ArticlePDF Available

Abstract and Figures

Satellite-derived products are of key importance for the high-resolution monitoring of the ocean surface at a global scale. Due to the sensitivity of spaceborne sensors to the atmospheric conditions as well as the associated spatio-temporal sampling, ocean remote sensing data may involve high-missing data rate. The spatio-temporal interpolation of these data remains a key challenge to deliver L4 gridded products to end-users. Whereas operational products mostly rely on model-driven approaches, especially optimal interpolation based on Gaussian process priors, the availability of large-scale observation and simulation datasets advocate for the development of novel data-driven models. This study investigates such models. We extend the recently introduced analog data assimilation to high-dimensional spatio-temporal fields using a multi-scale patch-based decomposition. Using an Observing System Simulation Expriment (OSSE) for sea surface temperature, we demonstrate the relevance of the proposed data-driven scheme for the real missing data patterns of the high-resolution infrared METOP sensor. It resorts to a significant improvement w.r.t. state-of-the-art techniques in terms of interpolation error (about 50 % of relative gain) and spectral characteristics for horizontal scales smaller than 100km. We further discuss the key features and parameterizations of the proposed data-driven approach as well as its relevance with respect to classical interpolation techniques.
Content may be subject to copyright.
Data-driven Models for the Spatio-Temporal
Interpolation of satellite-derived SST Fields
Ronan Fablet, Senior Member, IEEE, Phi Huynh Viet and Redouane Lguensat, Student Member, IEEE
Satellite-derived products are of key importance for the high-resolution monitoring of the ocean surface on a
global scale. Due to the sensitivity of spaceborne sensors to the atmospheric conditions as well as the associated
spatio-temporal sampling, ocean remote sensing data may be subject to high-missing data rates. The spatio-temporal
interpolation of these data remains a key challenge to deliver L4 gridded products to end-users. Whereas opera-
tional products mostly rely on model-driven approaches, especially optimal interpolation based on Gaussian process
priors, the availability of large-scale observation and simulation datasets calls for the development of novel data-
driven models. This study investigates such models. We extend the recently introduced analog data assimilation to
high-dimensional spatio-temporal fields using a multi-scale patch-based decomposition. Using an Observing System
Simulation Experiment (OSSE) for sea surface temperature, we demonstrate the relevance of the proposed data-driven
scheme for the real missing data patterns of the high-resolution infrared METOP sensor. It has resulted in a significant
improvement w.r.t. state-of-the-art techniques in terms of interpolation error (about 50% of relative gain) and spectral
characteristics for horizontal scales smaller than 100km. We further discuss the key features and parameterizations
of the proposed data-driven approach as well as its relevance with respect to classical interpolation techniques.
Index Terms
Ocean remtote sensing data, data assimilation, optimal interpolation, analog and exemplar-based models, multi-
scale decomposition, patch-based representation
Satellite-derived products are of key importance for the high-resolution monitoring of the ocean surface on a
global scale. A variety of sensors record observations of geophysical parameters, such as Sea Surface Temperature
(SST) [1], Sea Surface Height (SSH) [2], Ocean Color [3], Sea surface Salinity (SSS) [4], etc. In all cases, the
delivery of L4 gridded products for end-users involves a number of pre-processing steps from the L1 data acquired
and transmitted by spaceborne sensors. Due to both the space-time sampling geometry of satellite sensors and their
sensitivity to the atmospheric conditions (e.g., rains, aerosols, clouds), ocean remote sensing data may involve very
large missing data rates as illustrated in Fig.3. Hence, spatio-temporal interpolation is of key importance to deliver
gap-free gridded sea surface fields for further analysis.
R. Fablet, P.H. Viet and R. Lguensat are with IMT Atlantique; Lab-STICC, Brest, France, e-mail:
September 6, 2017 DRAFT
Optimal interpolation is certainly the state-of-the-art approach for the spatio-temporal interpolation of satellite-
derived sea surface geophysical fields [5], [6]. Optimal interpolation relies on the modeling of the covariance of the
considered spatio-temporal fields. The choice of the covariance model is a critical step [7], [6], [8], [9]. Stationary
covariance hypotheses are generally considered, though they might not be verified. For instance, frontal areas as
illustrated in Fig.5 may involve time-varying and space-varying anisotropical features. In such cases, considering
mean covariance model typically results in the smoothing out of the fine-scale SST details. Data assimilation
techniques for missing data interpolation may be regarded as another important category of model-driven approaches
([10], [8], [11]). A critical aspect of their implementation lies in the choice of the dynamical model, more precisely
the trade-off between its computational complexity and its ability to correctly represent real sea surface dynamics.
The tremendous amount of satellite observation data pouring from space, along with the wider availability of
reanalysis and/or numerical simulation datasets supports the development of data-driven approaches as an alternative
to model-driven schemes. In this respect, statistical and machine learning models offer new computational means
to account for space-time variabilities that cannot be completely captured by simplified physical models. The
application of Principal Component Analysis (PCA), also referred to as Empirical Orthogonal Functions (EOF) in
the geoscience field, to remote sensing missing data interpolation [12], [13] may be regarded as an example of
such data-driven schemes, though it proves mainly relevant for large-scale variabilities [13]. One may also cite the
development of exemplar-based models in image processing and their applications to missing data interpolation for
single-date remote sensing data [14], [15].
In this study, we investigate such data-driven and exemplar-based models for the spatio-temporal interpolation
of missing data in ocean remote sensing time series. We aim to exploit the implicit knowledge conveyed by
available multi-annual satellite-derived datasets to improve the interpolation of high-resolution spatio-temporal sea
surface geophysical fields. We rely on analog data assimilation [9], [16] and develop, to our knowledge, the first
application of analog data assimilation to high-dimensional spatio-temporal fields. Our methodological contributions
lie in the introduction of a multiscale analog data assimilation applied to local patch-based and PCA-constrained
representations. We demonstrate the relevance of the proposed scheme through an application to SST time series.
We report significant gain compared to state-of-the-art approaches, namely optimal interpolation [17], [7] and
PCA-based interpolation [12], [13].
This paper is organized as follows. Section II reviews the related work. Section III presents the proposed multiscale
analog data assimilation scheme. Numerical experiments are reported in Section IV and we further discuss our key
contributions and future work in Section V.
A. Model-driven approaches
As previously mentioned, model-driven approaches are the state-of-the-art techniques for the spatio-temporal
interpolation of missing data in ocean remote sensing observations [6], [11]. In particular, optimal interpolation
relates to the following formulation:
X∝ G(Xb,Γ) (1)
September 6, 2017 DRAFT
Y(t, s) = X(t, s) + (t, s),st(2)
where G(Xb,Γ) is a spatio-temporal Gaussian field with mean background field Xband covariance function Γ, and
the observation noise assumed to be Gaussian. trefers to the region domain for which observations are truly
available at time t. Given a series of observation fields Yand a known covariance function Γ, optimal interpolation
leads to an analytical MAP (Maximum A Posteriori) solution for field X, equivalent to the minimization of a
reweighted least-square criterion w.r.t. the covariance of noise . The choice of the covariance function Γis a
critical step. Exponential and Gaussian covariance models [8], [7] are the most classical choices with both constant
parameters as well as space-time-varying parameterization [18].
When dealing with high-dimensional fields, such as ocean remote sensing observations, the numerical computation
of the solution of the optimal interpolation may not be feasible, as it involves the inversion of a very large covariance
matrix. Sequential approaches, such as ensemble Kalman techniques [11], are then considered. They may be restated
as data assimilation formulations. Considering a discrete setting, they amount to the following model for field X:
X(t) = M(X(t1), η(t1)) (3)
where Mis referred to as the dynamical model and ηis a random perturbation. Model (1) may be restated according
to this formulation with a linear model Mand a Gaussian process ηderived from the considered Gaussian field
with covariance Γ. Other parameterizations of the dynamical model may be derived from fluid dynamics equations,
including for instance advection-diffusion models [10]. Ensemble Kalman schemes [11] are the state-of-the-art
techniques to numerically solve for the reconstruction of spatio-temporal field Xgiven partial observation field Y
under model (3). Using a sample-based representation of Gaussian distributions, they provide forward-backward
filtering schemes to approximate the optimal interpolation solution. We let the reader refer to [11] and reference
therein for additional details on stochastic data assimilation. We may also point out variational data assimilation
[19], [10], which exploits a continuous formulations of Model (3) and involves a gradient-based minimization of
the observation error under model (3).
A typical example of the optimal interpolation of an SST field from a series of partial observations is reported in
Fig.3. An important limitation of model-driven approaches lies in modeling uncertainties. Due to the autocorrelation
structure of sea surface geophysical structures and the observation sampling rate, optimal interpolation results to
accurate reconstruction of the spatio-temporal fields for spatial scales larger than 100km. However, finer scales are
significantly filtered out (see Fig.3). This property directly relates to the correlation length of the covariance model
(here, 100km). This correlation is a trade-off between the spatial resolution of the observation fields (here, 5km)
and the size of the gaps.
As detailed below, we explore data-driven approaches to take advantage of available observation or simulation
datasets with a view to improving the reconstruction of the fine-scale structures of sea surface fields.
