TYPE 1 ERROR AND "POWER" OF ADHERENCE TESTS FOR PLUVIOMETRIC GAUGES IN CEARÁ
DOI:
https://doi.org/10.14393/19834071.2013.21752Abstract
Due to the short extent of available historical time series of rainfall and flow, adherence tests are carried out to verify the fitting of different probability distributions to such time series. When there is adherence, the distributions can estimate the critical values of the phenomena and their occurrence. Thus, the analysis of four adherence tests was conducted including - Kolmogorov-Smirnov, Chi-Square, Cramer-von Mises and Anderson-Darling - considering Type 1 Error and "Power", in order to determine the behavior of each test regarding the fitting of four probability distributions: Gamma, Normal, Lognormal and Gumbel. Such study has been done via Monte Carlo simulation. The results of the analysis of the tests were used in the definition of which probability distribution was a better fit for the historical time series of 24 pluviometric gauges in the State of Ceará, Brasil. It was verified that, considering the coefficient of variation and the extent of data of the gauges, the Type 1 Error test that is closer to the expected values and has greater "power" is the Chi-Square. Keywords: historical time series, hydrological phenomena, synthetic series, Monte Carlo Simulation.Downloads
Download data is not yet available.
Downloads
Published
2014-04-02
Issue
Section
Civil Engineering