Turbidity in the Amazon Floodplain Assessed Through a Spatial Regression Model Applied to Fraction Images Derived From MODIS/Terra

Abstract

The objective of this paper was to estimate turbidity in the Curuai floodplain during the high water level period. Spatial regression models were developed by using fraction images derived from a linear spectral mixture model applied to a Moderate Resolution Imaging Spectroradiometer/Terra image and turbidity <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">in</i> <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">situ</i> data. As the turbidity in situ data showed spatial autocorrelation, they were divided into four spatial regimes (clusters). Thus, a spatial regression model was developed for each spatial regime. Through the Akaike information criterion, it was verified which spatial regime showed the best fit in the spatial regression model. The best fit was presented by the spatial regime 4 ( <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">R</i> <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> = 0.80,p < 0.05). Then, the spatial regression model developed for the spatial regime 4 was applied to all floodplain lakes. The spatial regression models show potential for assessing the water turbidity in aquatic systems by considering a spatial dependence between samples.