ABSTRACT

Preventing the occurrence of extreme hydrological events exceeds the current abilities of humanity. However, using adequate tools to estimate the probability associated with different values of hydrological variables, such as heavy rainfall, can help mitigate social and environmental damage. This study aimed to verify the hypothesis that the records of maximum rainfall, at different intervals of duration, in the three main rainfall stations in the state of Acre fit the log-normal distribution of two parameters (LN2) and, therefore, to evaluate the events of heavy rainfalls, based on the association between the duration of the rain and the return period. The equations conformed to the classic model of the LN2 model using the method of maximum likelihood once the baseline assumption of stationarity was met using the Mann-Kendall test. Adherence to the model was verified using the Anderson-Darling test. The adjusted distributions allowed for estimating the rainfall intensity at different durations and return periods for the cities of Cruzeiro do Sul, Tarauacá, and Rio Branco and are relevant elements for planning actions aimed at strengthening the components of water safety in the state of Acre.

Key words:
hydrological modeling; log-normal distribution; maximum rainfall; extreme events

HIGHLIGHTS:

The temporal behavior of the rainfall regime was stationary at the analyzed places.

The two-parameter log-normal is recommended for the data among the proposed probability distributions.

Included in the Brazilian Amazon, the region has high rainfall values for different durations.