3.3 Filtering criteria of the recession flow
To effectively improve the filtered results as low-flow, the filtered data must meet certain criteria. Cheng, Zhang, and Brutsaert (2016) evaluated the sensitivity between the recession index and each filtering criterion. The results indicated that the criteria for the number of data points before and after the recession events and those affected by the major streamflow event can be reduced in a relatively smaller basin due to the faster hydrological response after the rainfall event. According to the small catchments used here, this study used and revised the filtering criteria based on Brutsaert (2008) as follow:
  1. Eliminate the data where the flow variation is equal to or larger than zero.
  2. Eliminate the one data point before and two data points after each data point that meets criterion (1).
  3. Eliminate three data points after a major streamflow event defined from the flow duration curve (Kingsford, 2000).
  4. Eliminate data points with singular values.
3.4 Dynamic storage properties
To explore the physical-based representative of impact parameter, this study estimated the change in dynamic storage and storage-discharge sensitivity as dynamic storage properties derived from recession analysis.
Based on Kirchner’s (2009) storage-discharge sensitivity function, substituting equation (2) with environmental change can be expressed as follows: