Assessing the digital filter method for base flow estimation in glacier melt dominated basins

AbstractBase flow is an important component of streamflow. Although the simple digital filter method is widely used for base flow separation, the applicability in alpine rivers mainly dominated by glacier melt has not been described in detail. To assess and improve the performance of base flow estimates using the filter method for catchments dominated by glacier melt, the enhanced Soil Water Assessment Tool (SWAT) is used to obtain the estimates of streamflow and base flow for three catchments with different glacier melt contribution in arid and cold Northwestern China. The digital filter is then applied to the simulated streamflow to separate base flow and assess how well the base flow by the filter method matches these obtained using the SWAT model. In order to obtain the best match between the base flow by the filter method and those using the SWAT model, the linear regression model is used to estimate the relation between the filtered base flow and the glacier melt. It was found that the filtered base flow was matched well with base flow using the SWAT model during the low‐flow period. However, the base flow based on the digital filter method was overestimated during the high‐flow period, especially for the Manas River and Kumarik River. The base flow indexes by the digital filter estimates were 2.9%, 33.3% and 100% larger than those of the model method for the Gongnaisi River, Manas River and Kumarik River, respectively. The differences are larger with bigger glacier melt contribution. The performance of the digital filter is affected by the glacier melt, and it can be improved significantly by the combination of filtered base flow and the glacier melt. The base flow indexes by the improved filter are 1.5%, 4.4% and 10.7% larger than those of the model method for the Gongnaisi River, Manas River and Kumarik River, respectively. Copyright © 2015 John Wiley & Sons, Ltd.