Influence of Precipitation Data Consistency Analysis on the Estimation of the Frequency Factor for Regional Statistical Probable Maximum Precipitation Calculation

Código

I-EBHE0187

Autores

Thacio Carvalho Pereira, Karine Dias Nogueira, Jeniffer da Conceição Lino, Fernanda Trivellato

Tema

WG 1.10: Hydrologic Design - Solutions & Communication

Resumo

The estimation of the Probable Maximum Precipitation (PMP) using the statistical approach involves determining the frequency factor. Recent studies indicate the need to adjust this factor since the traditional methodology proposed by Hershfield (1965), determined locally and through the use of abacuses, may overestimate the PMP for regions with less pronounced precipitation regimes. Thus, in order to calibrate Hershfield's original approach, studies suggest defining PMP values based on a regional analysis to determine the frequency factor k, where k values are estimated for each rain gauge station. To this end, a large number of series of annual maximum rainfall heights should be considered, according to the durations of interest, within a climatologically homogeneous region, to then analyze the distribution of the highest k values within the area of interest. Since evaluating the quality of data from a rain gauge station can be quite subjective, this study sought to verify the influence of a robust consistency analysis of the raw data inventoried on the definition of regional frequency factors for a specified region in the state of Minas Gerais, near Serra do Sapo. To delimit the study's area of influence, the climate classifications of the Brazilian Institute of Geography and Statistics (IBGE) and Köppen were used. Thus, rain gauge stations publicly available on the portal of the National Water and Sanitation Agency (ANA) located within the stipulated delimitation were evaluated, filtering the stations with at least 15 years of monitored data. For each historical series, the number of failures was verified, especially in the rainy season months, excluding the years with a significant number of monitoring failures. The presence of physical or statistical outliers was also evaluated based on the implementation of box plots and the use of the Interquartile Range (IQR) methodology. Once the annual maximum precipitation series for the durations of interest were defined, the k values for each station were calculated, which depend on variables (e.g., sample maximum, mean, and standard deviation without considering the maximum). For stations that presented physical outliers, the k value was calculated first by considering them and then by excluding these data from the sample. The study showed that the frequency factor calculation equation is very sensitive to extreme data in the samples, which can significantly increase the k and consequently impact the PMP estimate calculated by the regional approach. Given the large amount of data that the study of regional PMP demands, the importance of a meticulous evaluation of the raw data quality was verified, as their descriptive statistics have a significant influence on the results.