B. Data-driven approaches
With the increasing availability of representative observation datasets, data-driven models become more and more
appealing to solve inverse image problems, including missing data interpolation. Initially mostly investigated for
September 6, 2017 DRAFT
computer vision and computer graphics applications, such as synthesis, inpainting and super-resolution issues [20],
[21], they have also gained interest for applications to remote sensing data [14], [15]. Patch-based and exemplar-
based models have emerged as powerful representations to project images onto large sets of patch exemplars
and/or dictionaries. Non-local means and non-local priors [22], [23] are state-of-the-art examples of such models
for image reconstruction issues. Developments for multivariate time series have also recently been investigated,
especially exemplar-driven data assimilation referred to in the geoscience field as analog data assimilation [9],
[16], [24]. Two main features make the direct application of these exemplar-based strategies to spatiotemporal
fields poorly efficient: their computational complexity and their ability to jointly capture large-scale and fine-scale
structures. Patch-based techniques generally involve small image patches (typically, from 3x3 to 11x11 patches for
2D images), which cannot resolve large structures, with a typical scale greater than the width of the patches. In
addition, the considered minimization schemes involve repeated iterations over the entire set of exemplars, which
may make them extremely computationally-demanding for applications to spatio-temporal data. By contrast, analog
data assimilation provides an efficient sequential scheme, but remains limited to relatively low-dimensional space
(up to a few tens of dimensions in [9], [16]).
PCA-based models are popular in the geoscience field. They have also gained interest for application to missing
data interpolation, especially DINEOF approaches [12], [13]. These involve two key steps: i) the estimation of basis
functions, which provide a lower-dimensional representation of the variability spanned by the considered spatial or
spatio-temporal data, ii) the interpolation of the missing data from projections onto the basis functions. VE-DINEOF
[13] has recently improved compared to the original DINEOF scheme [12]. In both cases, applications to ocean
remote sensing data, especially SST, were considered. Applied on a global or regional scale, the lower-dimensional
PCA-based representation is mostly relevant to recover large-scale structures and not as appropriate to reconstruct
fine-scale details. Overall, PCA-based decompositions are regarded as relevant representations to encode the spatial
patterns exhibited by geophysical fields. It may be noted that PCA representations are also often used in patch-based
image processing (see for instance [28], [25]).
From the above review of the related work, a key issue in developing a data-driven framework is the decomposition
and representation of the spatio-temporal variabilities of the considered fields using appropriate models both in terms
of computational efficiency and modeling uncertainties, especially for fine-scale details. This naturally calls for a
multi-scale representation. Formally, we considered a three-scale model where field Xwas decomposed as follows:
where ¯
Xrefers to the large-scale (low-frequency) component of X,dX1,2to details at two scales and ξto unresolved
scales. By convention, dX1refers to the intermediate scale and dX2to the finest scale. The spatio-temporal
interpolation of field Xresults in the estimation of components ¯
Xand dX1,2in (4) given Observation Model (2).
Following Formulation (3), this led to the definition of relevant dynamical priors for each component. For large-
scale component ¯
X, a model-driven representation associated to optimal interpolation naturally arises as a relevant
September 6, 2017 DRAFT
representation given its efficiency in recovering large-scale structure. We thus assume that ¯
Xis a spatio-temporal
Gaussian process with mean ¯
Xband spatio-temporal covariance Γ. Here, ¯
Xbis given by the climatological mean
over the entire times series and we therefore considered a Gaussian covariance model. We let the reader refer to
[26], [17], [27] for additional details on optimal interpolation using a Gaussian prior for Γ. By contrast, for detail
components dX1,2, we introduced data-driven spatio-temporal priors as detailed in the next section.
A. Multi-scale data-driven priors
The definition of detail fields dX1,2combines patch-based and PCA-based representations. For scale i= 1 or 2,
let us consider Pi×Pipatches, such that P1> P2(typically P1= 40 and P2= 20). We proceed as follows for
the scale i= 1. Given ¯
Xin multi-scale decomposition (4), each P1×P1patch of detail field dX1is given by the
projection of the associated patch for residual field X¯
Xonto a low-dimensional PCA decomposition. This PCA
decomposition is learnt from P1×P1patches of a training dataset of residual fields X¯
X. We apply the same
procedure for detail field dX2from residual field X¯
XdX1. Formally, this leads to the following definition of
detail fields dX1,2:
where P1,2are patch-based PCA image projection operators [28], [25]. They result in the decomposition of any
patch Psaround point sat time tof detail field dXias a linear combination of the principal components of the
PCA for scale i:
dXi(Ps, t) =
αi,k(s, t)Bi,k (6)
with Bi,k the kth principal component of the PCA at scale iand αi,k(s, t)the associated coefficient for patch Ps
at time t.NP CA,i refers to the number of vectors of the PCA basis at scale i. The spectral properties of PCA
decompositions along with the lower patch size at scale i= 2,i.e. P1> P2, lead to a scale-space decomposition
[31]. Contrary to a wavelet decomposition, we only implicitly set the considered scale ranges through the number
of principal components kept at each scale. The key interest here is a local adaption with point-specific PCA bases
which can also account for any image geometry (e.g., the presence of land points in the considered region).
Given these definitions for detail fields dX1,2, we considered an analog (data-driven) formulation of the associated
dynamical models (3). As stated in Section II, analog dynamical models introduced in [9], [24] do not directly apply
to high-dimensional fields and we considered patch-based models. We first assumed that we were provided with
representative catalogs C1,2of patch exemplars of the dynamics of details fields dX1,2. Each catalog is composed
of a set of patch exemplars {dXi(Psk, tk)}k, referred to hereafter as analogs, and of their temporal successors
{dXi(Psk, tk+ 1)}k. For a given patch Psand scale i, the definition of the analog dynamical model leads to the
definition of an exemplar-driven sampling strategy for the distribution of the state at time t,dXi(Ps, t), conditionally
to the state at time t1,dXi(Ps, t 1). Formally, we considered Gaussian conditional distributions of the form
dXi(Ps, t)|dXi(Ps, t 1) = u G (µi(u, Ci),Σ (u, Ci)) (7)
September 6, 2017 DRAFT
where G(·)is a Gaussian distribution. Mean µi(u, Ci)is defined as a weighted function of the successors of the
Knearest-neighbor of uin catalog Ci. Similarly, covariance Σ (u, Ci)is issued from the weighted covariance of
the successors of the Knearest neighbors. Theses weights and the nearest-neighbor search involve a predefined
kernel Kas detailed below. Let us denote by (Ak(u),Sk(u)) the analog-successor pair of the kth nearest-neighbor
to uin Ci. Following [9], [24], we investigate three different analog dynamical models corresponding to different
parameterizations of the above mean and covariance:
Locally-constant analog model: mean µi(u, Ci)and covariance Σ (u, Ci)are given by the weighted mean
and covariance of the Ksuccessors {Sk(u)}k.
Locally-incremental analog model: it proceeds similarly to the locally-constant analog model, but for the
differences between the successors and the analogs, such that mean µi(u, Ci)is given by the sum of uand of
the weighted mean of the Kdifferences {Sk(u)− Ak(u)}k.Σ (u, Ci)results in the weighted covariance of
these differences.
Locally-linear analog model: given the K analog-successor pairs {Ak(u),Sk(u)}k, it first comes to the
weighted least-square estimation of the linear regression of the state at time tgiven the state at time t1.
Denoting by Ai(s, t)the estimated local linear operator, mean µi(u, Ci)is given by Ai(s, t)·dXi(Ps, t 1)
and covariance Σ (u, Ci)by the weighted covariance of the residuals of the fitted linear regression.
KG(u(t), v(t)) = exp ku(t)v(t)k2
and a cone kernel KC, recently introduced for dynamical systems in [30]. For any pair of states u(t), v(t), it leads
KC(u(t), v(t)) = exp Lζ(u(t), v(t))
Lν(u, v) = kω(t)k21ζcos2θ1ζcos2φ1/2
where ω(t) = u(t)v(t),tu(t) = u(t)u(t1),tv(t) = u(t)u(t1),cosθ =hω(t), du(t)iand
cosφ =hω(t), dv(t)i. Compared to a classical Gaussian kernel, the cone kernel takes into account not only the
distance between the two states, but also the alignment of their instantaneous velocities with the difference between
the two states. It has been shown in [30] that the cone kernel may be more appropriate for analog forecasting
schemes. For the Gaussian (resp. cone) kernels, scale parameter σis locally-adapted to the median value of the
distances ku(t)v(t)k2(resp. Lν(u(t), v(t))) to the K nearest neighbors in the catalogs of exemplars. Parameter
νis set empirically between 0 and 1. In all cases, we take advantage of the considered PCA-based representation
of the patches to compute patch similarities within the associated low-dimensional spaces, and not in the original
patch space.
B. Numerical resolution
Given the data-driven priors M1,2, we proceed to the resolution of model (4) according to a MAP criterion.
We might consider a direct discrete gradient-based numerical resolution as the considered parameterization for
September 6, 2017 DRAFT
model (4) can be regarded as a spatio-temporal Markov Random field [14], [29]. This would however lead to
an extremely-demanding computational scheme. We prefered to exploit the multi-scale nature of our model to
develop a coarse-to-fine strategy and cast the global minimization problem as series of smaller problems, which
can be solved more efficiently. More precisely, we proceeded as follows. We first solved for the reconstruction of
large-scale component ¯
Xusing optimal interpolation with covariance model Γ. We then successively solved for
the reconstruction of detail fields dX1,2. This step runs independent resolution along the temporal dimension for
each patch position using sequential data assimilation algorithms, namely an analog Ensemble Kalman Smoother
(AnEnKS) and an HMM-based analog smoother (AnHMM). We refer the reader to [24] for additional details on the
implementation of the analog EnKS. The independent solutions computed for each patch position were recombined
using averaging. To reduce the computational complexity, we did not process all possible patch positions, but only
overlapping patches with a 35x35 (resp. 15x15) spatial sampling for P1×P1patches (resp. P2×P2). To remove
potential block artifacts, we apply a PCA-based filtering step onto 10 ×10 patches. As initialization for the analog
data assimilation iterations, we use a VE-DINEOF solution [13].
