Gap Filling of Precipitation Data by SSA - Singular Spectrum Analysis

27th IUPAP Conference on Computational Physics

物理学;计算机科学

Filho, A.S.F.^1 ; Lima, G.A.R.^1

Departamento de Engenharia Sanitária e Ambiental, Universidade Federal de Mato Grosso, Av. Fernando Corrêa da Costa n° 2367, Cuiabá MT

78060-900, Brazil^1

关键词: Environmental science;    Forecasting algorithm;    Natural systems;    Precipitation data;    Precipitation time series;    Relative length;    Singular spectrum analysis;    Spectral Estimation;   

Others  :  https://iopscience.iop.org/article/10.1088/1742-6596/759/1/012085/pdf
DOI  :  10.1088/1742-6596/759/1/012085

学科分类：计算机科学（综合）

来源: IOP

【 摘 要 】

From the macroscopic standpoint, the precipitation time series is obtained from observation of natural systems rather than in the laboratory. These time series are often full of gaps (missing values) due to the conditions under which the measurements are made. Missing values give rise to various problems in spectral estimation, inhibit statistical analysis and in specifying boundary conditions for numerical models. Hence, gap filling is necessary in environmental science. The aim of this study is to highlight the application of the SSA forecasting algorithm to fill in missing values to real-life time series. It was applied to several monthly precipitation time series recorded over a large savannah area in Brazil. The results are promising and the accuracy and reliability depend on the pattern and relative length of the gaps with respect to the total length of the time series and presence of noise.