CHAPTER 3

Prepare and Monitor

Key Focus

Establishing workflows to anticipate flood impacts from forecasted events.

Key Takeaways

• Identifying priority assets, such as scour-critical bridges, and corresponding critical impact thresholds is essential to establishing monitoring protocols.
• Sensors provide key data inputs for establishing flood forecasts in real time.
• Standard operating procedures are useful tools to define monitoring timeframes, thresholds, and roles of key personnel.
• Partnerships can increase monitoring capacity.
  • Services from federal agencies, such as the NWS and USGS, offer forecast and monitoring support.
  • State and local public agencies may also offer on-site monitoring support.

Actions to Accomplish

• Identify who will monitor flood forecasts for which DOT assets.
• When preparing a monitoring plan, establish roles and communication channels between the state DOT and district/regional offices.
• Determine a timeline for the monitoring that is to be done before an event.

3.1 Overview

“Prepare and monitor” refers to the established DOT protocols and partnerships that lead to proactive flood event management. Figure 3-1 highlights mature flood preparation and monitoring capabilities.

3.2 Introduction

DOTs often rely on historical knowledge of flooding impacts and reported conditions in the field to prepare for flooding events. In a nationwide survey of state DOTs (Park et al. 2021), 38% of participants reported that their state DOT did not perform flood prediction modeling. Monitoring protocols help establish a more proactive approach to flood event management. This chapter will focus primarily on the steps taken before a flooding event. Response during a flooding event is covered in Chapter 4: Operations.

This chapter presents key questions for DOTs to consider when documenting workflows for anticipating flood impacts and draws from case studies and examples of existing DOT practices for monitoring and preparing for flooding events. DOTs that want to start flood forecasting will find information to begin building a monitoring plan and identify key personnel to engage. DOTs that already flood forecast and want to enhance their capabilities will find effective practices that leverage data sources or tools they may be using.

A mature flood monitoring capability will include:

• The ability to anticipate threats to system level of service and customer mobility,
• A documented monitoring plan that establishes thresholds for critical assets and identifies personnel roles and responsibilities, and
• Protocols for sharing flood forecasts with internal and external partners.

This chapter is organized into the following sections to cover how DOTs may start or expand their monitoring capabilities:

• Developing a Monitoring Plan. Provides key questions to build or enhance a flood risk monitoring plan and examples of key features to enhance monitoring efficiency.
• Leveraging Real-Time Data. Describes data types available that may support a DOT’s efforts to build out real-time road network monitoring, including sensors, and options to manage data from multiple sources.
• Partnerships in Practice. Identifies opportunities for coordination within a DOT and with external partners to monitor flooding events.

3.3 Developing a Monitoring Plan

Monitoring plans detail protocols for the types of information and frequency at which it should be collected when preparing for a flooding event. Such plans also describe what tools should be leveraged in the process. A documented monitoring protocol can support the transfer of knowledge and employee onboarding, as well as communicating processes with partners to leverage resources—both within the DOT and externally. The foundational questions in Table 3-1 can be used to guide the development of a monitoring plan. DOTs can also use these questions to evaluate existing monitoring plans for completeness and identify opportunities for improvement.

Table 3-1. Foundational questions to guide the development or completion of a monitoring plan.

Uncertainty: How Do DOTs Measure or Assess Uncertainty?

One of the most challenging elements of beginning and operating a flood forecasting system is being able to answer the question: “How sure are you?” Clear communication about what forecasts mean and their limitations is essential to building organizational confidence. DOTs use a variety of sources to measure uncertainty, including:

• Evaluating Probabilities. Quantifying the likelihood of the flood event occurring. In the project survey of state DOTs, 50% of respondents reported evaluating probabilities as their method to account for and communicate uncertainty when taking action for specific assets.
• False Alarm Rates. Tracking in real time what happened, along with a history of the prediction.
• Confusion Matrix. Classifying events to compare when observations align with the forecast (true positive); when the prediction showed an event, but the event did not happen (false positive); and when a prediction did not show an event, but an event occurred (false negative). This can be used as a tool to understand quality.
• Tool Assessment Reviews. Conducting an independent review of flood prediction models in terms of accuracy, applicability, timeliness, and usability (Park et al. 2021).

As part of this research effort, a confusion matrix was used to assess the performance of the National Water Model (NWM) stream discharge forecasts,

which had a resulting global 73% true positive rate for discharge spikes from medium-range forecasts. The research team established a threshold of 2 cubic meters per second to weed out smaller bumps in a forecast. An additional filter was added to account for unusually high flow for a specific stream. For example, it is more typical for a small stream to double in size than a large river. These predictions were only included if they occurred more than 16 hours before an event.

