Methodology for classification of geographical features with remote sensing images: Application to tidal flats

NOELIA REVOLLO, MARINA CIPOLLETTI, MAURICIO PERILLO, CLAUDIO DELRIEUX, GERARDO PERILLO

GEOMORPHOLOGY

ELSEVIER SCIENCE BV

Lugar: Amsterdam; Año: 2016 vol. 257 p. 10 - 22

0169-555X

Tidal flats generally exhibit ponds of diverse size, shape,orientation and origin. Studying the genesis, evolution, stability anderosive mechanisms of these geographic features is critical to understandthe dynamics of coastal wetlands. However, monitoring these locationsthrough direct access is hard and expensive, not always feasible, andenvironmentally damaging. Processing remote sensing images is a naturalalternative for the extraction of qualitative and quantitative data dueto their noninvasive nature. In this work, a robust methodology forautomatic classification of ponds and tidal creeks in tidal flats usingGoogle Earth images is proposed. The applicability of our method istested in nine zones with different morphological settings. Each zone isprocessed by a segmentation stage, where ponds and tidal creeks areidentified. Next, each geographical feature is measured and a set ofshape descriptors is calculated. This dataset, together with a-prioriclassification of each geographical feature, is used to define aregression model, which allows a massive automatic classification oflarge volumes of data discriminating ponds and tidal creeks against otherspurious geographical features. In all cases, we identified andautomatically classified different geographic features with an averageaccuracy over 90% (89.7% in the worst case, and 99.4% in the best case).These results show the feasibility of using freely available Google Earthimagery for the automatic identification and classification of complexgeographical features, as ponds and tidal creeks. Also, the presentedmethodology may be easily applied in other wetlands of the world and evenemploying any other kind of remote sensing imagery.