Abstract
Abstract. A method to reconstruct missing data in sea surface temperature data using a neural network is presented. Satellite observations working in the optical and infrared bands are affected by clouds, which obscure part of the ocean underneath. In this paper, a neural network with the structure of a convolutional auto-encoder is developed to reconstruct the missing data based on the available cloud-free pixels in satellite images. Contrary to standard image reconstruction with neural networks, this application requires a method to handle missing data (or data with variable accuracy) in the training phase. The present work shows a consistent approach which uses the satellite data and its expected error variance as input and provides the reconstructed field along with its expected error variance as output. The neural network is trained by maximizing the likelihood of the observed value. The approach, called DINCAE (Data INterpolating Convolutional Auto-Encoder), is applied to a 25-year time series of Advanced Very High Resolution Radiometer (AVHRR) sea surface temperature data and compared to DINEOF (Data INterpolating Empirical Orthogonal Functions), a commonly used method to reconstruct missing data based on an EOF (empirical orthogonal function) decomposition. The reconstruction error of both approaches is computed using cross-validation and in situ observations from the World Ocean Database. DINCAE results have lower error while showing higher variability than the DINEOF reconstruction.
Highlights
The ocean temperature is an essential variable to study the dynamics of the ocean, because density is a function of temperature and the ocean velocity variability depends partially on ocean temperature
It is expected that the cost function will fluctuate using any optimization method based on minibatch, because the cost function is evaluated using a different minibatch at every iteration
Even if the dataset is small and the gradient could be computed over the entire dataset at once, using minibatches is still advised because these fluctuations allow the cost function to get out of local minima (Ge et al, 2015; Masters and Luschi, 2018)
Summary
The ocean temperature is an essential variable to study the dynamics of the ocean, because density is a function of temperature and the ocean velocity variability depends partially on ocean temperature. The ocean sea surface temperature (SST) has been routinely measured since the beginning of the 1980s. DINEOF (Data INterpolating Empirical Orthogonal Functions; Beckers and Rixen, 2003; Alvera-Azcárate et al, 2005) is an iterative method to reconstruct missing observations reducing noise in satellite datasets using empirical orthogonal functions (EOFs). DINEOF has been applied to several oceanographic variables at different spatial resolutions (e.g., Alvera-Azcárate et al, 2005, for SST; Alvera-Azcárate et al, 2007, for ocean color; Alvera-Azcárate et al, 2016, for sea surface salinity), providing accurate reconstructions. A truncated EOF decomposition will focus primarily on spatial structures with a “strong” signature
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