Visualizing projected flood impacts can also play an important role in communicating uncertainty. Displaying potential impacts in terms of event likelihood can help DOTs direct resources to monitor areas with higher risk. The Colorado Water Conservation Board issues a Flood Threat Bulletin that provides an overview of the potential flood threat facing the state of Colorado. The bulletin is issued regularly from May 1st through September 30th. A precipitation map is included, displaying the quantitative precipitation forecasts (QPFs) in terms of the chance of precipitation, as shown in Figure 3-2 (Colorado Water Conservation Board 2023). An additional product, a Flood Threat Outlook, is produced twice weekly and provides an overview of the flood threat and precipitation amount/probabilities in the state over the following 15 days.

While this type of quantitative display can be helpful for technical staff, it is important to also consider how risk and uncertainty can be communicated to different audiences. The Flood Threat Bulletin presents flood threat information in qualitative terms using the following definitions for each threat level:

• None: less than 10% expected probability of flooding within a given county
• Low: 10%–30%
• Moderate: 30%–60%
• High: greater than 60%
• High impact: greater than 60% and a severe threat to life and property.

These threat levels are based on several factors, including the magnitude and duration of the expected event. These qualitative categories account for a margin of error in the precipitation

estimates while providing a quick summary of flood threats that can support DOT personnel and the public in identifying elevated flood threats.

A flood forecasting team can assess uncertainty based on the level of agreement between multiple forecasts or hydraulic models. A study published in the Journal of the American Water Resources Association, “A Hydraulic MultiModel Ensemble Framework for Visualizing Flood Inundation Uncertainty,” developed an interactive web-based visualization of flood inundation (Zarzar et al. 2018). Uncertainty is classified based on model ensemble agreement, meaning that areas with high model agreement have higher confidence and higher flood risk than areas with low model agreement, as shown in Figure 3-3. While this ensemble hydraulic modeling approach is intensive, it illustrates an innovative way to visualize uncertainty and support focusing resources where the flood risk is greatest (Zarzar et al. 2018).

Monitoring Interface Elements

Data presentation interfaces can streamline a proactive monitoring approach by supporting the quick identification of assets with potential flood impact. To inform the development of this guide, the research team created a dashboard to demonstrate the potential functionality, use cases, and operational utility that a flood forecast capability would offer at locations of interest for participating DOTs. The demonstration dashboard used the NWM, research team-derived, or DOT-supplied rating curves and estimates of asset impact thresholds to forecast impacts at locations of interest provided by the 10 DOTs participating. The platform enabled the engagement with DOT participants to move beyond hypothetical questions (such as “How would you use flood forecast information?”) to more practical application insights. Participating DOT partners were better able to discuss the operational possibilities and challenges having such a capability would pose.

Using the demonstration dashboard, DOTs identified key features that would support their monitoring efforts, as shown in Figure 3-4.

Interface Maintenance

Maintenance is important to take into consideration when assessing the level of effort needed to operate a flood monitoring interface. This maintenance can include troubleshooting system errors and routine checks to ensure proper interface function. Some commercial options may offer interface maintenance as part of the service provided. The following key questions can assist DOTs in developing a maintenance plan for a flood monitoring interface:

• Will interface maintenance be conducted internally or by an external partner?
• Who will be the designated point of contact for reporting issues encountered in the monitoring interface?
• How will issues encountered be recorded and tracked?
• What end-users need to be notified if the monitoring system is not functioning properly?
• How often will routine maintenance checks be performed on the monitoring interface?

3.4 Leveraging Real-Time Data

Real-time data can support predictive flood hazard modeling efforts and provide actionable information in the moment. Data types available include broad-scale weather reporting and forecasting, video feeds of critical network nodes, and sensors monitoring water-body or road surface conditions. DOTs can leverage existing data sources across various data types to develop monitoring plans and supplement them with additional video or sensor networks to increase monitoring specificity. Table 3-2 and Table 3-3 summarize available data sources that may support DOTs’ efforts to build out real-time road network monitoring and future flood hazard forecasting.

Data Types

Real-time data can come from a variety of sources, such as those shown in Table 3-2.

The use of real-time data is already widespread for DOTs across a range of data types. Nearly all survey respondents indicated that they use some form of real-time data to monitor roadway conditions. Many survey respondents indicated that their teams already leverage water level and discharge data from USGS gauge services and, to a lesser extent, local or state gauge data. The DOTs of Ohio, North Carolina, California, Washington, Missouri, Minnesota, and Kansas indicated that they rely on closed-circuit television (CCTV) footage to monitor roadway conditions; however, none of these DOTs currently use CCTV footage to monitor roadways specifically for flooding or ponding. The California DOT (Caltrans) is the only one that reported using drones to monitor roadways; this method could benefit DOTs interested in monitoring roadway conditions at various locations that do not need continuous monitoring.

