Assessment of seasonal characteristics of streamflow droughts under semiarid conditions

Abstract

Streamflow regime of semiarid environments is strongly influenced by rainy and non-rainy seasons, an indication that streamflow drought analysis should consider seasonality of streamflow behavior. The streamflow drought index (SDI), developed for continuous streamflow data, is modified to accommodate seasonal time series data. Possible SDI associations with standardized precipitation index are investigated at 3-, 6-, 9- and 12-month timescales. Case study area is semiarid Karkheh watershed (western Iran), with 42 years of data from selected rain gauge/hydrometric stations. Boxplots were used to establish rainy/non-rainy and wet/dry seasons from monthly data. According to Mann–Kendall test results, eight rain gauge and ten hydrometric stations indicated nonsignificant temporal trend, which were kept for further analysis. Markov chain results showed first-order dependency in all stations, indicating that drought predictability characteristics are based on the conditions of previous time period. Dry season streamflow droughts generally indicated longer durations compared to wet season droughts, and streamflow droughts were more persistent than meteorological droughts. According to results, historic drought of 1998–2001 has not recovered after more than a decade, indicating critical levels of SDI values during the dry season. Consideration of streamflow seasonality has been instrumental in detecting severe dry season droughts, not readily detectable when continuous time series data are used.