The probability distribution of daily streamflow in perennial rivers of Angola

Research output: Contribution to journalArticlepeer-review

2 Citations (Scopus)


Hydrological observations in Angola are quite scarce and, as such, the water allocation process, one of the main components of water resources management, is commonly undertaken with a high degree of uncertainty. It is, therefore, vital to increase our understanding of the hydrological processes, namely the frequency distribution of daily streamflow, in this part of southern Africa and validate efficient methods that facilitate the extrapolation of information from gauged to ungauged catchments. All the components of this study were carefully designed to address all the above-mentioned goals. This was achieved with the modelling of 121 flow-duration curve samples observed in different parts of the country, including large datasets (e.g., 1954/55 to 1968/69) but mainly focused on the hydrological years of 1967/1968 and 1973/1974 and the implementation of regional frequency analysis based on the L-moments approach. The frequency distribution of daily streamflow was approximated with nine different probability distribution functions considering different subsets of the daily streamflow time series: (i) daily streamflow (ii) different subsets of daily streamflow divided into two ‘flows seasons’, wet and dry, (iii) and the previous subsets transformed with the definition of a thirty-day time lag, thereby reducing the serial dependency of daily streamflow and enabling the use of the L-moments approach. Overall, the results enabled two probability distribution functions to be identified able to provide a remarkable approximation to all the above-mentioned daily streamflow datasets (the four-parameter Kappa and three-parameter Generalized Pareto distributions). Furthermore, the regional frequency analysis supported the prediction of daily streamflow quantiles for eight test catchments with impressive accuracy (Nash-Sutcliffe efficiency coefficient: μ = 0.86; σ = 0.10; Pearson correlation coefficient: μ = 0.97; σ = 0.02), clearly showing that this approach represents a sound alternative for the prediction of daily streamflow in ungauged catchments located in this region.

Original languageEnglish
Article number126869
JournalJournal Of Hydrology
Publication statusPublished - Dec 2021


  • Daily streamflow
  • Frequency distribution
  • L-moments
  • Ungauged catchment


Dive into the research topics of 'The probability distribution of daily streamflow in perennial rivers of Angola'. Together they form a unique fingerprint.

Cite this