Catchment compatibility via copulas: A non-parametric study of the dependence structures of hydrological responses

• Published in: Advances in water resources 2016-04, Vol.90, p.116-133
• Main Authors: Grimaldi, S., Petroselli, A., Salvadori, G., De Michele, C.
• Format: Article
• Language: English
• Subjects: Basins; Brackish; Catchment compatibility; Catchments; Compatibility; Copula; Discharge; Enrichment; Hydrology; Non-parametric test; Rainfall-runoff transformation; Signatures; Statistics
• ISSN: 0309-1708; 1872-9657
• DOI: 10.1016/j.advwatres.2016.02.003

•Dependence structures as signature of hydrological responses.•Compatible catchments have similar dependence structures.•It is possible to merge observed data for improving copula estimation in compatible catchments. The similarity of catchment responses is a fundamental issue for regionalization studies, and hydrograph attributes (i.e., Discharge Peak, Volume, and Duration) can reveal the signature and the synthesis of local scale processes. Here, we focus the attention on the “compatibility” between catchments, viz. on the possibility to transfer, from one catchment to another, the information about the dependence structures at play. In particular, we statistically investigate the possible relationships between the features of different Basin Scenarios (characterized via the Concentration Time Tc and the Curve Number CN) and the corresponding dependence structures ruling the joint statistics of Discharge, Volume, and Duration. Given a large set of synthetic runoff time series, generated via a rainfall-runoff model, recent non-parametric tests, based on empirical copulas, are used to compare the dependence structures associated with different soil uses and concentration times. The results indicate how the hydrological properties may affect the dependence structure. The outcomes of the investigation could be particularly effective in two practical applications: (1) for determining the degree of compatibility of the dependence structures associated with different basin scenarios, and (2) for enriching scanty data bases, in order to improve the estimation of multivariate copulas.