A survey administered for a previous NCHRP project reinforces this finding. Based on 43 DOT respondents, the three most reported instruments for flood monitoring were federal stream gauges (41 respondents), federal rain gauges (25 respondents), and nonfederal stream gauges (17 respondents) (Park et al. 2021).

Table 3-2. Real-time data source types.

Table 3-3. Real-time data sources to support flood monitoring.

Sources: National Weather Service (n.d.), U.S. Geological Survey (n.d.), Delaware River Basin Commission (2023). Note: N/A = not applicable.

Table 3-4. Two state systems for data management.

Data Management

Aggregating real-time data across various data types and sources can improve monitoring resolution. However, managing data across types and sources can be challenging. At the national scale, the USGS National Water Information System (NWIS) is a data-sharing platform that supports the acquisition, processing, and long-term storage of water data and includes all historical USGS gauge records in the United States. In a survey administered for a previous NCHRP project, approximately half of state DOTs reported using NWIS for flood monitoring (Park et al. 2021). At the state level, Table 3-4 highlights existing systems from North Carolina DOT (NCDOT) and Texas DOT (TxDOT) that combine various systems to provide information at a higher spatial and temporal specificity that would not be possible with only one data source.

In addition to DriveTexas.org, a public-facing web app with current highway conditions across the state, TxDOT is working on a pilot effort to extend monitoring coverage to 80 radar-based gauges, installed and operated by USGS, that can measure both stage and velocity and provide the information necessary to calculate discharge. These gauges also connect to a cellular network, provide real-time data updates, and can produce water elevation data accurate to 2 millimeters. TxDOT found that coupling the radar streamflow gauges with lidar data for stream cross-sections can support predicting road inundation (McCarty et al. 2022).

3.5 Partnerships in Practice

An array of key players are needed to operationalize, monitor, and make decisions based on flood forecasting information. A central office will typically monitor flood forecasting in partnership with district or field offices and will also determine an activation approach based on triggers. When an asset appears to be subject to a hazard impact, a field office or district will inspect or validate that prediction by going to the site to support the data integrity of the forecasting capability. If an asset is washed out or otherwise affected, then someone from the field will assess the damage and come up with a repair, potentially working with others when the damage is severe. All these steps require clear communication processes, roles, and effective tools.

Private partners are often core team members involved in operationalizing a flood forecasting system. Some are involved in contributing key datasets. Others, such as BridgeWatch, monitor assets via a spatially explicit system and can provide real-time alerts or warnings. For example, NCDOT uses BridgeWatch to monitor 15,000 structures. Federal and state partners can also be integrated into the DOT monitoring protocols. For example, Ohio relies on its Office of Local Weather Prediction, and many DOTs rely on NWS briefings. DOT participants cited USGS as a potential partner in helping with field data collection.

Communication Methods

Clear roles within internal DOT departments enable regular and consistent communications. Examples of communications are shown in Table 3-5, and a few examples of how to approach these communications are discussed in the following:

• Create a method for the field to input information to a central office. DOTs have mentioned using a customer service phone number, a form for responders, and a GIS map or Excel spreadsheet to track key information.
• Make information digestible quickly. For example, Caltrans created a consistent format across its 10 offices that could be understood in 15 seconds.
• Solicit feedback from field staff and others with institutional knowledge of where it floods to identify locations of interest. For example, Minnesota sent a message to districts to ask each to send 5 to 10 spots with problems.

Common ways to collect flood monitoring information are directly from a gauge as well as from reports from staff, emergency services, citizens, and images and videos (Park et al. 2021).

Table 3-5. Sample messaging.

External Partners

Once internal roles and communications are established, state DOTs will want to engage partner departments or agencies. There needs to be a base of trust and an understanding of the consequences of impact messaging before bringing in key partners as part of a community of practice. Otherwise, there is a risk external partners will hear about an event and not act or will miscommunicate the forecasts. A validation practice is needed to improve certainty as a DOT’s communications expand to a broader audience. Transparency around uncertainties with the information is important. However, once that base of trust exists, external partners can help maximize the reach and impact of forecasts.

Each external department will have a different purpose and may ask for communication about different types of data or information. For example, an emergency manager will want to know, in ways that allow them to respond to events quickly, which assets may be most affected by a natural hazard, while, to inform prioritizing future investments, state hazard mitigation staff will want to know how often a particular asset has been affected. A triggering event may be an all-hands-on-deck event for field staff, emergency managers, and others.

Some DOTs may also want to include local or county agencies. For example, to leverage their state tools, the Minnesota and Virginia DOTs were interested in including counties and cities that own bridges and other transportation assets. Some cities and counties may already monitor assets, or DOTs may host and report data for locally owned assets.