Design-based validation of the MODIS Global Burned Area Product
Abstract
The potential research, policy and management applications of satellite products place a high priority on providing statements about their accuracy. Product comparison with independent reference data is needed to determine product accuracy. Design-based accuracy assessment methods select the independent reference data using a probability sampling and are preferred as the reference data can subsequently be used to compute accuracy metrics that explicitly take into account the sampling probability. To date, no global burned area product has been validated using a design-based methodology due to the technical challenges of designing an appropriate independent reference data sampling strategy, and, until recently, due to the high cost of independent reference data collection. Burned area product errors are not randomly distributed in space and time and tend to be linked to seasonal phenomena that can be confused with burning. This paper describes the design-based sampling strategy developed for the validation of the MODIS Global Burned Area Products using more than 200 globally distributed multi-temporal pairs of Landsat scenes, processed following the recommendations of the CEOS Cal/Val Burned Area Validation Protocol. The sampling design is illustrated and the advantages of the design-based sampling discussed.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2012
- Bibcode:
- 2012AGUFMIN52B..06B
- Keywords:
-
- 1640 GLOBAL CHANGE / Remote sensing;
- 1950 INFORMATICS / Metadata: Quality;
- 1986 INFORMATICS / Statistical methods: Inferential