Graphical and recursive digital filter techniques in the separation of base flow, A comparison in Ardabil Province rivers

Background and objectives: Base flow is an important component of flow hydrograph, which feeds by ground water or shallow subsurface flow, Sub surface flow and plays an important role in ecological function of a river. Identification of base flow contribution provides a basis to assess the watershed response and information on base flow availability and contributions is needed to develop water quantity and water quality management strategies in watershed scale. Also, the separation of the base of the important factors in hydrologic modeling and increase the accuracy of the model calibration in rainfall-runoff prediction. Comparison of different watersheds based on flow recession and low flow components can provide useful information regarding the storage characteristics and flow recharge. Also, the base flow contribution amount can be used in the analysis of environmental flow assessment and river flow health condition. Base flow component strongly affected by flow extraction through the flood plain wells which can be a source of uncertainty in the base flow separation by well-known numerical methods. The aim of this study was to compare graphically and recursive digital filtering techniques in base flow separation and selection of appropriate methods to estimate the base flow contribution in number of rivers over Ardebil Province.
Materials and methods: The base flow was estimated on daily basis using three graphical methods, including, 1) Fixed time interval, 2) Sliding time interval, 3) Local minima and six recursive digital filters including, 1) One-parameter algorithm, 2) Bughton two-parameter algorithm, 3) IHACRES three parameter method, 4) Lynne and Hollick, 5) Chapman algorithm and 6) Exponentially Weighted Moving Average methods. The results of these methods were compared with the recession analysis method as a reference. Estimated and analysis the results of base flow methods were compared for a 22-year period (1989-2011) for which stream flow data were available in 5 river gauge station in different ranges of flow rates over Ardabil Province.
Results: According to the results and the comparison of graphical and filtering methods of base flow separation, it was found that the accuracy of graphical methods is not acceptable (with negative and less than unity values) according to Nash-Sutcliffe efficiency criteria. Accordingly, the performance of Fixed time interval method according to Nash-Sutcliffe efficiency criteria – 0.04 did not give a good result in the study area. While, Two-parameter Boughton and One-parameter algorithms are most suitable methods in estimating the base flow component of hydrograph with 0.88 and 0.86 efficiency criteria respectively in the study area.
Conclusion: The results showed that graphical methods had not satisfactory results, but the digital filtering techniques based on signal processing theory had high accuracy in separating the flow hydrograph to speed and delayed components. The estimated base flow hydrographs by recursive digital filter methods showed good agreement overall with the recession analysis method. As a concluding remark, the recursive digital filters are more suitable methods in the study area with nonlinear nature complying the sequence of discharge data time series with respect to linear graphical methods. The suggested appropriate methods can be used to low flow regional analysis, hydrologic modelling and determination of the base flow contribution in daily flow hydrograph data in the study region.