Application of Artificial Neural Network Models for Flow Estimation in a Large Constructed Wetlands — Stormwater Treatment Area 2 in South Florida

The South Florida Water Management District (District) is responsible for managing water resources in 16-counties over a 46,439-square kilometer (17,930 square-mile) area. The District's area extends from Orlando to Key West and from the Gulf of Mexico to the Atlantic Ocean and contains the country's second largest freshwater lake, Lake Okeechobee, and the world-famous Everglades wetlands. The District manages water in one of the most diverse and complex ecosystems in the world — the interconnected Kissimmee-Okeechobee-Everglades system. The District's routine work includes buying and managing land, restoring floodplains, revitalizing shoreline habitats, and protecting wetlands. The District's mission is to provide regional flood control, water supply, and water quality protection as well as ecosystem restoration. During last ten years, the District has constructed and operates six Stormwater Treatment Areas that are large constructed wetlands to reduce the nutrients in water. Specially, six Stormwater Treatment Areas are designed to reduce the level of phosphorus entering the Everglades. Stormwater Treatment Area 2 (STA-2) is located in southern Palm Beach County and captures stormwater runoff originating from the S-6 watershed, upstream of the S-6 pump station. The STA is populated with sawgrass and cattail to reduce phosphorus levels, before it is released into Water Conservation Area 2A. STA-2 has many water control structures including pumps and gated culverts and spillways. These water control structures have complex hydraulic conditions that require a set of complex physics-based flow estimating equations. The Artificial Neural Network (ANN) models were developed, tested and validated for flow estimation purpose at the water control structures. Their application in estimating flow and missing tail water elevations at water control structures culverts G331 and G333, and spillway G332 were successfully demonstrated in a case study.