Main content

The Application of Fuzzy Logic, Multiple Linear Regression and Artificial Neural Networks for Stream Assessment, Design and Monitoring

Show simple item record

dc.contributor.advisor Gharabaghi, Bahram
dc.contributor.author Gazendam, Edward
dc.date 2017-01-30
dc.date.accessioned 2017-03-15T18:43:25Z
dc.date.available 2017-03-15T18:43:25Z
dc.date.issued 2017-03-15
dc.identifier.uri http://hdl.handle.net/10214/10259
dc.description.abstract The applicability of the Qualitative Habitat Evaluation Index (QHEI) as a planning and design tool to restore streams in Southern Ontario was evaluated. QHEI assessments were made at 50 Ontario sites and correlated to %EPT, HBI and Taxa Richness. QHEI reference ranges of >67.5, 52.5 to 67.5 and <52.5 were determined for Exceptional, Good and Marginal/Poor habitats respectively. Predictive regression equations were developed for stream assessment and rehabilitation design. However, only about 50% of the variance in biologic indices was explained by geomorphic stressors within the stream. Artificial Neural Network (ANN) models were developed to integrate complex non-linear relationships between aquatic indices and key watershed-scale and reach-scale parameters. Data were collected at 112 sites on 62 stream systems located in Southern Ontario. Benthic data were collected separately for HBI and Richness. The ANN models were trained on the randomly selected 1/4 of the dataset of 112 streams in Ontario, Canada and validated on the remaining 3/4. The R2 values were 0.86 for HBI and 0.92 for Richness. Sensitivity analysis revealed that Richness was directly proportional to Erosion and Riparian Width and inversely proportional to Floodplain Quality and Substrate parameters. HBI was directly proportional to Velocity Types and Erosion and inversely proportional to Substrate, % Treed and 1:2 Year Flood Flow parameters. Finally, the Ontario Channel Susceptibility Methodology (OCSM) provides quantitative assessment of the factors influencing channel-habitat quality and the sensitivity of stream channels to deterioration when subject to increased stormwater flows. A risk-based, Fuzzy Logic methodology integrated a broad spectrum of in-channel, near-channel, and watershed-surface data sets to rank stream susceptibility. The methodology was applied to Grindstone Creek and Highland Creek, two channel systems with varying susceptibility to the effects of hydrology, water quality and physical conditions. Results show that Highland Creek has a greater risk of deterioration than Grindstone Creek when subject to enhanced stormwater flows. OCSM was determined to be an effective tool for evaluating and prioritizing stream channels. Further data collection and research is required to ensure that appropriate QHEI design ranges and ANN modelling is completed for higher resolution biogeographical areas. en_US
dc.description.sponsorship NSERC OMOEECC OMAFRA en_US
dc.language.iso en en_US
dc.subject QHEI en_US
dc.subject Artificial Neural Network en_US
dc.subject Multiple Linear Regression en_US
dc.subject Fuzzy Logic en_US
dc.subject Stream Assessment en_US
dc.subject Stream Restoration en_US
dc.title The Application of Fuzzy Logic, Multiple Linear Regression and Artificial Neural Networks for Stream Assessment, Design and Monitoring en_US
dc.type Thesis en_US
dc.degree.programme Engineering en_US
dc.degree.name Doctor of Philosophy en_US
dc.degree.department School of Engineering en_US


Files in this item

Files Size Format View Description
Gazendam_Edward_201702_PhD.pdf 60.83Mb PDF View/Open Thesis

This item appears in the following Collection(s)

Show simple item record