C4.5: programs for machine learning
C4.5: programs for machine learning
Data Mining: Concepts and Techniques
Data Mining: Concepts and Techniques
Object-based change detection using correlation image analysis and image segmentation
International Journal of Remote Sensing
The WEKA data mining software: an update
ACM SIGKDD Explorations Newsletter
| Hi-index | 0.00 |
Reconstructing the evolution of large tropical fluvial systems over the geological time is challenging, particularly in areas such as the Amazonian lowlands where basic geological and geomorphological data are still scarce relatively to the large dimension of the region. In such areas, remote sensing data are useful for detecting ancient morphological features that may reveal past fluvial dynamics. In this study, we explored object-based image analysis (OBIA) in the Madeira-Purus interfluve, Southwestern Brazilian Amazonia, integrating geospatial data including Landsat satellite multispectral images, the digital elevation model (DEM) acquired during the Shuttle Radar Topography Mission (SRTM), and stream channels digitized from topographic maps. This approach provided the basis to categorize automatically classes with contrasting vegetation and/or topographic characteristics within the dense tropical forest over an extensive and relatively flat forested area. The main goal was to use these classes as a surrogate for the recognition of ancient geomorphic features consisting mainly of paleochannels that may help reconstructing fluvial history in space and time. Landsat optical images with stream vector were appropriate to classify open vegetation areas that grow over paleochannels, but failed to identify these objects when they were located over forested areas. However, the digital elevation model (DEM) derived from the Shuttle Radar Topography Mission (SRTM) was successful to detect these objects even in forested areas. Topographic survey undertaken in the field increased the classification reliability by demonstrating true terrain variations along transects measured across the paleochannels. Based on this technique, networks of dendritic paleochannels were mapped and related to ancient tributaries of the Madeira River that had their courses flowing opposite to main modern streams. This denotes a significant change in fluvial dynamics over time, most likely resulting from tectonic tilting.