Reconstruction of chlorophyll-a satellite data in Bohai and Yellow sea based on DINCAE method. Issue 9 (3rd May 2022)

• Record Type: Journal Article
• Authors: Luo, Xiaomin; Song, Jun; Guo, Junru; Fu, Yanzhao; Wang, Linhui; Cai, Yu
• Is Part Of: International journal of remote sensing. Volume 43:Issue 9(2022)
• Page Start: 3336
• Page End: 3358
• Publication Date: 2022-05-03
• Subjects: DINEOF -- DINCAE -- Bohai and Yellow seas -- chlorophyll-a -- neural network; Remote sensing -- Periodicals; Télédétection -- Périodiques; 621.3678
• DOI: 10.1080/01431161.2022.2090872
• Languages: English
• ISSNs: 0143-1161

Abstract

ABSTRACT: The Data Interpolation Empirical Orthogonal Functions (DINEOF) method is accurate in filling missing remote sensing datasets. This study used the Data-Interpolating Convolutional Auto-Encoder (DINCAE) method to reconstruct chlorophyll-a concentration (Chl-a) data from the Bohai and Yellow seas and compare it with DINEOF. The reconstruction power of the two methods was evaluated using cross-validation and in situ data, and the accuracy of both data validated by DINCAE was higher than that of DINEOF. The reconstruction results of the two methods were analysed from two perspectives: time and space scale. The reconstruction errors of DINCAE were lower than those of DINEOF at different temporal scales, and the temporal sensitivity of DINEOF was greater than that of DINCAE. DINCAE outperforms DINEOF in terms of controlling the small-scale features and Chl-a values. In terms of efficiency, DINCAE is five times faster than DINEOF for similar reconstruction conditions.