Upscaling Fill-and-Spill Hydrologic Processes

Abstract

Low-gradient landscapes found in parts of the Taiga Plains and the North American Prairies can be dominated by many depressional wetlands with variable storage capacity. Runoff from these regions is influenced by the local storage capacity of individual wetlands and water exchange between the wetlands. Fill-and-spill conceptual models have been proposed to consider the connectivity-controlled process in wetland dominated catchments. Although fill-and-spill phenomenon has been locally observed, few studies examine the response of a landscape to thousands of cascading wetlands, as is seen in a number of Canadian landscapes. Being able to characterize, understand, and parameterize this response in hydrological models may enable successful simulation of the contribution area and runoff response in wetland-dominated regions. Current probabilistic fill-and-spill models consider individual features rather than the cumulative connections between adjacent wetlands in a cascade. The lack of understanding of the regional effects of wetland distributional characteristics on landscape hydrology, combined with insufficiently resolved elevation data, particularly in flat terrains, are two concerns that signify the need for an improved probabilistic runoff model. We propose an upscaled wetland fill-and-spill (UWFS) algorithm to investigate the response of large-scale wetland systems in low gradient areas to rainfall or snowmelt events. The research addressed in this thesis consists of the following:

1. An explicit probabilistic-analytic model is developed and tested for cascades of wetlands, providing an upscaling approach to understand and characterize system responses. To do this, first, a probabilistic analytic model is developed based on the fill-and-spill conceptualization, which considers each wetland in the basin as a member of an ensemble. The mathematical solution requires information about the initial deficit distribution and distribution of wetland local contributing areas which may be estimated via a combination of spatial analysis and field observation. Then, by using the derived distribution approach, the response of a landscape with a single wetland cascade is upscaled to the response of a landscape with thousands of wetlands. This event model is extended to evaluate the continuous response of a heterogeneous wetland complex to rainfall and snowmelt events by evolving the deficit distribution based on evaporation and precipitation. 2. A Monte Carlo based approach is proposed here that samples from initial deficit and concentrating factor distributions and finds the generated runoff from water balance equation applied to wetland cascade networks. This model along with the analytical model enables us to explore the impacts of network depth, branching, and gatekeeping on fill-and-spill runoff responses from complex wetland networks. The accuracy of the probabilistic analytical solution is also assessed by comparing the results with those from the Monte Carlo approach. 3. The proposed probabilistic analytical runoff model has been implemented into an existing two-dimensional semi-distributed hydrologic model, Raven, to test the ability of the upscaling method in lumped runoff simulation of wetland-dominated basins influenced by fill-and-spill hydrology. The model has been tested at 10 subbasins inside the Qu’Appelle River Basin in Prairie and the simulation results has been compared to an existing Prairie model named HYdrological model for Prairie Region (HYPR). 4. The proposed UWFS algorithm has been applied to a discontinuous permafrost region, Scotty Creek basin in the Northwest Territories, to simulate runoff generation from secondary runoff areas (the wetlands not directly connected to the fen network). The streamflow responses to different landcover transitions and meteorological forcings from different climate change scenarios are applied to quantify the effects of lateral permafrost thaw on the hydrological response of the study basin.

The UWFS algorithm is applied to improve our understanding of the effects of distribution characteristics, network branching, wetland deficit conditions, and cascade depth upon the contributing area and effective runoff from heterogeneous wetland-dominated basins. We can use the proposed model to understand potential long-term hydrological impacts of climate change located in regions where climate warming changes the role of wetlands from storage features to water conveyors.