Performance assessment and improvement of recursive digital baseflow filters for catchments with different physical characteristics and hydrological inputs

• Verify optimal filter parameter & assess performance over wider range of conditions.
• Develop regression equation between filter parameter, performance and conditions.
• Lyne and Hollick (LH) filter performs better than Boughton and Eckhardt filters.
• Optimal filter parameters vary considerably over wider range of conditions.
• Regression equations can predict filter parameter and filter performance well.

Recursive digital filters (RDFs) are one of the most commonly used methods of baseflow separation. However, how accurately they estimate baseflow and how to select appropriate values of filter parameters is generally unknown. In this paper, the output of fully integrated surface water/groundwater (SW/GW) models is used to obtain optimal parameters for, and assess the accuracy of, three commonly used RDFs under a range of physical catchment characteristics and hydrological inputs. The results indicate that the Lyne and Hollick (LH) filter performs better than the Boughton and Eckhardt filters, over a larger range of conditions. In addition, the optimal values of the filter parameters vary considerably for all three filters, depending on catchment characteristics and hydrological inputs. The dataset of the 66 catchment characteristics and hydrological inputs, as well as the corresponding simulated total streamflow and baseflow hydrographs obtained using the SW/GW model, can be downloaded as Supplementary material.