Quantification of flooded areas of Pantanal by sub-pixel classification of modis time-series data.

Floods in the Pantanal affect the fish production and influence the dynamics of vegetation, also changing the meat production. The understanding of floods dynamics is crucial to infer the level of flooding, once it promotes changes in the whole plain. The understanding of floods dynamics is crucial to infer the level of flooding. MODIS (Moderate Resolution Imaging Spectroradiometer) images provide wide coverage of the Earthís surface with high temporal resolution, which are important features for flood monitoring. However, its moderate spatial resolution may cause the spectral mixing of different land cover classes within a single pixel. In this context, the objective of this study was to apply a methodology for sub-pixel classification using MODIS time-series data, in order to quantify the flooded areas in the Pantanal. Data from the mid-infrared channel of MODIS sensor allowed the monitoring of flood prone areas in the Pantanal during the 2008/2009 and 2007/2008 hydrological years. The drought and flood periods are quite variable, occurring from North to South and from East to West. The sub-pixel classification models, generated from Fuzzy ARTMAP neural network, demonstrated excellent suitability for the mapping and quantification of flooded areas of the Pantanal based on the Commitment measure.

Bibliographic Details

Main Authors:	ANTUNES, J. F. G., ESQUERDO, J. C. D. M.
Other Authors:	JOÃO FRANCISCO GONÇALVES ANTUNES, CNPTIA; JÚLIO CÉSAR DALLA MORA ESQUERDO, CNPTIA.
Format:	Artigo de periódico biblioteca
Language:	English eng
Published:	2016-02-02
Subjects:	Áreas úmidas, Processamento de imagem, Reconhecimento de padrões, Redes neurais, Lógica difusa, Redes neuro-fuzzy, Pattern recognition, Neuro-fuzzy networks, Sensoriamento remoto, Remote sensing, Image analysis, Wetlands, Fuzzy logic, Neural networks,
Online Access:	http://www.alice.cnptia.embrapa.br/alice/handle/doc/1035883
Tags:	Add Tag No Tags, Be the first to tag this record!