Missing data imputation in meteorological datasets with the GAIN method

Aim of this work is to present the preliminary results obtained using Generative Adversarial Imputation Networks (GAIN) to face the problem of incomplete time series in high frequency meteorological data. Meteorological data such as temperature, rain, wind etc., are being measured with many differen...

Full description

Saved in:
Bibliographic Details
Published in2021 IEEE International Workshop on Metrology for Industry 4.0 & IoT (MetroInd4.0&IoT) pp. 556 - 560
Main Authors Popolizio, Marina, Amato, Alberto, Politi, Tiziano, Calienno, Roberto, Di Lecce, Vincenzo
Format Conference Proceeding
LanguageEnglish
Published IEEE 07.06.2021
Subjects
Data models
Generative Adversarial Imputation Networks (GAIN)
meteorological time series data
missing data
Neural networks
Rain
Temperature distribution
Temperature measurement
Time series analysis
Training
Online AccessGet full text

Cover

More Information
Summary:Aim of this work is to present the preliminary results obtained using Generative Adversarial Imputation Networks (GAIN) to face the problem of incomplete time series in high frequency meteorological data. Meteorological data such as temperature, rain, wind etc., are being measured with many different techniques from many years. This fact has led to the creation of long time series that often are incomplete due to various reasons: temporary instrument faults, natural events, etc. Since these time series are used in many applications (from weather forecast to civil engineering design), their quality and completeness are becoming an even more critical question. In literature there are many studies aiming at meteorological time series completion using various approaches. In this context, authors show the results obtained using the GAIN method to complete high frequency temperature time series. The first obtained results seem to be very encouraging.
DOI:10.1109/MetroInd4.0IoT51437.2021.9488451