Geostatistical extrapolation of area-density relations of water bodies from local to regional scale

English Abstract:

The northern prairie landscape in North America is typified by topographic depressions which vary from small ephemeral ponds to large permanent lakes. These features are hydrologically and ecologically important because they providing habitats for migratory birds as well as focusing recharge to underlying aquifers. Quantifying the contribution of ephemeral ponds to groundwater recharge for large areas has proven challenging due to the difficulty in delineating the presence of topographic depressions within the landscape at a high enough resolution. Previous work has identified a power law relationship between the surface area of water bodies in depressions and their spatial density (i.e., number of water bodies per area). This power law relationship is influenced by landscape characteristics, in particular landform type (e.g., hummocky terrain). High-resolution (0.3 m) digital images were used to delineate water bodies in a 250-km² watershed of West Nose Creek (WNC) near Calgary, Canada. This dataset contains two well-defined power law relationships, one for hummocky and one for undulating terrains, which collectively comprise 75% of the watershed. Training images, based on the hummocky and undulating landform types within the WNC watershed were used to produce power law statistics for each of the landform types. The power law statistics were used to generate stochastic realizations of depression patterns which conform to the power law relationship determined from the WNC watershed. The stochastic images can then be used to generate distributions of the small ephemeral water bodies for larger areas with these landform types in order to investigate the spatial distribution of groundwater recharge.