Parametrization and calibration of a quasi-analytical algorithm for tropical eutrophic waters

• Autor(a) principal: Watanabe, Fernanda [UNESP]
• Data de Publicação: 2016
• Outros Autores: Mishra, Deepak R., Astuti, Ike, Rodrigues, Thanan [UNESP], Alcantara, Enner [UNESP], Imai, Nilton N. [UNESP], Barbosa, Claudio
• Tipo de documento: Artigo
• Idioma: eng
• Título da fonte: Repositório Institucional da UNESP
• DOI: 10.1016/j.isprsjprs.2016.08.009
• Texto Completo: http://dx.doi.org/10.1016/j.isprsjprs.2016.08.009; http://hdl.handle.net/11449/162154
• topic: Quasi-analytical algorithm; Inland waters; Algal bloom; Bio-optical model; Remote sensing reflectance; Inherent optical properties

Quasi-analytical algorithm (QAA) was designed to derive the inherent optical properties (IOPs) of water bodies from above-surface remote sensing reflectance (R-rs). Several variants of QAA have been developed for environments with different bio-optical characteristics. However, most variants of QAA suffer from moderate to high negative IOP prediction when applied to tropical eutrophic waters. This research is aimed at parametrizing a QAA for tropical eutrophic water dominated by cyanobacteria. The alterations proposed in the algorithm yielded accurate absorption coefficients and chlorophyll-a (Chl-a) concentration. The main changes accomplished were the selection of wavelengths representative of the optically relevant constituents (ORCs) and calibration of values directly associated with the pigments and detritus plus colored dissolved organic material (CDM) absorption coefficients. The re-parametrized QAA eliminated the retrieval of negative values, commonly identified in other variants of QAA. The calibrated model generated a normalized root mean square error (NRMSE) of 21.88% and a mean absolute percentage error (MAPE) of 28.27% for a(t)(lambda), where the largest errors were found at 412 nm and 620 nm. Estimated NRMSE for a(CDM)(lambda) was 18.86% with a MAPE of 31.17%. A NRMSE of 22.94% and a MAPE of 60.08% were obtained for a(phi)(lambda). Estimated a(phi)(665) and a(phi)(709) was used to predict Chl-a concentration. a(phi)(665) derived from QAA for Barra Bonita Hydroelectric Reservoir (QAA_BBHR) was able to predict Chl-a accurately, with a NRMSE of 11.3% and MAPE of 38.5%. The performance of the Chl-a model was comparable to some of the most widely used empirical algorithms such as 2-band, 3-band, and the normalized difference chlorophyll index (NDCI). The new QAA was parametrized based on the band configuration of MEdium Resolution Imaging Spectrometer (MERIS), Sentinel-2A and 3A and can be readily scaled-up for spatiotemporal monitoring of IOPs in tropical waters. (C) 2016 International Society for Photogrammetry and Remote Sensing, Inc. (ISPRS). Published by Elsevier B.V. All rights reserved.