Comparison of gap-filling methods for CO2 flux data

• Autor(a) principal: Goltz, Bernardo Ivo
• Data de Publicação: 2023
• Outros Autores: Aimi, Daniele Morgenstern, Mergen, Alecsander, Souza, Vanessa de Arruda, Veeck, Gustavo Pujol, Bremm, Tiago, Stefanello, Michel Baptistella, Roberti, Débora Regina
• Tipo de documento: Artigo
• Idioma: por
• Título da fonte: Revista Ciência e Natura (Online)
• Texto Completo: https://periodicos.ufsm.br/cienciaenatura/article/view/80997
• topic: CO2 flux; Gap filling; Missing data; Fluxo de CO2; Preenchimento de falhas; dados faltantes

Data collected by sensors are always subject to possible failures, whether due to power failure, external interference, among others.. Moreover, much of the data is not considered during the filtering process because it is physically inconsistent. These failures result in the need to implement various methods of data processing, with a focus on filling in missing records. In the case of CO2 flux data, filling in the missing data is crucial to obtain annual data and the carbon balance. The REddyProc package is widely used and documented in terms of filling this type of data. However, modern methods have been increasingly explored to optimize this process. In this study, we compare data filling between the REddyProc package and the KNN Imputer method. Preliminary results show that the REddyProc package has better statistical indices when filling CO2 streams compared to the KNN method.