Multi-image ensemble classification

Ensemble classification, in which results from multiple single-scene classifications are combined through a voting procedure, is shown to improve classification accuracy over the classification based on a single best scene. In addition, ensemble classification can produce a complete land cover map in areas without cloud-free remote sensing data. The ensemble classification approach is illustrated with a case study from the shallow-water environment surrounding Inhaca Island, Mozambique (26.0 °S, 32.6 °E), for which 17 Landsat scenes, all from 2013 and containing <; 50% cloud cover, were combined in the ensemble. Freely available software implementing the described ensemble classification approach with Landsat data is under development.