Author(s): Luisa Da Cunha Vieira, Talita Silva, Philippe Maillard
Linked Author(s): Talita Silva
Keywords: Serra Azul Reservoir; Landsat-5; Landsat-7;mono channel algorithm; Surface water temperature;
Abstract: Surface Water Temperature (SWT) is an important driver of physical and biochemical processes in lentic ecosystems. SWT monitoring allows the assessment of changes in the ecological functioning of lakes and reservoirs, including those related to climate change. Remote sensing offers the potential to complement in situ monitoring of inland water bodies. This article presents a study of how Landsat images were used to monitor SWT in the Serra Azul Reservoir in Southeast Brazil. Twenty seven images from Landsat-5 and Landsat-7 were selected from 1984 to 2002 by combining them with in situ monitoring dates and clear sky conditions. Atmospheric correction using a mono channel calibration algorithm was performed to obtain SWT from the thermal band of these satellite images. The advantage presented by this algorithm is that it uses solely the atmospheric water vapor content as input data. SWT estimated from Landsat images was compared to measured data from four monitoring stations in the Serra Azul Reservoir. Good results were obtained (Pearson's correlation coefficient of 0.85 and Root Mean Square Error of 2.22°C) which were compatible with other studies. Surface temperature maps were generated showing SWT spatial variability. Errors between measured and estimated SWT increase with the water vapor content before SWT correction by mono channel algorithm. This bias was eliminated after the atmospheric correction. Based on the mono channel algorithm validation, 156 new images from 1984 to 2002 were selected to generate temporal series of SWT. A seasonal cycle was observed and no temporal trend was detected.