All implementations were run under Python. We used [5] for optimal interpolation, and Analog Data Assimilation
toolbox [24]. The Python code of the proposed patch-based and multi-scale analog data assimilation is made available
( ANDA).
A. Experimental setting
Considered case-study: To perform a qualitative and quantitative evaluation of the proposed framework, we used
a reference gap-free L4 SST time series from which we create a SST with missing data using real missing data
masks. As reference SST, we used OSTIA product delivered daily by the UK Met Office[6] with a 0.05spatial
resolution (approx. 5km) from January 2007 to April 2016. The OSTIA analysis combines satellite data provided
by infrared sensors (AVHRR, AATSR, SEVIRI), microwave sensors (AMSRE, TMI) and in situ data from drifting
and moored buoys. For the missing data mask series, we studied an infrared sensor, more specifically METOP,
which may involve very high missing data rates as illustrated in Fig.3 & 5.
As a case-study region, we selected an area off South Africa. This highly dynamic ocean region involves complex
fine-scale SST structures (e.g., filaments, fronts) as shown in Fig.3. Our evaluation focused on the interpolation of
the SST fields for year 2015, other years being used to build a catalog of exemplars for the analog frameworks. The
Python code used for the creation of the considered SST data is available ( TCI rfablet).
Parameter setting of the proposed approaches: We performed interpolation experiments with both AnHMM
and AnEnKF/KS schemes (see Section III for details). We exploited a three-scale model: the global scale (entire
region), 40x40 patches and 20x20 patches. At each scale, each patch was encoded by its PCA-based decomposition
using a 10-component PCA. As initialization for missing data areas, we used an optimal interpolation on the
global scale. The parameterizations of the optimal interpolation and of the DINEOF scheme were those used for
comparison purposes as detailed below. In the analog setting, the number of neighbors was varied from 10 to 110
and we compared Gaussian and Cone kernels.
September 6, 2017 DRAFT
Comparison to state-of-the-art approaches: For comparison purposes, we consider an optimal interpolation,
which is the interpolation technique used in most operational products (e.g., [6]), VE-DINEOF [13], a PCA-based
technique, and a direct region-level application of the analog data assimilation. Their parameter settings were as
Optimal interpolation (OI): we used a Gaussian kernel with a spatial correlation length of 100km and a temporal
correlation length of 3 days. These parameters were empirically tuned for the considered dataset using a cross-
validation experiment. We used the optimal interpolation package from [5]. The considered parameter setting
was consistent with previous work [6], [8] and stressed the strong temporal correlation of SST field [8]. In
our case-study, a direct implementation of the OI would have required a large memory: for a missing data
rate of 70%, the interpolation onto the considered 300 ×600 grid would have required the inversion of a
system of 5T.104equations with T the temporal correlation. Given the considered spatial correlation length
of 100km, we achieved an optimal interpolation onto a coarser grid with a resolution of 25km and applied a
bicubic interpolation onto the targeted high-resolution grid (5km resolution).
VE-DINEOF interpolation: we exploited a direct implementation of VE-DINEOF scheme [13] on the regional
scale using 200 PCA components, which amounted to 99.27% of the total variance of the dataset. This
VE-DINEOF setting is referred to as G-VE-DINEOF. We also considered a multi-scale version of the VE-
DINEOF procedure using the same three-scale decomposition as the multi-scale analog data assimilation. As
for MS-AnEnKS and MS-AnHMM, we used two detail components corresponding to 40x40 patches and 20x20
patches. At each scale, i.e. the coarse region scale and the two detail scale, we exploited 10-dimensional PCA
decomposition (NP CA,1=NP C A,2= 10). The resolution of this multi-scale VE-DINEOF, referred to as MS-
VE-DINEOF, applies a coarse-to-fine strategy, such that at each scale, the VE-DINEOF iteratively updated the
missing data area from the projection of overlapping patches onto the 10-dimensional PCA basis;
Global AnEnKS interpolation: to evaluate the relevance of the proposed multi-scale decomposition, we tested
a direct application of the AnEnKS at the region scale, referred to as G-AnEnKS. Similarly to G-VE-DINEOF,
we considered 200 PCA components, which amounted to 99.27% of the total variance of the dataset. From
numerical experiments, the best parameter setting combined a locally-incremental analog forecasting with
K= 100 neighbors and a Gaussian kernel.
It may be noted that variational interpolation techniques, based on the minimization of regularization norms [32],
cannot be expected to lead to relevant results given the large missing data rates in the considered dataset (above
70% on average) and were not considered in our experiments.
Qualitative and quantitative evaluation: to assess the quality of the different interpolation schemes, we first
achieved a quantitative analysis according to mean square error (MSE) statistics for the SST reconstructed SST
fields, the associated gradient fields, and the detail fields of a 4-scale dyadic wavelet decomposition of the SST fields.
We also computed radially-averaged power spectral densities to analyze the fine-scale patterns of the reconstructed
field. In addition, we performed a qualitative analysis of these fields with a focus on the reconstruction of fine-scale
September 6, 2017 DRAFT
B. Interpolation performance
We shall begin with the results of our numerical experiments. We first present the quantitative evaluation of
interpolation performance, including a comparison to state-of-the-approaches. Second, we further illustrate this
performance using interpolation examples. Third, we report a sensitivity analysis of the best analog assimilation
setting. We also include an evaluation of interpolation performance when the creation of the catalogs of exemplars
involve observation datasets with missing data.
Quantitative comparison to state-of-the-art approaches: We first report the overall MSE statistics of the
considered interpolation approaches, namely OI, G-VE-DINEOF, MS-VE-DINEOF, MS-AnHMM, G-AnEnKS and
MS-AnEnKS, in Tab.I. The multi-scale Analog schemes are a clear improvement over the OI and VE-DINEOF
reconstruction, with a relative gain in SST RMSE up to 50% for MS-AnEnKS at the finest scale (dX2). MS-
AnHMM also leads to a significant improvement but is clearly outperformed by MS-AnEnKS. It may be noted
that the direct application of the analog data assimilation, G-AnEnKS, to field Xdoes not lead to very significant
improvement. This is regarded as a direct benefit of the multi-scale decomposition, which greatly increases the
representativity of the collected catalogs of exemplars. No such difference is reported for the application of global
and multi-scale VE-DINEOF schemes, which further stresses the relevance of the analog dynamical prior exploited
by MS-AnEnKS. The analysis of the MSE statistics at different scales of a dyadic wavelet decomposition indicates
that the improvement mainly refers to the third and fourth dyadic scales (i.e., spatial scales greater than 20km).
Most of the improvement is brought about by the resolution of component dX1(about 40% of relative gain w.r.t.
OI), when component dX2accounts for about 10% of relative gain w.r.t. OI. The MSE time series (Fig.1) lead
to similar observations. Interestingly, AnEnKS depicts a lower time variability of the MSE compared to OI and
VE-DINEOF (standard deviation of 0.06 vs. 0.13), the later being more sensitive to larger missing data rates. This
is viewed as a benefit of the exemplar-based time regularization conveyed by the analog framework.
Qualitative analysis of interpolation results from examples: To complement this global analysis, we report
interpolation results for two dates, corresponding to relatively low (60%) and greater (90%) missing data
rates, respectively in Fig. 5 and Fig. 3. For these two examples, we visually compare OI, MS-VE-DINEOF and
MS-AnEnKS interpolations to the groundtruth both for the SST field and the gradient magnitude fields. In Fig.3,
MS-AnEnKS clearly outperforms OI and MS-VE-DINEOF (SST (resp. SST gradient) rmse of 0.20 (resp. 0.24) vs.
0.42 (resp. 0.40) and 0.41 (resp. 0.40)). We also highlight areas in which the improvements in the reconstruction of
local SST details may be noticed. Visually, the improvement is more noticeable on the gradient amplitude. Whereas
OI and MS-VE-DINEOF lead to relatively coarse SST structures, MS-AnEnKS results in finer front details, which
are visually more similar to the groundtruth. This is further emphasized by the analysis of the power spectral densities
of the different fields (Fig. 6, left). OI clearly underestimates the spectral energy below 100km, as expected from
the associated spatio-temporal smoothing with a spatial correlation length of 100km. A similar underestimation
is observed for MS-VE-DINEOF for scales ranging between 70km and 150km. By contrast, MS-AnEnKS nicely
matches the spectral signature of the groundtruth up to 20km. These results appear consistent with the previous
observation that the improvement brought about by the analog assimilation was mainly noticeable in terms of MSE
September 6, 2017 DRAFT
for scales greater than 20km. The white noise plateau observed from 20km and below for the reference SST field
may indicate that the OSTIA field conveys little information for scales lower than 20km for this particular date.
This is further illustrated by the analysis of a one-dimensional transect at 36.525oS accross a strong SST front in
Fig.4. The MS-AnEnKS interpolation clearly leads to a better estimation of local SST variabilities, where OI and
MS-VE-DINEOF tends to oversmooth strong gradients. Overall, the same observation holds for the second example
(Fig.5), though the lower missing data rate (59%) slightly reduces the differences observed between the different
interpolation methods.
