Effects of temporal variability in ground data collection on classification accuracy
Links
- More information: Publisher Index Page (via DOI)
- Download citation as: RIS | Dublin Core
Abstract
This research tested whether the timing of ground data collection can significantly impact the accuracy of land cover classification. Ft. Riley Military Reservation, Kansas, USA was used to test this hypothesis. The U.S. Army's Land Condition Trend Analysis (LCTA) data annually collected at military bases was used to ground truth disturbance patterns. Ground data collected over an entire growing season and data collected one year after the imagery had a kappa statistic of 0.33. When using ground data from only within two weeks of image acquisition the kappa statistic improved to 0.55. Potential sources of this discrepancy are identified. These data demonstrate that there can be significant amounts of land cover change within a narrow time window on military reservations. To accurately conduct land cover classification at military reservations, ground data need to be collected in as narrow a window of time as possible and be closely synchronized with the date of the satellite imagery.
| Publication type | Article |
|---|---|
| Publication Subtype | Journal Article |
| Title | Effects of temporal variability in ground data collection on classification accuracy |
| Series title | Geocarto International |
| DOI | 10.1080/10106049908542123 |
| Volume | 14 |
| Issue | 4 |
| Year Published | 1999 |
| Language | English |
| Publisher | Taylor & Francis |
| Description | 8 p. |
| First page | 7 |
| Last page | 14 |
| Google Analytic Metrics | Metrics page |