A new method based on artificial bee colony programming for the regional standardized intensity-duration-frequency relationship


GÖRKEMLİ B. , ÇITAKOĞLU H. , Haktanir T., KARABOĞA D.

ARABIAN JOURNAL OF GEOSCIENCES, vol.15, no.3, 2022 (Journal Indexed in SCI) identifier

  • Publication Type: Article / Article
  • Volume: 15 Issue: 3
  • Publication Date: 2022
  • Doi Number: 10.1007/s12517-021-09377-1
  • Title of Journal : ARABIAN JOURNAL OF GEOSCIENCES
  • Keywords: Regional intensity - duration - frequency relationship, Artificial bee colony programming (ABCP), SYMBOLIC REGRESSION, GENETIC ALGORITHM, ELASTIC-MODULUS, RAINFALL, MODEL

Abstract

The aim of this study was to develop new analytical models to define the regional rainfall intensity - duration - frequency (IDF) relationship for the Inland Anatolia Region, which is determined discretely by the L-Moments method at many values of rainfall durations and frequencies. First, the parameters of each one of nine commonly used empirical equations were calibrated to provide the best possible definition of the IDF relationship for the Inland Anatolia Region. Next, analytical models best fitted to the IDF relationship of the L-Moments method were generated by the artificial bee colony programming (ABCP) approach, such that a combination of nine different sets were simulated, taking into account three cost functions and three maximum depths. Mean absolute error, root mean square error, mean square error, Nash-Sutcliffe efficiency coefficient, Willmott's refined index, performance index, and coefficient of determination were computed to assess the accuracies of the empirical equations and of the ABCP models. These criteria revealed that the ABCP models defined the IDF relationship better than the empirical equations over the entire range of frequencies from 2 to 10,000 years. The accuracy of the empirical equations is much worse than the ABCP model, especially for frequencies smaller than 2000 years. Finally, Kruskal - Wallis tests were applied on all of the IDF relationships given by (1) the L-Moments method, (2) the empirical equations, and (3) the ABCP approach. These results indicated that the numerical values of these three models were from the same population.