We also illustrate the relevance of the post-processing step in the AnEnKS (Fig.2). The spatially-independent
assimilation of overlapping patches may result in block artifacts at patch boundaries as clearly highlighted by
the gradient field. The considered EOF-based filtering for 10 ×10 patches successfully removes most of these
block artifacts and retrieves a visually consistent gradient field as discussed above. It may be noted that a different
implementation of the analog assimilation using non-sequential iterative scheme for patch-based image processing
[22], [14] would be an alternative, however at the expense of an increased computational complexity. By contrast,
the independent assimilation of each spatial patch only involves one forward and one backward iteration, such
that each space-time patch is visited only twice. We evaluate more precisely the computational complexity of the
different interpolation models in Tab.VI. MS-VE-DINEOF is clearly involves the lowest computational complexity.
In this respect, given relatively similar interpolation performance, VE-DINEFO appears as a relevant alternative to
OI for the interpolation of the coarse-scale component. By contrast, even if MS-AnHMM significantly reduces the
computational complexity of the analog assimilation, the differences in interpolation performance reported in Tab.I
clearly recommend the selection of the MS-AnHMM as the relevant fine-scale analog assimilation scheme for SST
Sensitivity analysis for MS-AnEnKS: Given the overall qualitative and quantitative analysis reported above,
we further analyze the MS-AnEnKS setting, especially its sensitivity to the selected parameter setting. In Tab.II
We report relative MSE statistics while varying the number of neighbors in the analog models. Tab.III reports a
similar analysis for different kernel parameterizations. Overall, the best parameterization combines a cone kernel
[30] using 100 neighbors and a locally-incremental analog model. It might be noted that the choice of the kernel
weakly affects interpolation performance. By contrast, the locally-incremental analog model significantly improves
the relative MSE of the locally-linear and locally-constant strategies (Tab.IV) by about 10% and 25%. This is in
accordance with the conclusions drawn in [24]. The lower performance of the locally-linear analog model may
relate to an unfavourable trade-off between estimation uncertainty and local adaption. We may point out that all
these parameterizations of the proposed interpolation framework outperforms both OI and MS-VE-DINEOF.
Creation of catalog Cfrom observation datasets: In the experiments reported above, the catalog of patch
exemplars is built from the gap-free SST time series from 2008 to 2014. This experimental setting is representative
of an application context where one aim to exploit previous reanalyses and/or numerical simulations for the
interpolation of upcoming observations. The key interest of the analog assimilation is to facilitate the implicit
synergy between possibly computationally-expensive high-resolution models and/or reanalyses and satellite-derived
observation datasets. A second application context is also investigated. We may also directly build the catalog of
September 6, 2017 DRAFT
50 100 150 200 250 300 350
RMSE of SST field on missing areas
50 100 150 200 250 300 350
RMSE of gradient magnitude
Fig. 1: Time series of the relative MSE: OI (black,-), VE-DINEOF (blue,-) and AnEnKS (red,-) for the estimated
SST fields (left) and gradient magnitude fields (right)
Fig. 2: Illustration of the postprocessing step for the removal of blocky artifacts: gradient magnitude field of the
an interpolated SST field using MS-AnEnKS before (a) and after (b) the application of the considered PCA-
based postprocessing step with 10×10 patches. We also report the radially-averaged power spectral density of the
interpolated SST fields w.r.t. the true SST field (GT, black-).
exemplars from the satellite-derived observation datasets, which involve missing data. To simulate this experiment,
we created a representative catalog from the SST time series with the METOP missing data mask from 2008
to 2014. We proceeded similarly to the scheme described for year 2015 in Section III.3. We only retained SST
patches with less than 20% of missing data. We compared the resulting interpolation performance to that of the
first experiment in Tab.V. Although lower MSE values are reported for this second experiment (0.22 vs. 0.20 in
terms of mean relative MSE of the interpolated SST fields), the relative gain compared to OI and VE-DINEOF
is still significant (0.22 vs. respectively 0.40 and 0.41). The qualitative analysis of the interpolated fields leads
to conclusions similar to those drawn for the first experiment. These results further stress the relevance of the
proposed data-driven approach in order to benefit either from high-resolution simulations and/or re-analyses or real
satellite-derived observation datasets. It may be noted that our multi-scale approach may also allow us to combine
observation datasets from different sensors [14].
September 6, 2017 DRAFT
TABLE I: Comparison of global interpolation performance: mean relative MSE of OI, G-VE-DINEOF, MS-VE-
DINEOF, G-AnEnKS and MS-AnEnKS: we report mean relative MSE statistics in terms of the SST fields, the
gradient magnitude of the SST fields and of the detail coefficients for a four-level dyadic wavelet decomposition.
For MS-ANEnKS, we report both the interpolation performance at intermediate scale i= 1 (MS-ANEnKS|dX1),
i.e. with dX2= 0 in (4), and at scale i= 2 (MS-ANEnKS|dX2). We let the reader refer the main text for details
on the associated parameter setting of the different interpolation models.
Criterion SST k∇k k=1 k=2 k=4 k=8
OI 0.4157 0.3986 0.0053 0.0212 0.0897 0.1897
G-VE-DINEOF 0.4064 0.3967 0.0124 0.0221 0.0873 0.1969
MS-VE-DINEOF 0.4052 0.3765 0.0052 0.0192 0.0803 0.1697
G-AnEnKS 0.3842 0.3922 0.0120 0.0219 0.0967 0.1902
MS-AnHMM dX20.3350 0.3529 0.0057 0.0208 0.0838 0.1711
dX10.2536 0.3349 0.0057 0.0212 0.0848 0.1622
dX20.2009 0.2357 0.0053 0.0173 0.0579 0.1067
TABLE II: Influence of the number of analogs on MS-AnEnKS performance: mean relative MSE of MS-AnEnKS
interpolation w.r.t. the number of analogs for the three considered analog strategies (7).
Number of analogs (K)10 20 30 40 50 60 70 80 90 100 110
Locally-constant 0.2746 0.2778 0.2822 0.2852 0.2884 0.2904 0.2926 0.2948
Locally-Linear 0.2449 0.2369 0.2325 0.2301 0.2288 0.2280 0.2278 0.2271 0.2266 0.2266
Locally-incremental 0.2119 0.2113 0.2083 0.2051 0.2030 0.2028 0.2020 0.2012 0.2009 0.2009 0.2011
TABLE III: Influence of the kernel on MS-AnEnKS performance: mean relative MSE of the interpolated SST fields
using different kernel parameterizations using a Gaussian kernel and a cone kernel [30].
Gaussian Cone ζ=0.995 Cone ζ=0.5 Cone ζ=0
0.2030 0.2028 0.2036 0.2009
TABLE IV: MS-AnEnKS performance depending on the selected analog model (7): we let the reader refer to Tab.I
for the description of the considered evaluation criteria
Criterion SST k∇k k=1 k=2 k=3 k=4
Locally-constant 0.2725 0.3214 0.0063 0.0208 0.0783 0.1529
Locally-Linear 0.2245 0.2730 0.0059 0.0186 0.0637 0.1265
Locally-Incremental 0.2009 0.2357 0.0053 0.0173 0.0579 0.1067
September 6, 2017 DRAFT
TABLE V: Influence of missing data in catalogs C1,2: we let the reader refer to Tab.I for the description of the
considered evaluation criteria.
Criterion SST k∇k k=1 k=2 k=3 k=4
Catalogs C1,2built from gap-free 2008-2014 data 0.2009 0.2357 0.0053 0.0173 0.0579 0.1067
Catalogs C1,2built from 208-2015 dataset with missing data 0.2230 0.2643 0.0056 0.0194 0.0653 0.1212
TABLE VI: Computational complexity of the interpolation models evaluated in Tab.I
Exe. time 3.5h 0.5h 1.2h 3h
We presented a novel data-driven model for the spatio-temporal interpolation of satellite-derived SST fields. To our
knowledge, this study reported the first application of the analog data assimilation framework to high-dimensional
satellite-derived geophysical fields. We demonstrated its relevance with respect to state-of-the-art techniques, namely
optimal interpolation [6] and a PCA-based matrix completion scheme [12], [13]. Our model significantly outperforms
these two techniques in terms of reconstruction error, especially for fine-scale structures in the range [20km, 200km].
The considered case-study involves real missing data patterns from the METOP-AVHRR sensor. It is therefore
representative of the irregular space-time sampling of the sea surface associated with infrared satellite sensors. The
relative gain in the mean interpolation MSE of about 50% stresses the potential of data-driven computational models
in the exploitation of large-scale observation datasets to improve the reconstruction of geophysical fields from partial
satellite-derived observations. We have made our case-study dataset available as a supplementary material to our
paper with a view to favoring the benchmarking of interpolation methods for satellite-derived geophysical products1.
As demonstrated by our experimental evaluation, the first key feature of the proposed model is the use of a multi-
scale decomposition. Whereas a classic model-driven interpolation (OI) applies to the coarse-scale component, the
reconstruction of the fine-scale components exploit the analog data assimilation [24]. A critical aspect of analog
methods is the availability of a representative catalog of exemplars. In this respect, the considered multi-scale
decomposition is regarded as a crucial means to stationarize the fine-scale spatial variabilities depicted by sea
surface geophysical fields and make more relevant exemplar-based representations of these variabilities. Wavelet
analysis is generally the classic scheme to derive a multi-scale decomposition [31]. Here, we exploited PCA-based
representations for different patch sizes, so that we naturally combined a multi-scale decomposition to a low-
dimensional representation of the spatial variabilities on each scale. Such PCA-based representations also efficiently
deal with complex image geometries (e.g., the presence of land areas in the considered ocean case-study region).
