Low-flow assessment for ungauged sub-basin in upper ping river basin, Thailand

Water scarcity has become one of the most remarkable problems in Thailand. An assessment of low-flow may lead to better water resources management and reduce the risk of water scarcity. The assessment of low-flow in gauged basins where the flow time series are available is straightforward. The challenge exists in ungauged or poorly-gauged basins where the flow data are unavailable or of low quality. Due to the studies of low-flow assessment in ungauged basins are of limited, this study aims to address the low-flow assessment in 25 sub-basins in the Upper Ping River basin in Thailand with available data from 1995-2014 by defining an applicable regionalization method for extrapolating beyond the limitations of observed flow data. Three regionalization methods namely regional regression method, sub-basin similarity method, and climate adjustment method are investigated for the selected sub-basins. The regional regression method is based on a stepwise regression procedure as the relationship between low-flow characteristics and basin physical characteristics. The sub-basin similarity method considered the weight of donor basins according to a combination of physical similarity and spatial proximity. The climate adjustment method considered the distance, choice of record augmentation technique, and the length of overlap period between the subject and the donor basins. Ninety-five-percentile flow (Q95), baseflow index (BFI), and the annual minimum 7-day moving average streamflow with a 10-year recurrence interval (7Q10) are selected in representing the low-flow characteristics of the sub-basins. The result indicated that the three regionalization methods are applicable to predict low-flow in the Upper Ping River basin but with a different predictive degree. However, the comparison further indicated that the climate adjustment method performs well in predicting 7Q10 and Q95 while for BFI it yields a moderate performance when there are available flow records of at least 5 years. Alternatively, applying the regional regression method with Q95 is more recommended than the sub-basin similarity method or with 7Q10 and BFI when there is no flow record available or available with a period of fewer than 5 years.