Modeling hydrological processes using hybrid modeling approach (Completed)
The hybrid modeling approach is defined here as a modeling system that includes or integrates multiple (at least two) models of different types (e.g., the mechanistic SDW model with ANN model; GP model with the SDW model; ANN with GP models). The models can be mechanistic and physically based or conceptual models, where complete quantitative knowledge of the interactions among processes is lacking and some reliance on observations is required to calibrate and finalize the structure of the model, or data-driven models.
The global view of this research includes (i) developing a hybrid modeling approach applicable to simulate individual hydrological processes as well as watersheds; (ii) evaluating the hybrid models in terms of simulation (prediction) accuracy, output uncertainty, and computational efficiency and tractability; (iii) recalibrating and validating the hybrid models with coarser resolution hydrometeorological input data, such as the North American Regional Renalysis (NARR) data available on 32 x 32 km grids for the period of 1979-2007. This task will allow the effect of coarse resolution data on the output uncertainty of the hybrid models to be addressed, and thus minimize the need for on site monitoring of particular parameters. It will also allow for testing the impacts of future climate change scenarios on the studied watersheds and hydrological processes through using either the outputs of Regional Climate Models (RCMs) or the downscaled outputs of the Global Circulation Models (GCMs).
The proposed modeling approach and the resulting hybrid hydrologic models will be validated through a primary application as well as some other secondary applications conditioned on time and resource availability. The primary application will focus on modeling the restored (reclaimed) watersheds in northern Alberta, Canada. The watersheds are reclaimed after disturbance due to oil sands mining. This research program steps beyond previous hydrological research in the oil sands in that it will integrate the extensive collection of hydrological and meteorological data into a tool which will allow mine and closure planners to assess and quantify the risk of hydrological failure of various soil cover and reclamation options. This tool will be incorporated into, or will be used to improve the existing deterministic guidelines for landform and soil cover (restored watersheds) design in the Athabasca Oil Sands Region, which evaluate soil hydrologic functions from a static perspective. By working with CEMA, a consortium of oil sands operators and stakeholders, the maximum possible dissemination of this work will occur. The results will also provide new insight into the ecohydrology of restoration covers, and how vegetation response to climate forcing varies between natural and restored sites. In addition to the scientific validation for the proposed modeling approach, this primary application will allow for serious evaluation of the modeling exercise in light of its economic impact on the reclamation practices of the oil sands industry, which in turn impacts the regional economy and the environment. The secondary applications will include the flood (runoff) prediction on the Red River basin (Manitoba), as a relatively large watershed, and the runoff in the Corman Park rural municipality (Saskatchewan) as a pilot area chosen by the PFRA to study the impacts of extreme events on the watershed infrastructure. The applicant is currently the Hydrology Advisor for the Corman Park pilot study.