The second critical feature of the proposed multi-scale analog assimilation is the combination of a relevant analog
1The Python code used for the creation of the considered SST data is available at: TCI rfablet
September 6, 2017 DRAFT
Fig. 3: Reconstruction of a SST field on June, 30, 2015 with a large missing data rate (87%): (a) first row, reference
SST field (groundtruth (GT)), its associated gradient magnitude, observed field; second row, interpolated fields by
OI, MS-AnEnKS, MS-VE-DINEOF; third row, gradient magnitude of the fields depicted in the second row.
5 10 15 20 25 30
Longitude(°E) ( Latitude = 36.525°S )
SST (°C)
5 10 15 20 25 30
Longitude(°E) ( Latitude = 36.525°S )
Fig. 4: Analysis of a SST transect at 36.525oS for the interpolation results depicted in Fig. 3: we depict a one-
dimensional profile at latitude 36.525oS (c) for both the SST (bottom) and the SST gradient magnitude (top)
for the reference SST field (black,-) as well as OI (magenta,-), MS-VE-DINEOF (blue,-) and MS-AnEnKS (red,-)
interpolated SST fields.
forecasting strategy, namely a locally-incremental strategy, with a generic sequential assimilation algorithm, namely
an Ensemble Kalman Smoother. By contrast, the choice of the kernel and the number of analogs seem to be only
of secondary importance. It may be noted that Model (4) could be straightforwardly extended to a greater number
of scales. For the considered case-study however, numerical experiments did not lead to significant improvements
with 3 or 4 detail scales.
We believe that this study opens new research avenues for the development of new data-driven models for
September 6, 2017 DRAFT
Fig. 5: Reconstruction of an SST field on February, 19, 2015 with a relatively low missing data rate (56%): see
Fig.3 for details.
Wavelength (km)
Fourier power spectrum
Radially−averaged PSD
Fig. 6: Spectral analysis of interpolation results depicted in Fig.3 and 5: we report the radially-averaged power
spectral densities of the reference SST field (black,-) as well as OI (magenta,-), MS-VE-DINEOF (blue,-) and
MS-AnEnKS (red,-) interpolated SST fields for June, 30, 2015 (left) and February, 19, 2015 (right).
the reconstruction of upper ocean dynamics from satellite-derived observations, in the same way that data-driven
schemes have led to major advances in other imaging domains such as photography, microscopy, astronomy....
The exploitation of analogs for interpolation may be interpreted in a climatological sense, the key idea being that
previously observed fine-scale geophysical variabilities will probably occur again, though not necessarily with the
same seasonal timing. The application to other sea surface tracers, such as ocean color, is then natural [33]. The
proposed multi-scale analog assimilation also seems particularly appealing for the downscaling of low-resolution
September 6, 2017 DRAFT
satellite-derived products, such as sea surface salinity [34] and sea surface height [35]. From a methodological
point of view, multimodal extensions would be of interest to account for multi-sensor observations as well synergies
between different tracers [36], [14].
Analog strategies are particularly appealing when large and representative observation datasets are available, as
illustrated in the case-study considered here. By contrast, one may question their relevance in addressing scarce
observation datasets as well as extreme events, which are by essence rare events. In this context, the creation of
catalogs of analogs from realistic high-resolution numerical simulations [37], [38], which are becoming increasingly
available, appears to be a relevant path to be further investigated in future work.
This work was supported by ANR (Agence Nationale de la Recherche, grant ANR-13-MONU-0014), Labex
Cominlabs (grant SEACS) and TeraLab (grant TIAMSEA). We thank B. Chapron, P. Ailliot, P. Tandeo and J.F.
Piolle for discussing and commenting on the development of analog assimilation models for SST data interpolation.
We are also gratefull to A. Northan for a thorough proofread of our manuscript.
[1] D. Chelton and F. Wentz, “Global Microwave Satellite Observations of Sea Surface Temperature for Numerical Weather Prediction and
Climate Research - ProQuest,” Bulletin of the American Meteorological Society, vol. 86, no. 8, p. 1097, 2005.
[2] D. B. Chelton, J. Ries, B. J. Haines, L.-L. Fu, and P. S. Callahan, “Satellite Altimetry,” in International Geophysics, ser. Satellite Altimetry
and Earth SciencesA Handbook of Techniques and Applications, A. Cazenave and L.-L. Fu, Eds. Academic Press, 2001, vol. 69, pp.
[3] K. Baith, R. Lindsay, G. Fu, and C. McClain, “Data analysis system developed for ocean color satellite sensors,” Eos, Transactions
American Geophysical Union, vol. 82, no. 18, pp. 202–202, 2001.
[4] V. Klemas, “Remote Sensing of Sea Surface Salinity: An Overview with Case Studies,Journal of Coastal Research, pp. 830–838, 2011.
[5] R. Escudier, J. Bouffard, A. Pascual, P.-M. Poulain, and M.-I. Pujol, “Improvement of coastal and mesoscale observation from space:
Application to the northwestern Mediterranean Sea,” Geophysical Research Letters, vol. 40, no. 10, pp. 2148–2153, 2013.
[6] C. J. Donlon, M. Martin, J. Stark, J. Roberts-Jones, E. Fiedler, and W. Xindong, “The Operational Sea Surface Temperature and Sea Ice
Analysis (OSTIA) system,” Remote Sensing of Environment, vol. 116, pp. 140–158, 2012.
[7] B. Buongiorno Nardellia, A. Pisanoa, C. Tronconia, R. Santoleria, “Evaluation of different covariance models for the operational
interpolation of high resolution satellite Sea Surface Temperature data over the Mediterranean Sea,” Remote Sensing of Environment,
vol. 164, pp. 334343, 2015.
[8] P. Tandeo, P. Ailliot, and E. Autret, “Linear Gaussian state-space model with irregular sampling: application to sea surface temperature,”
Stochastic Environmental Research and Risk Assessment, vol. 25, no. 6, pp. 793–804, 2010.
[9] P. Tandeo, P. Ailliot, J. Ruiz, A. Hannart, B. Chapron, A. Cuzol, V. Monbet, R. Easton, and R. Fablet, “Combining Analog Method and
Ensemble Data Assimilation: Application to the Lorenz-63 Chaotic System,” in Machine Learning and Data Mining Approaches to Climate
Science, V. Lakshmanan, E. Gilleland, A. McGovern, and M. Tingley, Eds. Springer, 2015, pp. 3–12.
[10] S. Ba, T. Corpetti, and R. FABLET, “Multi-resolution missing data interpolation in SST Image Series,” in IEEE Int. Conf. on Image
processing, Bruxelles, Belgium, 2011.
[11] G. Evensen, Data Assimilation. , Springer Berlin, 2009.
[12] J. M. Beckers and M. Rixen, “EOF Calculations and Data Filling from Incomplete Oceanographic Datasets,Journal of Atmospheric and
Oceanic Technology, vol. 20, no. 12, pp. 1839–1856, 2003.
[13] B. Ping, F. Su, and Y. Meng, “An Improved DINEOF Algorithm for Filling Missing Values in Spatio-Temporal Sea Surface Temperature
Data,” PLOS ONE, vol. 11, no. 5, p. e0155928, 2016.
September 6, 2017 DRAFT
[14] R. Fablet and F. Rousseau, “Missing data super-resolution using non-local and statistical priors,” in 2015 IEEE International Conference
on Image Processing (ICIP), 2015, pp. 676–680.
[15] L. Lorenzi, F. Melgani, and G. Mercier, “Inpainting strategies for reconstruction of missing data in VHR images,” IEEE Geoscience and
Remote Sensing Letters, vol. 8, no. 5, pp. 914–918, 2011.
[16] F. Hamilton, T. Berry, and T. Sauer, “Ensemble Kalman Filtering without a Model,Physical Review X, vol. 6, no. 1, p. 011021, 2016.
[17] J. Li and A. D. Heap, “Spatial interpolation methods applied in the environmental sciences: A review,Environmental Modelling &
Software, vol. 53, pp. 173–189, 2014.
[18] P. Tandeo, E. Autret, B. Chapron, R. Fablet, and R. Garello, “SST spatial anisotropic covariances from METOP-AVHRR data,Remote
Sensing of Environment, vol. 141, pp. 144–148, 2014.
[19] F. Le-Dimet and O. Talagrand, “Variational algorithms for analysis and assimilation of meteorological observations: theoretical aspects,
Tellus, pp. 97–110, 1986.
[20] A. A. Efros and W. Freeman, “Image Quilting for Texture Synthesis and Transfer,” in Proceedings of the 28th Annual Conference on
Computer Graphics and Interactive Techniques, ser. SIGGRAPH ’01. New York, NY, USA: ACM, 2001, pp. 341–346.
[21] A. Criminisi, P. Perez, and K. Toyama, “Region filling and object removal by exemplar-based image inpainting,IEEE Transactions on
Image Processing, vol. 13, no. 9, pp. 1200–1212, 2004.
[22] A. Buades, B. Coll, and J. M. Morel, “A non-local algorithm for image denoising,” in IEEE Conf. on Computer Vision and Pattern
Recognition, CVPR’05, vol. 2, 2005, pp. 60–65 vol. 2.
[23] G. Peyr, S. Bougleux, and L. Cohen, “Non-local Regularization of Inverse Problems,Inverse Problems and Imaging, vol. 5, no. 2, pp.
511–530, 2011.
[24] R. Lguensat, P. Tandeo, P. Ailliot, M. Pulido and R. Fablet, “The Analog Data Assimilation,Monthly Weather Review, 2017.
[25] T. Tasdizen, “Principal Neighborhood Dictionaries for Nonlocal Means Image Denoising,IEEE Transactions on Image Processing, vol. 18,
no. 12, pp. 2649–2660, 2009.
[26] N. Cressie and C. K. Wikle, “Statistics for Spatio-Temporal Data,” Wiley & Sons, 2015.
