Application of L band SAR JERS-1 data for the monitoring of hydroelectric reservoir filling process

摘要

SAR (synthetic aperture radar) data is an important tool for the study of reservoirs. An important variable for reservoir management is precise definition of the land/water boundary. The double bounce reflection present in SAR data allow one to recognize those areas present in the Amazon basin. The objective of this research is to define a methodology to determine the flooding variables (area and volume) of a hydroelectric reservoir. The flooded area by the Serra da Mesa reservoir is located over the Central Brazilian Shield covered by cerrado vegetation. This study used topographic maps at a scale of 1:25000, and SAR/JERS-1 images, from which information about the dam boundary was extracted. Several classification strategies were tested for the classification of SAR/JERS-1 images. The Serra da Mesa reservoir is located in Goias State, Central Brazil. A V-shaped valley characterizes the topography. This type of topography produces a flooded area of dentritic shape, which produces a large number of aquatic environments within the reservoir. The L band SAR image provides a good separability between the flooded forest and that not flooded. A large variation in water level is determined by the hydrologic cycle of the region. This change in water level affects the water volume and reservoir flow during the dry season.