Uni- and multivariate bias adjustment of climate model simulations in Nordic catchments: Effects on hydrological signatures relevant for water resources management in a changing climate

• Published in: Journal of hydrology (Amsterdam) 2023-08, Vol.623, p.129807, Article 129807
• Main Authors: Tootoonchi, Faranak, Todorović, Andrijana, Grabs, Thomas, Teutschbein, Claudia
• Format: Article
• Language: English
• Subjects: Bias adjustment; Bias correction; Climate change; Climate model; High latitudes; Hydrological signatures
• ISSN: 0022-1694; 1879-2707; 1879-2707
• DOI: 10.1016/j.jhydrol.2023.129807

[Display omitted] •Bias adjustment improves accuracy and consistency of simulated hydrological signatures.•Not a single bias adjustment method enhances performance in all analyzed signatures.•Univariate distribution scaling (DS) performs well, especially in rainfall-driven catchments.•Multivariate methods perform better for low-flow signatures in snowmelt-driven catchments. Hydrological climate-change-impact studies depend on climatic variables simulated by climate models. Due to parametrization and numerous simplifications, however, climate-model outputs come with systematic biases compared to the observations. In the past decade, several methods of different complexity and dimensionality for adjustment of such biases were introduced, but their benefits for impact studies and accurate streamflow projections are still debated. In this paper, we evaluated the ability of two state-of-the-art, advanced multivariate bias-adjustment methods to accurately reproduce 16 hydrological signatures, and compared their performance against two parsimonious univariate bias-adjustment methods based on a multi-criteria performance evaluation. The results indicated that all bias-adjustment methods considerably reduced biases and increased the consistency of simulated hydrological signatures. The added value of multivariate methods in maintaining dependence structures between precipitation and temperature was not systematically reflected in the resulting hydrological signatures, as they were generally outperformed by univariate methods. The benefits of multivariate methods only emerged for low-flow signatures in snowmelt-driven catchments. Based on these findings, we identified the most suitable bias-adjustment methods for water-resources management in Nordic regions under a changing climate, and provided practical guidelines for the selection of bias-adjustment methods given specific research targets and hydroclimatic regimes.