Application of wavelet transform for extracting and analysing evapotranspiration-induced diel fluctuations in streamflow records

Abstract

Diurnal fluctuations in groundwater and streamflow are produced due to evapotranspiration, indicating a strong connection between streamflow and the groundwater reservoir. Studying the patterns of diel fluctuations can provide valuable information on the hydrological processes in a catchment. Analysing these fluctuations makes it possible to estimate the evapotranspiration rate as well. In this paper, the signal analysis technique of the wavelet transform is applied to the streamflow time series to extract and analyse diel fluctuations. The performance of two main types of wavelet transform, continuous and discrete, is assessed against widely applied methods of trend extraction, such as moving average. The results show that wavelet transform can be used successfully to identify periodic and non-periodic features of the time series, such as seasonal and trend components, and to distinguish between signal and noise. The continuous wavelet transform demonstrates that the diurnal component of streamflow exhibits significant variation over different temporal scales, with the dominant periods ranging from 12 to 36 h. In conclusion, the findings suggest that wavelet transform can effectively capture evapotranspiration-induced diurnal streamflow fluctuations and provide insights into the hydrological processes at different temporal scales.