Abstract: Seasonal Delineation of Flooded Areas for Management of Hydroelectric Power Plants and Pipelines in the Brazilian Amazonia Using Spaceborne Radar Data

Miranda, F. P. - Petrobras/Cenpes; C. H. Beisl - Gorceix Foundation

Spaceborne radars provide unique information about flooding in tropical rainforest regions. This is relevant for management of pipelines and hydroelectric power plants. Two pilot projects in the Brazilian Amazonia are described here: Balbina Dam and Rio Urucu-Coari Pipeline. These focus on JERS-1 SAR and RADARSAT-1 data application for monitoring inundation phenomena.

The Balbina Dam was built in 1987 on the Uatumã River near Manaus , the capital of Amazonas State. Seasonal distribution of precipitation accounts for the 9.0-meter difference between the lowest and highest recorded values of water level. JERS-1 SAR and RADARSAT-1 (Standard 3 beam mode) data were obtained in the high water period (26 May, 1996). The Rio Urucu-Coari Pipeline is 280 km long; construction started in February 1997. In the Urucu River and Lake Coari, water level changes between dry and wet seasons attain 13.0 meters. JERS-1 SAR images were acquired in both high (01 to 06 June, 1996) and low (25 to 28 October, 1995) water periods; RADARSAT-1 data (Standard 4 and 5 beam modes) were obtained in the dry season (13 October to 13 November, 1996).

The unsupervised semi-variogram textural classifier (USTC) was selected as the classification algorithm for flood mapping using JERS-1 SAR and RADARSAT-1 data. It is a deterministic classifier which provides the option of combining both textural and radiometric information.

At the Balbina Dam, a digital elevation model (DEM) was built from 1:50,000 scale topographic maps made in 1984 based on aerial photographs. Such a raster file was then co-registered to the spaceborne radar images and used to estimate the surface extent of flooding taking into account stage height readings at the time of JERS-1 SAR and RADARSAT-1 acquisition. Results have shown that USTC is an effective tool to map the upland/flooded forest boundary in regions of low topographic gradients in the vicinity of the dam. In these regions, DEM-based flood area estimation is poor due to errors in altimetry induced by aerial photographs, which record the upper level of the canopy instead of the ground underneath

Application of USTC to JERS-1 SAR data acquired in the Urucu River and Lake Coari highlighted flooded forest areas along the pipeline which are extremely sensitive to oil spills. USTC-classified RADARSAT-1 images were instrumental in the detailed delineation of open water bodies. Seasonal inundation maps obtained by means of semi-variogram textural classification of spaceborne radar data are therefore useful value added products for pipeline site planning, operational support, and infrastructure maintenance in the Brazilian Amazonia.

AAPG Search and Discovery Article #90933©1998 ABGP/AAPG International Conference and Exhibition, Rio de Janeiro, Brazil