Abstract
Lake Manzala is considered the largest lagoon in Egypt. The lake has significant economic and environmental impacts that should be managed based on management models. The bathymetry of this lake, which is time consuming and laborious to obtain in the field, is one of the key input files for developing hydrological models of the lake. Remote sensing technology is used to determine satellite bathymetric maps with reasonable accuracy. The objective of the present work was to utilize Landsat 8 satellite imagery to determine the bathymetry of Lake Manzala. Generalized linear model (GLM), artificial neural network (ANN), decision tree, bagging (BAG, an ensemble regression algorithm), least-squares boosting fitting ensemble (LSB), and support vector machine (SVM) approaches were used in this study to process the images and manage the database of each image. The Landsat images were corrected for atmospheric conditions and the sunglint effect. Then values from the logarithms of corrected reflectance bands (coastal, blue, green, and red) and their ratio logarithms at locations corresponding to GPS surveys were extracted. Two assessing metrics, root mean square error (RMSE) and correlation (R), were used to calibrate the derived logarithm values for the model using observed data for the lake. The results show that the BAG and decision tree approaches perform well for Lake Manzala. Such methodologies should be applied for bathymetry determination, especially for shallow lakes, to save monitoring effort and costs. This approach facilitates the development of management models for lakes.