Can everything that involves hard data and decision-making be automated? Probably not everything, but things that are otherwise time consuming and subjective certainly should be considered for automation. Dam site evaluation is certainly one of those tasks.
The process of taking a dam from potential to reality can take 10 years or more, and involves a significant desktop study before anyone’s boots even hit the ground. These desktop studies are generally undertaken manually and involve a pre-feasibility evaluation of all available elevation (DTM and DEM), satellite imagery and aerial photos. The manual process is both time consuming and rarely applied using set standards.
Semi-automated techniques, such as composite indexing – overlaying relevant GIS data, such as topography and hydrologic regime, to produce a summed, ranked layer, can speed up the process. The result is a map of high-graded areas that have suitable social need, appropriate hydrologic regime, and sufficient topographic relief for a dam and reservoir. However, in the past these studies tended to be adhoc and completed using a variety of inconsistent methods, making them impossible to compare.
Cuan Petheram and colleagues from the CSIRO recognised that the aforementioned techniques are useful for capturing broad-scale considerations that determine dam suitability for a region, but realised that they still can’t model the factors that will directly contribute to the performance of an individual large dam. Yield and reliability are central to the feasibility equation, but to do that, the precise location of a potential dam needs to be designated first.
To address the knowledge gap, Dr Petheram and colleagues have created DamSite. DamSite is a program that automates both the identification of all suitable dam locations within a region, and then quickly calculates reservoir dimensions for individual dam sites.
The program uses a series of algorithms that identify and rank potential dam wall locations across a landscape. Rankings are based on calculations of reservoir dimensions to predict yield, and reliability within a landscape. The result is a rigorous and consistently applied model to evaluate individual dam sites for suitability and deliverability. Using the wall locations, a digital elevation model, and hydrologic regime, the yield and reliability can be calculated.
The model was run across 3 million km2 of northern Australia in support of the recent Northern Australia White Paper (Northern rivers and dams: A preliminary assessment of surface water storage potential for northern Australia). More than 2 billion potential dam sites were recognised and assessed.
Petheram notes they have recently made some big improvements since the white paper, including using Landsat imagery to create a raster of river thickness, from which they can then calculate spillway width. Using this information helps them estimate the flood rise for a significant flood event, which is an important consideration in costing a dam. These improvements will also soon be submitted for publication. Those interested in using the DamSite model should contact Cuan Petheram (Cuan.Petheram@csiro.au).
To read the peer-reviewed paper, click here.