Regionalizing Flood Magnitudes using Landscape Structural Patterns of Catchments

Abstract

Emerging as an important issue in the disciplines of landscape ecology and landscape hydrology which inspired it, defining the concept of landscape metrics in a hydrological context has become a challenge to both landscape planners and engineers. Accordingly, the present study addresses the relationships existing between flooding phenomena and landscape metrics (shape index, fractal dimension index, perimeter-area ratio, related circumscribing circle, and contiguity index) of land use/land cover, hydrological soil groups and geological permeability classes. A regionalization approach was adopted employing 39 select catchments (33—4800 km2 in area, 0.47—21 m3 s−1 in mean discharge), located within the southern basin of the Caspian Sea. These catchments were predominantly covered by forest (57.4%), while rangeland, farmland and urban areas accounted for 25.9%, 11.7%, and 1.6%, respectively. Class-level landscape structural metrics of land use/land cover, hydrological soil groups and geological permeability classes have then been served as inputs to stepwise multiple linear regression analysis in an attempt to explain the flood magnitudes. The regression models (0.69 ≤ r2 ≤ 0.84) suggested that the catchments’ flood magnitude could explicitly be predicted using average measure of the shape and related circumscribing circle indices for the land use/land cover classes and those of hydrologic soil groups and geological permeability classes of the catchments. This indicated that regularity (vs. irregularity) of the landscape, pedoscape, and lithoscape, as represented by the shape index as well as the circumscribing circle index (for elongation and convolution), explained 69–84% of the variation in the flood magnitudes in the catchment.