Long-term Prairie Wetlands Extraction and Change Detection with Multi-spatial and Multi-temporal Remote Sensing Data
| dc.contributor.advisor | Guo, Xulin | |
| dc.contributor.advisor | Martz, Lawrence W | |
| dc.contributor.committeeMember | Pomeroy, John | |
| dc.contributor.committeeMember | Hackett, Paul | |
| dc.creator | Qiao, Ning 1987- | |
| dc.date.accessioned | 2018-04-24T22:21:12Z | |
| dc.date.available | 2018-04-24T22:21:12Z | |
| dc.date.created | 2018-04 | |
| dc.date.issued | 2018-04-24 | |
| dc.date.submitted | April 2018 | |
| dc.date.updated | 2018-04-24T22:21:12Z | |
| dc.description.abstract | Prairie wetlands, also called “potholes”, provide both ecological and hydrological functions and have experienced dramatic change over the past century. This research aims to: 1) compare the capacity of Landsat and SPOT in mapping open water and wet areas with advanced classification methods; 2) monitor and quantify the changes in wetlands and drainage channels, between 1948 and 2009, with aerial photography; and 3) evaluate Landsat’s ability to extract historical wetland coverage data across seasons using a variety of methods. Results indicate that Landsat is capable for mapping open water, wet areas and other LULC types in PPR; however only 48.5% of wetland areas are identified as compared with air photos. Historical analysis of air photo generated wetland and drainage channels show that the whole basin’s wetlands rapidly decreased from 1958 to 1990 (24% to 13%) and slowly decreased from 1990 to 2009 (13% to 10%) with the least reduction in sub basin 1. Drainage channels slowly increased from 1958 to 1990 (119 km to 269 km) and dramatically increased from 1990 to 2009 (269 km to 931km). Wetland area is highly correlated with accumulated snowfall in the previous three years in sub basin 2 (r=0.91, p<0.05) due to its memory effect to previous water conditions. For the full basin, however, there were not enough years of data to prove this correlation. Even though the minimum distance algorithm in early spring is optimal for mapping wetlands in the Prairie Pothole Region (PPR), comparing with air photos, SPOT imagery underestimated wetlands smaller than 1200 m2, while Landsat imagery is not able to detect wetlands smaller than 900 m2 and underestimates areas smaller than 1600 m2. Although free-archived Landsat can detect water bodies larger than 900 m2, its ability to detect prairie wetland is limited due to missing numerous small-scale wetlands and misclassification of seasonal wetlands. | |
| dc.format.mimetype | application/pdf | |
| dc.identifier.uri | http://hdl.handle.net/10388/8518 | |
| dc.subject | prairie wetland, classification, change detection | |
| dc.title | Long-term Prairie Wetlands Extraction and Change Detection with Multi-spatial and Multi-temporal Remote Sensing Data | |
| dc.type | Thesis | |
| dc.type.material | text | |
| thesis.degree.department | Geography and Planning | |
| thesis.degree.discipline | Geography | |
| thesis.degree.grantor | University of Saskatchewan | |
| thesis.degree.level | Masters | |
| thesis.degree.name | Master of Science (M.Sc.) |