IEEE Geoscience and Remote Sensing Magazine - September 2021 - 31

algorithm that combines information from the raw satellite
data with additional external data or by employing climatological
information to compensate for the lack of information
content in the raw data. Generally, a high level of
prior dependence reduces the independent value carried by
the satellite-derived product.
The last four attributes-continuity, robustness, resiliency,
and data timeliness-are attributes of the system,
not the sensor. For continuity, the sensors might be of
different types as long as the geophysical observables
can be rigorously intercalibrated. For climate monitoring,
for example, relatively large random errors can be
acceptable if the observations are free of bias and stable
over time. Therefore, long-term stability and continuity
are critical attributes for climate applications. It is worth
noting that the data collected from a specific mission
will be used for climate studies for decades past the mission's
lifetime.
There is no established metric used to measure the robustness
(the capability to perform reliably without failure
over a wide range of conditions and time) of a particular
variable observed by the EOSC. Here robustness represents
the overall global observing system, not the robustness
of a single sensor (typically quantified as the mean time
between failures). And this overall robustness does not require
sensors that measure the same thing at exactly the
same time with the same observation geometry. For example,
a constellation that measures global atmospheric temperatures
with six independent sensors is more robust than
one that uses only two sensors.
Resiliency, the ability to quickly replenish the EOSC
(as measured by the time necessary for replacement),
may become much more critical as the risk from space
debris increases. Our definition of data timeliness (often
called data latency) includes both downlink and processing
time as well as all the delays introduced by data
exchange and delivery to users, both of which can be
significant. The attributes of robustness, resiliency, and
timeliness are particularly important for operational applications
where providing warnings, predictions, and
other anticipatory information has come to be expected
in real time, all the time, and almost everywhere.
When assessing the value of an observing system in the
framework of all the applications (or, more practically, a
set of representative applications), it is important to keep
in mind all of the attributes listed in Table 1. The current
EOSC, a product of both long-standing and novel international
partnerships, provides a wide range of environmental
observations with varying degrees of complementarity,
including observations of different accuracy, coverage,
density, and resolution. Typically, each application will
have different requirements for the attributes of the geophysical
variables.
In some cases, several sensors of different types provide
information about the same geophysical variable. This
might lead to the perception of redundancy (and an implied
SEPTEMBER 2021 IEEE GEOSCIENCE AND REMOTE SENSING MAGAZINE
lack of efficiency), but, in reality, different types of sensors
may cover several nonoverlapping, nonredundant aspects
of the observable itself, which serve to complement each
other and therefore address a diverse set of applications.
For example, MW, IR, and RO sensors all contribute different,
independent, and valuable information on the vertical
profiles of temperature and water vapor. IR has a modest
vertical resolution (a vertical resolving power of ~1 km) but
cannot " see " through clouds.
MW provides useful information
in clouds but has a lower
vertical resolution (a resolving
power ~2-3 km). MW and
IR have horizontal footprints
(GIFOV) at a nadir of approximately
15-50 km2 [17].
RO provides high vertical
resolution (a resolving power
of 200-500 m) [18] and is
nearly unaffected by clouds
but has a larger horizontal
footprint (~150 km × 1 km)
[19], which limits RO's achievable horizontal resolving
power. Therefore, these three sensor types are complementary:
Collectively, they provide improved estimates of
atmospheric temperature compared to any single sensor
type. It is also of great value when assimilating these three
types of sensors into NWP models that the errors of the
different types are uncorrelated.
RESILIENCY, THE ABILITY
TO QUICKLY REPLENISH THE
EOSC (AS MEASURED BY
THE TIME NECESSARY FOR
REPLACEMENT), MAY
BECOME MUCH MORE
CRITICAL AS THE RISK FROM
SPACE DEBRIS INCREASES.
FROM NOISE TO SIGNAL: A HISTORICAL PERSPECTIVE
AND THE IMPLICATIONS FOR REQUIREMENTS
The science and methods used to derive products from
satellite data are constantly evolving, sometimes in major
ways. For example, NWP systems originally assimilated
temperature profiles estimated from radiance information,
but research showed that it was more effective to assimilate
radiance information directly, and positive impacts
have been observed since this change was made [20]. In
some cases, the measurements originally assumed to be
contaminated by " spurious signals " and thus not useful
for some applications became more useful as the causes of
contamination were accounted for, while at the same time
yielding new and innovative products for the geophysical
variables that cause the " contamination. " In NWP,
for instance, some MW data measured in the presence of
cloud and rain have long been considered contaminated.
Only recently have these data become valuable for NWP
as knowledge of how to use them has improved. Now, extensive
efforts are being undertaken to maximize the use
of these data in global models at all of the major NWP
centers [21], [22].
New measurement techniques are regularly being proposed
and tested, with some evolving into operational
practice. One striking example is the use of GNSS signals
to infer profiles of atmospheric radio wave-bending angles
31
IEEE Geoscience and Remote Sensing Magazine - September 2021

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