The use of remote-sensing techniques to monitor dense reservoir networks in the Brazilian semiarid region

Abstract

Reservoirs are the main water source in the Brazilian semiarid region, especially in the crystalline-geology watersheds, forming high-density reservoir networks (HdRN). However, in most cases, the construction of these reservoirs has been done without technical supervision. The objective of this work was to map and evaluate the spatial distribution of the 25,000 km2 Orós Reservoir Basin (ORB) HdRN, in semiarid Brazil, with the help of remote-sensing tools associated with geographic information systems (GIS). Using LANDSAT 5 images of the end of the 2011 rainy season of the ORB, the remote-sensing technique allowed the identification of 6002 polygons, which corresponded to only 4717 reservoirs (implying a misidentification of 21%). Between 2002 and 2011, a 17.5% increase (and 1.81% annual increase) in the number of reservoirs in the basin was observed, still lower than the annual increment from 1970 to 2002, when an average increase of 2.64% per year was observed, in other studies. The perimeter of the reservoirs ranged from 0.250 to 560 km and the individual water surface area ranged from 0.004 to 195.0 km2, resulting in a total surface of 465.0 km2. Analysing the surface area of the strategic reservoirs, results showed that the estimation of the surface area (from remote sensing with manual polygon adjustment) yielded values very close to those of the on-site monitored areas, with R2 = 0.99 and normalized difference index ranging from −0.02 to +0.09. The reservoir density in the ORB in 2011 was 0.19 reservoirs km–2, higher than the recommended optimum density of 0.15 reservoirs km–2 basin. Analysis of reservoir density by municipality recorded values ranging from 0.02 to 0.40 reservoirs km–2. The sedimentary-geology municipalities presented a reservoir density on average 80% lower than the that of the crystalline-geology municipalities, indicating a strong relationship between geology and reservoir density. Neither population density nor rainfall explained the spatial distribution of reservoirs within the basin, both yielding R2 lower than 0.1. This remote-sensing survey of reservoirs demonstrated two major flaws: the misidentification of shadows as reservoirs and the inability to identify the presence of macrophytes, which negatively affected the number and surface area of the target reservoirs. Despite these problems, remote sensing has been shown to be a technique of great potential in the planning and management of water resources in regions with dense reservoir networks.