Publications

Wang, YQ; Gao, ZQ; Liu, DY (2019). Multivariate DINEOF Reconstruction for Creating Long-Term Cloud-Free Chlorophyll-a Data Records From SeaWiFS and MODIS: A Case Study in Bohai and Yellow Seas, China. IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING, 12(5), 1383-1395.

Abstract
A long-term reliable satellite chlorophyll-a (chl-a) data record is essential in understanding the state of ocean biology and quantifying its changes. Creating a long-term data record requires a combination/merger of multiple satellite products into one data record, since the lifetime of any single ocean color sensor is finite. However, because of differences in sensor design, calibration, and retrieval models, apparent cross-mission biases are usually observed between different sensor products. To attain a coherent multisensor chl-a data record, the observed cross-mission biases should be accurately addressed in the data combination/merging schemes. In this study, a multivariable data interpolating empirical orthogonal functions (M-DINEOF) approach was used to create long-term chl-a records by applying the sea-viewing wide field-of-view sensor and moderate resolution imaging spectroradiometer products. Under the assumption that the single-sensor chl-a product is free from spurious temporal artifacts and can be reference time series representing the actual variability of chl-a, the discrepancies of trends derived from different chl-a series were quantitatively evaluated based on statistical t-test and Taylor diagram analyses. Compared with direct concatenation and linear regression methods, the M-DINEOF method more effectively reproduced the main trend patterns observed in reference data series during their overlapped periods. The results highlight the importance of a cross-mission bias correction when combining multisensor satellite data records and suggest that the M-DINEOF reconstruction provides a simple and effective path forward for creating reliable multisensor ocean color records suitable for long-term trend analysis.

DOI:
10.1109/JSTARS.2019.2908182

ISSN:
1939-1404