IEEE Geoscience and Remote Sensing Magazine - September 2021 - 130

LSTs in time and space, which provides a new approach
for obtaining continuous LST data. With this technique, a
series of LST data can be simulated, and likely trends can
be identified.
Such trend information provides a physical constraint
on LST and can enhance the accuracy of LST reconstruction
and spatiotemporal fusion methods. This point has
recently attracted the attention of scholars. Based on the
spatiotemporal fusion method and CLDAS LST data, Long
et al. generated MODIS-like LSTs suitable for all-weather
conditions [91]. It can be predicted that the generation of
SCHR-LST using LST from data assimilation will become
more common.
SYNERGIES BETWEEN RECONSTRUCTION
AND SPATIOTEMPORAL FUSION
All of the reviewed LST reconstruction and spatiotemporal
fusion methods have been implemented as separate, independent
processes. In general, input data for spatiotemporal
fusion should be spatially continuous or filtered with a
common cloud mask [17], [43], [104], [105]. Unfortunately,
cloud contamination, defective sensors, and scanning gaps
between orbits cause numerous abnormal or missing values,
which hinders LST spatiotemporal fusion [13]. Several
applied studies on spatiotemporal fusion simply ignore spatially
discontinuous images or use synthetic products (e.g.,
eight-day LST composite products and MOD11A2) [100],
[139], [152], [153].
Although synthetic products are helpful for analyzing dynamic
changes in a long time series, they increase the uncertainty
in location- and time-specific quantitative studies.
LST fusion results can be used as input data for reconstruction
[91]. Therefore, from an application perspective, it is
highly desirable to synergistically combine LST reconstruction
and spatiotemporal fusion.
CONCLUSIONS
Missing information and data gaps are prevalent occurrences
in satellite-retrieved LST values, and a tradeoff always exists
between temporal and spatial resolutions. Obtaining
SCHR-LST production is crucial to many fields of research
and applications [18], e.g., Earth's energy balance, material
and energy exchange on a global scale, subpixel wildfire
temperature detection, drought assessment, and urban heat
island monitoring on a local or fine scale.
The reconstruction and spatiotemporal fusion of LST
products have become areas of abundant research that aims
to address the limitations. Although reconstruction methods
attempt to obtain spatially continuous LST fields, i.e.,
by filling existing data gaps, spatiotemporal fusion methods
generate gap-free LST images at high spatial and temporal
resolution simultaneously.
This study reviews the recent advances made in the reconstruction
and spatiotemporal fusion of LST products and
puts forward prospective research. The main contributions
of this work are summarized in the following three aspects:
130
◗ Brief theoretical concepts of LST retrieval and processing
are presented. Currently used standard LST satellite
products along with their specifications and the limitations
of satellite-derived LST products are summarized.
The problems of reconstruction and spatiotemporal fusion
are described, and the connections between them
are discussed in the " Background " section.
◗ A thorough overview of the current advancements made
in reconstruction and spatiotemporal fusion is provided
in the " Reconstruction for Spatial Continuity " and
" Spatiotemporal Fusion for High Resolution " sections.
The survey covers the classifications of existing methods,
provides detailed comparisons of significant algorithms
from experimental and theoretical perspectives,
presents limitations related to SCHR-LST, and introduces
prevalent validation strategies. We present three types of
reconstruction methods for missing pixels (the spatial,
temporal, and spatiotemporal approaches), two groups of
reconstruction methods for cloudy pixels (the PMW- and
SEB-based techniques), three types of spatiotemporal fusion
methods (weighted function-, unmixing-, hybrid,
and learning-based approaches), and four validation
methods (SiV, GrV, OsV, and DsV) for the reconstructed
and fused LST products. In addition, we summarize the
main evaluation indicators in the " Validation of Reconstructed
and Fused LST Products " section.
◗ Several future prospects of reconstruction and spatiotemporal
fusion are put forward in the " Future Prospects "
section. The exploitation of geographical laws,
integration of physical properties, application of DLbased
frameworks, combination of assimilated data,
and creation of synergies between reconstruction and
spatiotemporal fusion are discussed. The application
conditions and features as well as the data sources of
these approaches are summarized in Table 7. This review
is designed to provide researchers with guidelines on selecting
the proper reconstruction or fusion method.
ACKNOWLEDGMENTS
The work was supported by the National Key Research and
Development Program under grant 2018YFA0605500 and
the National Natural Science Foundation of China (grants
41501376 and 41871275). The corresponding authors are
Penghai Wu and Huanfeng Shen.
AUTHOR INFORMATION
Penghai Wu (wuph@ahu.edu.cn) received his Ph.D. degree
in cartography and geographical information engineering
from Wuhan University, Wuhan, China, in 2014.
He is currently an associate professor with the School of
Resources and Environmental Engineering, Anhui University,
Hefei, 230601, China. He has authored more
than 40 research articles in peer-reviewed international
journals. He is a member of the editorial board of Journal
of Sensors and has frequently served as a referee for
more than 20 international journals for remote sensing
IEEE GEOSCIENCE AND REMOTE SENSING MAGAZINE SEPTEMBER 2021
IEEE Geoscience and Remote Sensing Magazine - September 2021

Table of Contents for the Digital Edition of IEEE Geoscience and Remote Sensing Magazine - September 2021
