Wetland mapping in the Congo Basin using optical and radar remotely sensed data and derived topographical indices

This study reports results of an approach to mapping the wetlands of the Congo Basin, focusing on the Cuvette Centrale or Central Basin of the Congo River watershed. Challenges to mapping the wetlands of the Congo Basin include data limitations, vegetative structural heterogeneity, seasonal variability, and limited opportunities for ground-based training and validation data collection. To overcome these limitations, regional expert knowledge was used to train multisource passive optical and active radar remote sensing data and topographical indices using a classification tree algorithm. The final product is a per pixel probability map of wetland cover at a spatial resolution of 57 meters. Thresholding the probability map at 50% yields a proportion of wetland cover for the study area of 43%. The method used inputs that are freely available and employed a fully automated process, except for training data collection. Comparisons to existing maps and in situ field observations indicate an improved map characterization compared to past efforts. Local topography depicting relative elevation for sub-catchments proved to be the most valuable discriminator of wetland cover. However, all sources of information featured prominently in contributing to the classification tree procedure, reinforcing the idea that multi-source data are important in the characterization of wetland land cover.

BWANGOY Jean-Robert;  HANSEN Matthew;  ROY David;  DE GRANDI Gianfranco;  JUSTICE Chris; 

2013-10-09

ELSEVIER SCIENCE INC

JRC47003

0034-4257,   

http://www.sciencedirect.com/science/article/pii/S0034425709002533,    https://publications.jrc.ec.europa.eu/repository/handle/JRC47003,   

10.1016/j.rse.2009.08.004,