Designing green stormwater infrastructure for hydrologic and human benefits: an image based machine learning approach

Urbanization over the last century has degraded our natural water resources by increasing storm-water runoff, reducing nutrient retention, and creating poor ecosystem health downstream. The loss of tree canopy and expansion of impervious area and storm sewer systems have significantly decreased infiltration and evapotranspiration, increased stream-flow velocities, and increased flood risk. These problems have brought increasing attention to catchment-wide implementation of green infrastructure (e.g., decentralized green storm water management practices such as bioswales, rain gardens, permeable pavements, tree box filters, cisterns, urban wetlands, urban forests, stream buffers, and green roofs) to replace or supplement conventional storm water management practices and create more sustainable urban water systems. Current green infrastructure (GI) practice aims at mitigating the negative effects of urbanization by restoring pre-development hydrology and ultimately addressing water quality issues at an urban catchment scale. The benefits of green infrastructure extend well beyond local storm water management, as urban green spaces are also major contributors to human health. Considerable research in the psychological sciences have shown significant human health benefits from appropriately designed green spaces, yet impacts on human wellbeing have not yet been formally considered in GI design frameworks. This research develops a novel computational green infrastructure (GI) design framework that integrates storm water management requirements with criteria for human wellbeing. A supervised machine learning model is created to identify specific patterns in urban green spaces that promote human wellbeing; the model is linked to RHESSYS hydrological model to evaluate GI designs in terms of both water resource and human health benefits. An application of the models to Dead Run Watershed in Baltimore showed that image mining methods were able to capture key elements of human preferences that could improve tree-based GI design. Hydrologic benefits associated with these features were substantial, indicating that increased urban tree coverage and a more integrated GI design approach can significantly increase both human and hydrologic benefits.

Name thesis:degree_name

M.S.

Level thesis:degree_level

Thesis

Discipline thesis:degree_discipline

Civil Engineering

Grantor

University of Illinois at Urbana-Champaign

Year dc:date

2013

Author dc:creator

• Rai, Ankit

Contributors dc:contributor

• Minsker, Barbara S.

• Machine Learning
• Human Preferences
• Preference matrix
• Hydrologic Benefits
• Regional Hydro-Ecologic Simulation System (RHESSYS)

Statement dc:rights

• Copyright 2013 Ankit Rai

Language dc:language

en

Handle dc:identifier

http://hdl.handle.net/2142/45469

OAI identifier oai:identifier

oai:www.ideals.illinois.edu:2142/45469