A Comparison of Reclassification Methods to Improve an NDVI Based Flood Map
| dc.contributor.advisor | Fuhrmann, Sven | |
| dc.contributor.author | Fayne, Jessica V | |
| dc.creator | Fayne, Jessica V | |
| dc.date | 2015-05-01 | |
| dc.date.accessioned | 2015-08-06T18:51:48Z | |
| dc.date.available | 2016-05-01T08:23:55Z | |
| dc.date.issued | 2015-08-06 | |
| dc.description | The work was embargoed by the author and will not be publicly available until May 2016. | |
| dc.description.abstract | In Cambodia and Vietnam, low-lying terrain is particularly susceptible to flooding during the monsoon season between May and November. To monitor flooding in the region, a near-real time NDVI percent decrease based Flood Extent Product was developed to be hosted on an online Flood Dashboard by the NASA DEVELOP team. The product was designed to be updated twice per day with 250-meter resolution from MODIS on the Aqua and Terra satellites. To increase the usage and usability of this product, the classification intervals were compared with other commonly used classification schemes to monitor flooding. The use of substantiated flood classification methods is essential to ensure understanding and usefulness of mapped flood products. Classification schemes can influence the usability and usefulness of these products, e.g. inappropriate flood mapping classification intervals and color selections may incorrectly classify flooded areas and distract from the interpretation of the phenomenon of interest. The percent change method proved to be very helpful in delineating flood boundaries compared to standard deviation and differencing methods. However, only the 100% decrease interval class had the highest accuracy ratings compared to three reference data sets, with an average producer’s accuracy of 67.8% and an average user’s accuracy of 74%. The results of the accuracy assessments indicate that only the 100% interval class can be reclassified to into a descriptive ‘flood’ classification. The use of an additional ‘wet’ category with 75% decrease will be useful to support the flooded area description and allow users to monitor changes in regions that are not currently flooded, but are more susceptible to flooding. The use of a descriptive two-class product eliminates confusion from understanding input data while removing extra information from lower interval change classes. | |
| dc.identifier.uri | https://hdl.handle.net/1920/9713 | |
| dc.language.iso | en | |
| dc.subject | NDVI | |
| dc.subject | MODIS | |
| dc.subject | Flood mapping | |
| dc.subject | Aqua Terra | |
| dc.subject | Water Mask | |
| dc.title | A Comparison of Reclassification Methods to Improve an NDVI Based Flood Map | |
| dc.type | Thesis | |
| thesis.degree.discipline | Geographic and Cartographic Sciences | |
| thesis.degree.grantor | George Mason University | |
| thesis.degree.level | Master's | |
| thesis.degree.name | Master of Science in Geographic and Cartographic Sciences |