstream of some railroad crossings. We will use a network closely following the official national hydrographic network (NHDplusV2) to better represent the railroad crossings. Using rainfall as the primary forcing, HLM will forecast and report hourly streamflow at every channel. The proposed system will provide accurate real-time forecasts for the railroad crossings in the Midwest, supporting logistics during storm events and flooding season.

"The project has the following tasks: • Data pre-processing: The system setup will start by processing elevation and NHDPlusV2 data to obtain a network representing most channels with a good resolution. The network delineation also includes tracing the sub-watersheds (hillslopes) corresponding to each channel element of the network. Finally, we will generate the forcing and parameters files required by HLM. • Open-loop validation: Using MRMS as the main forcing, we will validate HLM simulations against USGS observations. This step will allow us to identify required improvements in the HLM setup. • Forecast validation: Using QPFs, we will validate hypothetical previous flood forecasts with and without downstream data assimilation. The validation will also include comparisons with previous railroad crossing flooding events. • Model deployment: After the validation, we will perform the operational setup of the flood forecast system in a web platform."

"We expect to have the following results and products: • A river network that follows the official NHDPlusV2 network while representing most existing channels. The network segments will refer to the railroad crossings. • A distributed hydrological model with the skill to capture peak flows at multiple scales, upstream from the crossings. • A flood forecasting platform for visualization of the railroad system in the Midwest. • Papers describing the implementation."

A flood forecasting system oriented to railroad crossings will work as a support in the railway operation decision-taking. With this support, railroad operations will have better information regarding floods, reducing operational costs and associated risks. For Iowa, the system will be included in the Iowa Flood Information System, a browser based platform that has been visited nearly five million times since its deployment by the Iowa Flood center at the University of Iowa.