CGI Federal's flood forecasting solution

Flood events can impact every state in the country and have affected over 40 million people in the U.S. over the last decade. Some coastal communities experience chronic flooding. In Lafayette, Louisiana, flood occurs weekly.

The need for real-time predictions and forecasting is critical and why CGI Federal’s internal innovation program took on the challenge and developed a machine learning flood forecasting prototype in partnership with the University of Louisiana at Lafayette.

Their prototype [e focuses on using machine learning and artificial intelligence techniques to solve complex problems in hydrologic modeling. The work has created the building blocks for an AI-based flood prediction and modeling platform.

Advances in data science, machine learning and AI, along with the growth in environmental data from non-traditional sources has opened the door for innovative approaches, the company said.

Working with the university and a partnership with the Lafayette Consolidated Government, CGI is building a community-wide, multi-faceted digital transformation effort that is unique in its diversity and impact to education, workforce development, and government provided services to citizens and businesses.

This includes CGI’s commitment to create 1,000 digital media and transformation jobs in the Lafayette community, state funding to triple the number of IT graduations from the University of Louisiana at Laffayette, and the creation of a STEM camp for at risk middle school students that teaches computer programing and how to build computers.

CGI is also working with the university create Agile capstone and internship programs for graduating seniors.

The flood forecasting and warning system integrates data from space-based sensors, ground radar, field sensors, and social media data to feed a predictive analytics model. For instance, it predicts soil moisture by using a Deep Learning model that analyzes precipitation, temperature, vegetation, plant canopy surface water, and wave radiations.

While the work is concentrated on flood events, the same approach can be used for other applications, according to the company.