Critical data shortage may leave millions unaware of flood risk across nation
As flood disparities worsen, CSS researchers find that improving data networks with low-cost flood sensors can help burdened communities respond.
A record number of floods have struck the U.S. over the past decade, devastating communities across the nation. In particular, millions of low-income, predominately minority-owned households remain most at risk — an inequality that is predicted to grow worse as climate change intensifies.
The Center for Social Solutions (CSS) has been working with the Real-Time Water Systems Lab at the University of Michigan to improve data networks so that burdened communities can more effectively respond to major flood events. In September, CSS researchers Brad Bottoms and Julie Arbit presented their most recent findings — “Equity in Flood Risk” — at a national emergency management conference in Grand Rapids, MI.
“Studies show that historically disadvantaged communities are both most at risk to flooding, as well as less likely to be able to recover,” research associate Julie Arbit explained. “Improving data networks that forecast and assess flooding can help track the extent of disasters in these communities.”
Current flood forecasting and measurement systems rely on data from flood gauges to determine where floods are occurring. However, limitations in federal flood gauge infrastructure make it difficult to assess risk at a household level and ensure communities are getting the aid they need.
“Data helps vulnerable communities have their voice in the public eye. Federal governments and other organizations responding to floods don’t deal with qualitative descriptions — they deal with quantitative data,” CSS data scientist Brad Bottoms explained. “ So it’s really hard for communities to get access to funds and commodities during a disaster if they can’t prove that something happened to them and that their area should be prioritized.”
Unfortunately, with a single federal gauge costing upwards of $40,000, most rivers in the U.S. have sparse federal gauge networks, averaging 550km of stream line per gauge. As such, while federal gauges can provide highly accurate flood risk assessments in areas where they are located, flood risk for areas with sparse gauge networks often goes overlooked, limiting the ability of these communities to respond and recover.
Low-cost sensors
Engineers at the Real-Time Water Systems Lab have been developing a network of low-cost flood sensors to help address disparities in flood data. With significantly lower equipment and installation costs than federal stream gauges, low-cost sensors can detect and communicate flood hazards at a household-level for only $800.
CSS data scientist Brad Bottoms looked at how these low-cost sensors can be used to improve data networks in communities where current infrastructure is sparse. Using Mt. Clemens, MI as a pilot area, he analyzed the outcome of an August 2021 flood.
Bottoms found that when low-cost sensor data was added to existing federal gauge data, flood-response models improved. Flooded roads and buildings that would have otherwise gone overlooked were able to be accounted for.
For instance, although many areas in Mt. Clemens had high water levels during the August 2021 flood, only areas with dense federal gauges received targeted flood warnings. Without sufficient data, other flooded areas had to rely on general, county-level weather statements.
“Through the use of low-cost sensors, we were able to validate that flooding occurred in the northern areas of Mt. Clemens, but the public only received a broad, county-level alert instead of a targeted warning,” Bottoms stated. “Had a more dense network been in place, the National Weather Service could have identified specific flood risk for this area.”
Applications for Environmental Justice
Going forward, the Center for Social Solutions, in collaboration with the Real-Time Water Systems Lab, hopes to continue to advocate for the use of low-cost sensors in at-risk areas to ensure more equitable coverage.
“In light of these findings, we are promoting the use of low-cost sensors to densify gauge networks. We have shown that by adding more data to the network, the accuracy of disaster response models only increases. Even if low-cost sensors don’t have the same level of confidence or precision as federal gauges, by combining all available data we can reduce overall levels of error,” Bottoms explained.
The CSS research team will also be identifying areas where social vulnerability factors intersect with sparse federal gauge networks to prioritize where sensors are placed next.
“Our team is developing a vulnerability index that will accurately depict those who suffer most from natural disasters in the country,” Arbit added. “We will not only target areas of repetitive loss, but also areas of true vulnerability that have not yet been affected by flooding.”
With flood disasters projected to worsen over the next several decades, the ability to evaluate household risk is expected to become increasingly critical. CSS researchers are looking forward to understanding more about the ways in which existing stream gauge infrastructure can be bolstered to help those most in need.
“As climate change effects intensify, new areas are being affected every day, areas that have never before seen significant flooding,” said Arbit. “We want to get ahead of these disasters and strategically place new sensors in these areas so we can protect those that are most vulnerable.”
“Equity in Flood Risk” was presented by CSS research associates Julie Arbit and Brad Bottoms at the 2021 International Association of Emergency Managers (IAEM) Conference. Full poster details below.