[27] R. Reynolds, and T.M. Smith, “Improved global sea surface temperature analyses using optimum interpolation,” Jal of Climate, 7:229-249,
[28] C.-A. Deledalle, J. Salmon, and A. S. Dalalyan, “Image denoising with patch based PCA: local versus global.” Proc. 22nd British Machine
Vision Conf., BMVC’2011, 2011.
[29] W. T. Freeman and Liu, “Markov Random Fields for Super-Resolution,” in Advances in Markov Random Fields for Vision and Image
Processing, A. Blake, P. Kohli, and C. Rother ed. MIT Press, 2011.
[30] Z. Zhao and D. Giannakis, “Analog Forecasting with Dynamics-Adapted Kernels,” arXiv:1412.3831 [physics], 2014, arXiv: 1412.3831.
[31] S. Mallat, A wavelet tour of signal processing, second edition. Academic Press, 1999.
[32] M. Bertalmio, A. L. Bertozzi, and G. Sapiro, “Navier-Stokes, fluid dynamics, and image and video inpainting,” in Proc. IEEE Computer
Vision and Pattern Recognition, CVPR’01, 2001, pp. 355–362.
[33] B. Saulquin, F. Gohin, and R. Garrello, “Regional Objective Analysis for Merging High-Resolution MERIS, MODIS/Aqua, and SeaWiFS
Chlorophyll- a Data From 1998 to 2008 on the European Atlantic Shelf,” IEEE Transactions on Geoscience and Remote Sensing, vol. 49,
no. 1, pp. 143–154, 2011.
[34] M. Umbert, N. Hoareau, A. Turiel, and J. Ballabrera-Poy, “New blending algorithm to synergize ocean variables: The case of SMOS sea
surface salinity maps,” Remote Sensing of Environment, vol. 146, pp. 172–187, 2014.
[35] R. Fablet, J. Verron, B. Mourre, B. Chapron, and A. Pascual, “Improving mesoscale altimetric resolution/data in the Mediterranean Sea:
multi-tracer convolutional retreatment of standard products,” Submitted 2016,
[36] L. Gaultier, J. Verron, J.-M. Brankart, O. Titaud, and P. Brasseur, “On the inversion of submesoscale tracer fields to estimate the surface
ocean circulation,” Journal of Marine Systems, vol. 126, pp. 33–42, 2013.
[37] J. Gula, M. Molemaker, and J. McWilliams, “Submesoscale Cold Filaments in the Gulf Stream,Journal of Physical Oceanography,
vol. 44, no. 10, pp. 2617–2643, 2014.
[38] H. Sasaki and P. Klein, “SSH Wavenumber Spectra in the North Pacific from a High-Resolution Realistic Simulation,” Journal of Physical
Oceanography, vol. 42, no. 7, pp. 1233–1241, 2012.
September 6, 2017 DRAFT
... Satellite-based SST often loses a large amount of data for various reasons mentioned above, limiting the production of seamless daily SST from single satellite data sources. Therefore, studies to reconstruct missing SST data have been actively conducted [11][12][13][14][15][16][17][18]. Several widely used seamless daily SST products are based on Optimal Interpolation (OI) with data assimilation of multiple infrared and passive microwave satellite observations and in situ measurements, which include global products from Danish Meteorological Institute (DMI) on a grid scale of 0.05 • and Operational Sea Surface Temperature and Sea Ice Analysis (OSTIA) on a grid scale of 0.054 • [16][17][18][19]. ...
... Therefore, studies to reconstruct missing SST data have been actively conducted [11][12][13][14][15][16][17][18]. Several widely used seamless daily SST products are based on Optimal Interpolation (OI) with data assimilation of multiple infrared and passive microwave satellite observations and in situ measurements, which include global products from Danish Meteorological Institute (DMI) on a grid scale of 0.05 • and Operational Sea Surface Temperature and Sea Ice Analysis (OSTIA) on a grid scale of 0.054 • [16][17][18][19]. However, since such operational products are based on the OI approach resulting in smoothed surface, they have a relatively limited feature resolution [17]. ...
... Several widely used seamless daily SST products are based on Optimal Interpolation (OI) with data assimilation of multiple infrared and passive microwave satellite observations and in situ measurements, which include global products from Danish Meteorological Institute (DMI) on a grid scale of 0.05 • and Operational Sea Surface Temperature and Sea Ice Analysis (OSTIA) on a grid scale of 0.054 • [16][17][18][19]. However, since such operational products are based on the OI approach resulting in smoothed surface, they have a relatively limited feature resolution [17]. Several processes such as outlier removals and smoothing are conducted to improve the accuracy of the operational SST products, which often degrade the spatial pattern of the products, even though 1 km SST data are assimilated in the products [18]. ...
Full-text available
Sea SurfaceTemperature (SST) is a critical parameter for monitoring the marine environment and understanding various ocean phenomena. While SST can be regularly retrieved from satellite data, it often suffers from missing data due to various reasons including cloud contamination. In this study, we proposed a novel two-step data fusion framework for generating high-resolution seamless daily SST from multi-satellite data sources. The proposed approach consists of (1) SST reconstruction based on Data Interpolate Convolutional AutoEncoder (DINCAE) using the SSTs derived from two satellite sensors (i.e., Moderate Resolution Imaging Spectroradiometer (MODIS) and Advanced Microwave Scanning Radiometer 2(AMSR2)), and (2) SST improvement through data fusion using random forest for consistency with in situ measurements with two schemes (i.e., scheme 1 using the reconstructed MODIS SST variables and scheme 2 using both MODIS and AMSR2 SST variables). The proposed approach was evaluated over the Kuroshio Extension in the Northwest Pacific, where a highly dynamic SST pattern can be found, from 2015 to 2019. The results showed that the reconstructed MODIS and AMSR2 SSTs through DINCAE yielded very good performance with Root Mean Square Errors (RMSEs) of 0.85 and 0.60 °C and Mean Absolute Errors (MAEs) of 0.59 and 0.45 °C, respectively. The results from the second step showed that scheme 2 and scheme 1 produced RMSEs of 0.75 and 0.98 °C and MAEs of 0.53 and 0.68 °C, respectively, compared to the in situ measurements, which proved the superiority of scheme 2 using multi-satellite data sources. Scheme 2 also showed comparable or even better performance than two operational SST products with similar spatial resolution. In particular, scheme 2 was good at simulating features with fine resolution (~50 km). The proposed approach yielded promising results over the study area, producing seamless daily SST products with high quality and high feature resolution.
... s depends on the orbit of the satellite. The wind quality depends on multiple parameters related to the inversion algorithm (Chapron et al. 2001;Lin et al. 2008) structures by improving the optimal interpolation (Ubelmann et al. 2015), or through exploiting data-driven representations (Tandeo et al. 2015;Lguensat et al. 2017b;Lguensat et al. 2017a;R. Fablet et al. 2017;Barth et al. 2020;Beauchamp et al. 2020) as investigated in the next sections. ...
... Recently, data-driven approaches (Tandeo et al. 2015;Lguensat et al. 2017b;Lguensat et al. 2017a;R. Fablet et al. 2017;Barth et al. 2020;Beauchamp et al. 2020) have emerged as relevant alternatives to model-driven schemes. They take benefit from the increasing availability of remote sensing observations and simulation data to derive computationally efficient representations. Analog methods are one of the first data-driven techniques developed within a da ...
... to model-driven schemes. They take benefit from the increasing availability of remote sensing observations and simulation data to derive computationally efficient representations. Analog methods are one of the first data-driven techniques developed within a data assimilation framework (Tandeo et al. 2015;Lguensat et al. 2017b;Lguensat et al. 2017a;R. Fablet et al. 2017). Combining analog data assimilation (AnDA) with a patch-based representation have shown great results with respect to the state-of-theart OI and EOF-based schemes. However, the parametrization of the proposed framework involves tuning several parameters principally due to the data-driven formulation of the dynamical prior based on analo ...
This thesis focuses on the data-driven identification of dynamical representations of upper ocean dynamics for forecasting, simulation and data assimilation applications. We focus on practical considerations regarding the provided observations and tackle multiple issues, ranging from the parametrization of the models, their time integration, the space in which the models should be definedand their implementation in data assimilation schemes.The core of our work resides in proposing a new data-driven embedding technique. This framework optimises an augmented space as a solution of an optimization problem, parametrised by a trainable Ordinary Differential Equation (ODE) that can be used for several applications such as forecasting and data assimilation. We discuss the effectiveness of the proposed framework within two different parametrizations of the trainable ODE. Namely, the Linear-quadratic and Linear ones and show that both formulations lead to interesting applications and most importantly, connect with interesting state-of-theart theory that helps understanding and constraining the proposed architecture. Regarding data assimilation applications, we explore two distinct methodologies. The first technique can be seen as an alternative to the ensemble Kalman filtering and the second one relates to the proposed dynamical embedding technique and can be extended to match recent advances of state-of-the-art filtering techniques.
... For the other types of energy forms, their parameterizations are generally set a priori and not learned from the data. Regarding more particularly datadriven and learning-based approaches, most previous works (Peyr et al., 2011;Alvera-Azcarate et al., 2016;Fablet et al., 2017) have addressed the learning of interpolation schemes under the assumption that a representative gap-free dataset is available. This gap-free dataset may be the image itself (Criminisi et al., 2004;Lorenzi et al., 2011;Peyr et al., 2011). ...
