Comparison of methods for filling daily and monthly rainfall missing data: statistical models or imputation of satellite retrievals?
Nenhuma Miniatura disponível
Data
2022
Título da Revista
ISSN da Revista
Título de Volume
Editor
Resumo
Accurate estimation of precipitation patterns is essential for the modeling of hydrological
systems and for the planning and management of water resources. However, rainfall time series,
as obtained from traditional rain gauges, are frequently corrupted by missing values that might
hinder frequency analysis, hydrological and environmental modeling, and meteorological drought
monitoring. In this paper, we evaluated three techniques for filling missing values at daily and
monthly time scales, namely, simple linear regression, multiple linear regression, and the direct
imputation of satellite retrievals from the Global Precipitation Measurement (GPM) mission, in
rainfall gauging stations located in the Brazilian midwestern region. Our results indicated that,
despite the relatively low predictive skills of the models at the daily scale, the satellite retrievals
provided moderately more accurate estimates, with better representations of the temporal dynamics
of the dry and wet states and of the largest observed rainfall events in most testing sites in comparison
to the statistical models. At the monthly scale, the performance of the three methods was similar, but
the regression-based models were unable to reproduce the seasonal characteristics of the precipitation
records, which, at least to some extent, were circumvented by the satellite products. As such, the
satellite retrievals might comprise a useful alternative for dealing with missing values in rainfall time
series, especially in those regions with complex spatial precipitation patterns.
Descrição
Palavras-chave
Filling of missing values, Rain gauges, Linear regression, Global precipitation measurement
Citação
DUARTE, Luíza Virgínia; FORMIGA, Klebber Teodomiro Martins; COSTA, Veber Afonso Figueiredo. Comparison of methods for filling daily and monthly rainfall missing data: statistical models or imputation of satellite retrievals? Water, Basileia, v. 14, n. 19, e3144, 2022. DOI: 10.3390/w14193144. Disponível em: https://www.mdpi.com/2073-4441/14/19/3144. Acesso em: 11 Dez. 2024.