Impact of data time resolution in long term baseflow index assessment from mass balance filtering

Key words: Mass balance filtering, baseflow, BFI, time-resolution, electrical conductivity, permeability

Baseflow plays an important role in sustaining freshwaters quality and quantity at the global scale. Quantitative estimates of baseflow are necessary but no direct observations are available to the purpose, thus hydrograph filtering appears as a valid solution to quantify the baseflow process. The mass balance filter (MBF), based on electrical conductivity (EC) observations, is one of the most objective filtering techniques.

The aim of the present study is to analyze the impact of data time resolution in the assessment of the long-term scale baseflow index (BFI, the ratio between baseflow and total flow) by hydrograph filtering. The MBF method was used to estimate the long-term BFI of 64 catchments across Continental United States ranging from about 5 to 50000 Kmq, from arid to continental climate conditions and accounting for 5 to 10 years of continuous observations. Streamflow and EC data were collected from the United States Geological Survey (USGS) National Water Information System website at the 15 minutes time resolution and aggregated at the daily scale. BFI₁₅, the BFI computed at the 15 min time resolution, was compared with BFI₂₄, that is with the BFI computed at the daily scale. The difference among the two indices was investigated in relation to catchment climate and physiographic characteristics.

The large dataset was divided into two groups, a poorly-drained group and a well-drained group on the base of the catchment permeability assessed by the GLobal HYdrogeology MaPS (GLHYMPS) project. The first group corresponds to BFI values < 0.5, the second to BFI values > 0.5. Overall the BFI₁₅ was found to be larger than BFI₂₄, with an average difference of about 7%, probably caused by the fact that at finer time scale a wider spectra of streamflow processes and components can be detected, more evidently during the peak flow conditions. The average difference drops to 1% in the case of the well-drained hydrological systems, which appear likely the less impacted by the monitoring time resolution and increase on average to 11% (with maximum values approaching 50%) in the case of the poorly drained systems.