... The AnDa scheme combines the exploitation of a gap-free reference dataset to design an analog dynamical model with a state-of-the-art ensemble Kalman filter to address the interpolation issue. Here, we implement the patch-based version of AnDA as presented in Fablet et al. (2017). As AnDA and DINEOF schemes rely on some training on a gap-free dataset, we use for evaluation purposes a 20-day test period in December, with the remaining data being used as training data except 10 days before and after the 20-day test period. ...
Full-text available
Earth observation satellite missions provide invaluable global observations of geophysical processes in play in the atmosphere and the oceans. Due to sensor technologies (e.g., infrared satellite sensors), atmospheric conditions (e.g., clouds and heavy rains), and satellite orbits (e.g., polar-orbiting satellites), satellite-derived observations often involve irregular space–time sampling patterns and large missing data rates. Given the current development of learning-based schemes for earth observation, the question naturally arises whether one might learn some representation of the underlying processes as well as solve interpolation issues directly from these observation datasets. In this article, we address these issues and introduce an end-to-end neural network learning scheme, which relies on an energy-based formulation of the interpolation problem. This scheme investigates different learning-based priors for the underlying geophysical field of interest. The end-to-end learning procedure jointly solves the reconstruction of gap-free fields and the training of the considered priors. Through different case studies, including observing system simulation experiments for sea surface geophysical fields, we demonstrate the relevance of the proposed framework compared with optimal interpolation and other state-of-the-art data-driven schemes. These experiments also support the relevance of energy-based representations learned to characterize the underlying processes.
... Marullo et al. (2014) fused clearsky SSTs from the SEVIRI and Mediterranean Forecasting System to generate gap-free, hourly SSTs using diurnal optimal interpolation across a moving temporal window. Different model outputs have also been evaluated for filling invalid SST pixels (Nardelli et al., 2015); however, the resulting accuracies of cloudy-sky results were highly reliant upon simulation data, especially for continuous cloudy days, and no further correction was implemented (Fablet et al., 2017). Dumitrescu et al. (2020) fused daytime hourly SEVIRI LSTs and modeled skin temperature using multiple linear regression and generalized additive models, in addition to elevation, time, and solar radiation to improve GEO LST estimation under clouds. ...
Full-text available
By characterizing high-frequency surface thermal dynamics at a medium spatial scale, hourly land surface temperatures (LST), retrieved from geostationary satellite thermal infrared (TIR) observations, shows great potential to be used across a range of scientific applications; however, cloud cover typically leads to data gaps and degraded retrieval accuracy in TIR LST products, such as the official Advanced Baseline Imager (ABI) LST product. Studies have focused on the LST gap reconstruction; however, most interpolation-based methods only work for a short-term cloud duration and are unable to adequately compensate for cloud effects, and traditional surface energy balance (SEB)-based methods are able to handle cloud coverage while they are not feasible for use at night. Moreover, few studies have concentrated on recovering the abnormal retrievals of partial cloud pixels. In this study, an all-sky diurnal, hourly LST estimation method based on SEB theory was proposed; the proposed method involved three major steps: 1) an original spatiotemporal dynamic model of LST was constructed from ECMWF Reanalysis v.5 (ERA5); 2) clear-sky ABI LST was then assimilated to the dynamic model to generate a continuous LST series; 3) the diurnal cloud effects were superimposed on cloudy time estimated by an innovative optimization method from satellite radiation products. A 2-km, all-sky, hourly LST product was produced over the contiguous US and Mexico from July 2017 to June 2021. Validation was conducted using ground measurements at 18 sites from Surface Radiation (SURFRAD) and core AmeriFlux networks, and produced an overall root-mean-square error (RMSE) of 2.44 K, a bias of − 0.19 K, and an R 2 of 0.97 based on 408,300 samples. For clear-sky samples, the RMSE values were 2.37 and 2.24 K for day and nighttime, respectively, which was a notable improvement over the corresponding values of the official ABI LST product (2.73 and 2.86 K, respectively). The RMSE values on cloudy-sky were 2.78 and 2.23 K for day and nighttime, respectively. The daily mean LST by aggregating all-sky, hourly LST had an RMSE of 1.13 K and R 2 of 0.99. Overall, this product showed reliability under consistent cloud durations, although it was slightly affected by surface elevation. The diurnal temperature cycle climatology of major land cover types was also characterized. The product is freely available at:
... At present, the advances in remote sensing technologies have enabled the acquisition of global observation to be more accessible in ocean mesoscale eddy detection, tracking, prediction tasks. Mesoscale eddies have been investigated both intensively and extensively with long-term time series (Oey et al., 2005;Wang et al., 2015;Świerczyńska et al., 2016;Fablet et al., 2017;Imani et al., 2017;Fu et al., 2019), e.g., sea surface height, salinity, temperature, SLA, etc. The initialization of ocean dynamic state through assimilation underpins prediction with the optimal statistical estimation from the numerical models and the sparse ocean interior information (Thoppil et al., 2021). ...
Full-text available
Ocean mesoscale eddies are ubiquitous in world ocean and account for 90% oceanic kinetic energy, which dominate the upper ocean flow field. Accurately predicting the variation of ocean mesoscale eddies is the key to understand the oceanic flow field and circulation system. In this article, we propose to make an initial attempt to explore spatio-temporal predictability of mesoscale eddies, employing deep learning architecture, which primarily establishes Memory In Memory (MIM) for sea level anomaly (SLA) prediction, combined with the existing mesoscale eddy detection. Oriented to the western Pacific ocean (125�􀀀137.5�E and 15�􀀀27.5�N), we quantitatively investigate the historic daily SLA variability at a 0.25� spatial resolution from 2000 to 2018, derived by satellite altimetry. We develop the enhanced MIM prediction strategies, equipped with Gated Recurrent Unit (GRU) and spatial attention module, in a scheduled sampling manner, which overcomes the gradient vanishing and complements to strengthen spatio-temporal features for long-term dependencies. At the early stage, the real value SLA input guides the model training process for initialization, while the scheduled sampling intentionally feeds the newly predicted value, to resolve the distribution inconsistency of inference. It has been demonstrated in our experiment results that our proposed prediction scheme outperformed the state-of-art approaches for SLA time series, with MAPE, RMSE of the 14-day prediction duration, respectively, 5.1%, 0.023 m on average, even up to 4.6%, 0.018 m for the effective sub-regions, compared to 19.8%, 0.086 m in ConvLSTM and 8.3%, 0.040 m in original MIM, which greatly facilitated the mesoscale eddy prediction. This proposed scheme will be beneficial to understand of the underlying dynamical mechanism behind the predictability of mesoscale eddies in the future, and help the deployment of ARGO, glider, AUV and other observational platforms.
... Using CubeSats, Houborg and McCabe [18] present an approach to further leverage Landsat and MODIS for further spatiotemporal improvements. Similarly, nearest neighbor analog multiscale patch-decomposition data-driven models are used as state-of-the-art interpolation techniques for developing global sea surface temperature (SST) datasets [19]. In recent years, super-resolution techniques have presented state-of-the-art results for spatial enhancement of satellite images [20]- [22]. ...
Full-text available
Applications of satellite data in areas such as weather tracking and modeling, ecosystem monitoring, wildfire detection, and land-cover change are heavily dependent on the tradeoffs to spatial, spectral, and temporal resolutions of observations. In weather tracking, high-frequency temporal observations are critical and used to improve forecasts, study severe events, and extract atmospheric motion, among others. However, while the current generation of geostationary (GEO) satellites has hemispheric coverage at 10-15-min intervals, higher temporal frequency observations are ideal for studying mesoscale severe weather events. In this work, we present a novel application of deep learning-based optical flow to temporal upsampling of GEO satellite imagery. We apply this technique to 16 bands of the GOES-R/Advanced Baseline Imager mesoscale dataset to temporally enhance full-disk hemispheric snapshots of different spatial resolutions from 10 to 1 min. Experiments show the effectiveness of task-specific optical flow and multiscale blocks for interpolating high-frequency severe weather events relative to bilinear and global optical flow baselines. Finally, we demonstrate strong performance in capturing variability during convective precipitation events.
Full-text available
Satellite altimetry is a unique way for direct observations of sea surface dynamics. This is however limited to the surface-constrained geostrophic component of sea surface velocities. Ageostrophic dynamics are however expected to be significant for horizontal scales below 100~km and time scale below 10~days. The assimilation of ocean general circulation models likely reveals only a fraction of this ageostrophic component. Here, we explore a learning-based scheme to better exploit the synergies between the observed sea surface tracers, especially sea surface height (SSH) and sea surface temperature (SST), to better inform sea surface currents. More specifically, we develop a 4DVarNet scheme which exploits a variational data assimilation formulation with trainable observations and {\em a priori} terms. An Observing System Simulation Experiment (OSSE) in a region of the Gulf Stream suggests that SST-SSH synergies could reveal sea surface velocities for time scales of 2.5-3.0 days and horizontal scales of 0.5$^\circ$-0.7$^\circ$, including a significant fraction of the ageostrophic dynamics ($\approx$ 47\%). The analysis of the contribution of different observation data, namely nadir along-track altimetry, wide-swath SWOT altimetry and SST data, emphasizes the role of SST features for the reconstruction at horizontal spatial scales ranging from \nicefrac{1}{20}$^\circ$ to \nicefrac{1}{4}$^\circ$.
Full-text available
Frente a una narrativa distópica presente en los medios de comunicación y cultura popular que caracteriza el avance y desarrollo de la inteligencia artificial como una amenaza o riesgo existencial (e.g. desempleo tecnológico, sistemas de armas autónomas letales, robots asesinos, propaganda política computacional etc.) quiero valorar de manera crítica y constructiva el rol de la inteligencia artificial para el bien común (AI4SG). La tecnología digital también se puede aplicar para la solución de grandes problemas de la humanidad, como los 17 Objetivos de Desarrollo Sostenible y sus 169 metas de la agenda 2030. En este artículo, comentaré distintos casos de uso y aplicación de la inteligencia artificial y la robótica encaminados a conseguir los 17 Objetivos de Desarrollo Sostenible y qué principios éticos deben guiar su aplicación para que la inteligencia artificial consiga la ambiciosa agenda 2030. También comentaré el plan de acción de España y la estrategia nacional para cumplir la agenda 2030 y de qué manera incorpora las TIC.
Full-text available
This paper addresses physics-informed deep learning schemes for satellite ocean remote sensing data. Such observation datasets are characterized by the irregular space-time sampling of the ocean surface due to sensors’ characteristics and satellite orbits. With a focus on satellite altimetry, we show that end-to-end learning schemes based on variational formulations provide new means to explore and exploit such observation datasets. Through Observing System Simulation Experiments (OSSE) using numerical ocean simulations and real nadir and wide-swath altimeter sampling patterns, we demonstrate their relevance w.r.t. state-of-the-art and operational methods for space-time interpolation and short-term forecasting issues. We also stress and discuss how they could contribute to the design and calibration of ocean observing systems.
Full-text available
Conference Paper
This paper addresses physics-informed deep learning schemes for satellite ocean remote sensing data. Such observation datasets are characterized by the irregular space-time sampling of the ocean surface due to sensors' characteristics and satellite orbits. With a focus on satellite altimetry, we show that end-to-end learning schemes based on variational formulations provide new means to explore and exploit such observation datasets. Through Observing System Simulation Experiments (OSSE) using numerical ocean simulations and real nadir and wide-swath altimeter sampling patterns, we demonstrate their relevance w.r.t. state-of-the-art and operational methods for space-time interpolation and short-term forecasting issues. We also stress and discuss how they could contribute to the design and calibration of ocean observing systems.
Full-text available
In light of growing interest in data-driven methods for oceanic, atmospheric, and climate sciences, this work focuses on the field of data assimilation and presents the analog data assimilation (AnDA). The proposed framework produces a reconstruction of the system dynamics in a fully data-driven manner where no explicit knowledge of the dynamical model is required. Instead, a representative catalog of trajectories of the system is assumed to be available. Based on this catalog, the analog data assimilation combines the nonparametric sampling of the dynamics using analog forecasting methods with ensemble-based assimilation techniques. This study explores different analog forecasting strategies and derives both ensemble Kalman and particle filtering versions of the proposed analog data assimilation approach. Numerical experiments are examined for two chaotic dynamical systems: the Lorenz-63 and Lorenz-96 systems. The performance of the analog data assimilation is discussed with respect to classical model-driven assimilation. A Matlab toolbox and Python library of the AnDA are provided to help further research building upon the present findings.
Full-text available
In this study, an improved Data INterpolating Empirical Orthogonal Functions (DINEOF) algorithm for determination of missing values in a spatio-temporal dataset is presented. Compared with the ordinary DINEOF algorithm, the iterative reconstruction procedure until convergence based on every fixed EOF to determine the optimal EOF mode is not necessary and the convergence criterion is only reached once in the improved DINEOF algorithm. Moreover, in the ordinary DINEOF algorithm, after optimal EOF mode determination, the initial matrix with missing data will be iteratively reconstructed based on the optimal EOF mode until the reconstruction is convergent. However, the optimal EOF mode may be not the best EOF for some reconstructed matrices generated in the intermediate steps. Hence, instead of using asingle EOF to fill in the missing data, in the improved algorithm, the optimal EOFs for reconstruction are variable (because the optimal EOFs are variable, the improved algorithm is called VE-DINEOF algorithm in this study). To validate the accuracy of the VE-DINEOF algorithm, a sea surface temperature (SST) data set is reconstructed by using the DINEOF, I-DINEOF (proposed in 2015) and VE-DINEOF algorithms. Four parameters (Pearson correlation coefficient, signal-to-noise ratio, root-mean-square error, and mean absolute difference) are used as a measure of reconstructed accuracy. Compared with the DINEOF and I-DINEOF algorithms, the VE-DINEOF algorithm can significantly enhance the accuracy of reconstruction and shorten the computational time.
Full-text available
Methods of data assimilation are established in physical sciences and engineering for the merging of observed data with dynamical models. When the model is nonlinear, methods such as the ensemble Kalman filter have been developed for this purpose. At the other end of the spectrum, when a model is not known, the delay coordinate method introduced by Takens has been used to reconstruct nonlinear dynamics. In this article, we merge these two important lines of research. A model-free filter is introduced based on the filtering equations of Kalman and the data-driven modeling of Takens. This procedure replaces the model with dynamics reconstructed from delay coordinates, while using the Kalman update formulation to reconcile new observations. We find that this combination of approaches results in comparable efficiency to parametric methods in identifying underlying dynamics, and may actually be superior in cases of model error.
Full-text available
Two general algorithms for solving constrained minimization problems are presented and discussed in the context of analysis and assimilation of meteorological observations. The first algorithm presented in the augmented lagrangian algorithm. It generalizes the more classical penalty and duality algorithms. The second algorithm, inspired from optimal control techniques, is based on an appropriate use of an adjoint dynamical equation, and seems to be particularly well adapted to the assimilation of observations distributed in time. Simple numerical examples show the ability of these algorithms to solve non-linear minimization problems of the type encountered in meteorology. Their possible use in more complex situations is discussed, in particular in terms of their computational cost. - from Authors
Full-text available
Conference Paper
We here address the super-resolution of a high-resolution image involving missing data given that a low-resolution image of the same scene is available. This is a typical issue in the remote sensing of geophysical parameters from different spaceborne sensors. Such super-resolution application involves large downscaling factor (typically from 10 to 20) and the super-resolution model should account for both texture patterns and specific statistical features, especially the spectral and non-Gaussian features. In this context, we propose a novel non-local approach and formally states the solution as the joint minimization of several projection constraints. We illustrate the relevance of the proposed model on real ocean remote sensing data, namely sea surface temperature fields, as well on visual textures.
Mallat's book is the undisputed reference in this field - it is the only one that covers the essential material in such breadth and depth. - Laurent Demanet, Stanford University The new edition of this classic book gives all the major concepts, techniques and applications of sparse representation, reflecting the key role the subject plays in today's signal processing. The book clearly presents the standard representations with Fourier, wavelet and time-frequency transforms, and the construction of orthogonal bases with fast algorithms. The central concept of sparsity is explained and applied to signal compression, noise reduction, and inverse problems, while coverage is given to sparse representations in redundant dictionaries, super-resolution and compressive sensing applications. Features: * Balances presentation of the mathematics with applications to signal processing * Algorithms and numerical examples are implemented in WaveLab, a MATLAB toolbox * Companion website for instructors and selected solutions and code available for students New in this edition * Sparse signal representations in dictionaries * Compressive sensing, super-resolution and source separation * Geometric image processing with curvelets and bandlets * Wavelets for computer graphics with lifting on surfaces * Time-frequency audio processing and denoising * Image compression with JPEG-2000 * New and updated exercises A Wavelet Tour of Signal Processing: The Sparse Way, third edition, is an invaluable resource for researchers and R&D engineers wishing to apply the theory in fields such as image processing, video processing and compression, bio-sensing, medical imaging, machine vision and communications engineering. Stephane Mallat is Professor in Applied Mathematics at École Polytechnique, Paris, France. From 1986 to 1996 he was a Professor at the Courant Institute of Mathematical Sciences at New York University, and between 2001 and 2007, he co-founded and became CEO of an image processing semiconductor company. Includes all the latest developments since the book was published in 1999, including its application to JPEG 2000 and MPEG-4 Algorithms and numerical examples are implemented in Wavelab, a MATLAB toolbox Balances presentation of the mathematics with applications to signal processing.
The EnKF has recently been taken into use with simulation models for oil and gas reservoirs, with the purpose of estimating poorly known parameters and to improve the predictive capability of the models. There are economical benefits of obtaining a model which best possible represents the reservoir. Optimally, it could be used for predicting the future production and to assist in the planning of new production and injection wells. A better model also provides insight and understanding regarding the properties of the reservoir.
Daily gap-free Sea Surface Temperature (SST) fields at high resolution are required by several operational users working on monitoring and forecasting the status of the marine environment. Existing instruments cannot provide such fields, and careful interpolation of available data is required to provide gap-free SST estimates. In the framework of the MyOcean projects (now turning into the European Copernicus Marine Service), several satellite (and/or satellite and in situ) interpolated products are distributed in real time, and continuous research and development activities are carried out to improve their quality. In this paper, we describe the work done to improve the High Resolution (1/16°) interpolated product covering the Mediterranean Sea, that is obtained by combining all available satellite infrared images. Three Optimal Interpolation schemes based on different space–time covariance models and background fields are derived, compared and validated versus in situ drifting buoy measurements (leading to a minimum standard deviation error (STDE) of 0.46 °K and mean bias error (MBE) of − 0.15 °K), as well as by adopting a holdout validation approach with artificial clouds (with STDE ranging between 0.22 °K and 0.55 °K, and MBE always between 0.03 °K and 0.05 °K, depending on the cloud configuration considered). Almost negligible differences are found between the three products, revealing only a slight improvement when using spatially varying covariance parameters and a daily climatology as background, which is less risky in case of prolonged cloudy conditions, though attaining the same performance with small/rapid clouds. This scheme has thus been implemented in the Mediterranean SST operational chain since April 2014.