Unveiling the Effects of Landscape Metrics on Regionalization of Runoff Parameters

Abstract Landscape Metrics (LMs) have an important role in linking spatial configuration with runoff to characterize hydrological response of a watershed. However, the effects of LMs on runoff parameters have not been investigated quantitatively, and predictive accuracy has not been examined in detail yet. To address this gap, runoff parameters are regionalized by LMs using simple linear and log-linear regressions, Catchment Similarity Approach (CSA), and a sound method the Ordered Weighted Averaging (OWA). Curve number method and Nash’s instantaneous unit hydrograph concept are used for lumped hydrological modelling at Kocaırmak and Darıören stream gauging stations located in Türkiye. In the watershed, 25 sub-basins are delineated, LMs are extracted, hydrological parameters are computed and regionalized by LMs. Prediction results are assessed by Nash–Sutcliffe Efficiency coefficient (E NS ), Peak Error (PE), Peak–Weighted Root Mean Square Error (PW–RMSE). E NS , PE, and PW–RMSE are computed using the extracted runoff parameters as 0.943, 0.792, 2.20%, 0.84%, 68.30 and 9.10 for Kocaırmak and Darıören, respectively. Regression-based methods reveal that number of patches, landscape shape index, area-weighted patch area, mesh and various statistics of shape indexes estimate more enhanced predictions than the extracted runoff parameters. CSA contributes to estimate satisfactory runoff hydrographs. Outcomes from regression-based methods and CSA provide satisfactory predictions by collaboration of sound decision strategy coefficients and threshold values of catchments similarity index in the OWA method. The methodology adapted in this study may initiate attempts encouragingly to link LMs with hydrological parameters to develop sustainable land use strategies by controlling runoff parameters.