GEO DATA

Original Paper

Imputation of Ocean-color Product in Polynya Region of Antarctica for Primary Productivity Estimates: Jinku Park, Sungjae Lee, Jeong-Hoon Kim, Hyun-Cheol Kim; GEO DATA. 2023;5(1):8-14. Published online March 28, 2023; DOI: https://doi.org/10.22761/GD.2023.0002

Abstract PDF: This study, focusing on the Antarctic polynyas, performed the imputation of chlorophyll-a concentration (Chl-a) dataset, which is one of the ocean color products mainly used for estimating primary productivity, using the Data Interpolating Empirical Orthogonal Function method and constructed accurate time-series data that excludes as much uncertainty as possible in long-term variability studies due to missing data. The polynya regions were classified into a total of 23 zones through quantitative criterions, and the statistical accuracy of imputation performance was 0.89 for R2 and 0.42, 0.24, and 0.15 for root mean square error, mean squared error, mean absolute error, respectively, on average, showing the ability to perform generally accurate reconstruction. Finally, the reconstructed Chl-a data showed a relatively stable fluctuation compared with standard satellite Chl-a data, and tended to be underestimated due to the expansion of the observable regions. We expect that securing these relatively stable and accurate estimates will be significantly different from the time-series data composed of standard Chl-a estimates, enabling more accurate variability and trend analysis.

Article

Reconstruction of Satellite Chlorophyll-a Concentration in the Southwestern East Sea using Imputation Method: Jinku Park, Sungjae Lee, Hyun-Cheol Kim; GEO DATA. 2021;3(4):11-17. Published online December 31, 2021; DOI: https://doi.org/10.22761/DJ2021.3.4.002

Abstract PDF: The chlorophyll-a concentration (CHL) data observed with the ocean color sensors have been widely used in the various studies related to phytoplankton. However, irregularly distributed missing data induced by clouds, a unique nature of optics, can cause large uncertainty, and a solution to this missing issue has been continuously demanded until now. We investigated the applicability of the data interpolating empirical orthogonal function and evaluated the reconstruction results in the southwestern East Sea. A total of 311 decomposed modes were used, showing a coefficient of determination of about 0.86 and a root mean square error of 0.37 mg m⁻³, compared to the truth data. Overall, it was confirmed that the observed CHL was overestimated compared to the reconstructed CHL when the spatio-temporal average was taken.

Search