Landsat-based remote sensing of the colored dissolved organic matter absorption coefficient in a tropical oligotrophic reservoir

Abstract

Colored dissolved organic matter (CDOM) is a photoactive fraction of organic carbon that can be detected via remotely sensed data (proximal and satellite). The retrieval of the absorption coefficient of CDOM (aCDOM) in reservoirs is important because it is related to the carbon budget and is a possible source of greenhouse gas that is emitted from aquatic ecosystems. In this context, we divided this study into three steps: (1) evaluating four retrieval algorithms for aquatic CDOM proposed for different types of waters; (2) recalibrating the algorithms using field data; and (3) adjusting new algorithms based on Landsat 5 TM bands. This research aimed to evaluate the use of Landsat 5 TM images for retrieving aCDOM in oligotrophic inland waters. The best performing model in the literature exhibited a normalized root mean square error (NRMSE) value of 29.5% and a mean absolute percentage error (MAPE) value of 25.4%, showing that some models proposed for marine water may be applied to specific inland waters. After recalibration, the best result was NRMSE = 29.9% and MAPE = 24.8%. The Landsat 5 TM bands yielded accurate results (NRMSE = 26.2% and MAPE = 33.6%), which indicated the potential for using Landsat data to monitor the bio-optical properties in inland waters.