Geoscience Reference
In-Depth Information
soil processes in boreal regions (Rautiainen et al.
2012
); disaggregation of SMOS data
(Merlin et al.
2012
); and various aspects of the validation of SMOS soil moisture data (Al
Bitar et al.
2012
; Bircher et al.
2012
; dall'Amico et al.
2012
; Jackson et al.
2012
; Lacava et al.
2012
; Mialon et al.
2012
; Peischl et al.
2012b
; Rowlandson et al.
2012
; Sanchez et al.
2012
;
Schlenz et al.
2012
; Schwank et al.
2012
).
In view of the applications discussed later in this paper, we briefly mention that snow
measurements are often provided by AMSR-E to measure snow water equivalent (SWE),
and MODIS (MODerate resolution Imaging Spectroradiometer, Morisette et al.
2002
)to
give a picture of the snow-covered area. Finally, it is worth to mention GRACE (Gravity
Recovery And Climate Experiment, Tapley et al.
2004
) for its ability to measure an
integrated water quantity of soil moisture and snow, as well as water in deeper layers.
Satellite instruments do not measure directly hydrological parameters. What they
measure is photon counts (level 0 data). Algorithms then transform the level 0 data into
radiances (level 1 data). Subsequently, using retrieval techniques (Rodgers
2000
), retri-
evals of layer quantities (e.g., of soil moisture) or integrated amounts (e.g., total water
storage) are derived (level 2 data). Fields derived from manipulation of level 2 data, for
example, by interpolation to a common grid are termed level 3 data. Analyses derived from
the assimilation of level 1 and/or 2 data are termed level 4 data.
Satellite observations (from level 0 and up) have associated with them a number of
errors, including random and systematic errors in the measurement, and the error of rep-
resentativeness (or representativity). Random errors (sometimes termed precision) have the
property that averaging the data can reduce them. This is not the case of the systematic
error or bias (sometimes termed accuracy). The error of representativeness is associated
with the extent to which the measurement represents a point or volume in space. In land
surface measurements, the error of representativeness is important to consider in com-
parisons of coarse-scale satellite data with point data.
Satellite-based hydrological data are becoming increasingly available, although little
progress has been made in understanding their observational errors. Evaluation of the
accuracy of land surface satellite data is a challenge, and novel methods to characterize
their errors are being applied. Examples include triple collocation (e.g., Scipal et al.
2008
;
Dorigo et al.
2010
; Parinussa et al.
2011
); the R-metrics approach (Crow
2007
; Crow and
Zhan
2007
; Crow et al.
2010
); and data assimilation (Houser et al.
2010
, and references
therein).
A number of in situ network and airborne hydrological studies have been set up in the
last decade for evaluation of satellite data. Examples of in situ networks include
SMOSMANIA in France (Calvet et al.
2007
; Albergel et al.
2009
); NVE (Norges vassd-
rags-og energidirektorat, Norwegian Water Resources and Energy Directorate) in Norway
and elsewhere (see Table 1 in Crow et al.
2012
); and several local- to regional-scale (larger
than 100 km
2
, smaller than 10,000 km
2
) networks in the USA and elsewhere (see Table 2
in Crow et al.
2012
). In situ soil moisture data from various networks across the world are
consolidated in the International Soil Moisture Network (ISMN;
http://www.ipf.tuwien.ac.
vides details. An example of an airborne study on evaluation of satellite data is the
Australian Airborne Cal/Val Experiments for SMOS (AACE, Peischl et al.
2012a
). An
example of in situ ground-based station data used to evaluate satellite data is SMOSREX
(de Rosnay et al.
2006
). The temporal scale of in situ platforms ranges from minutes to
hours; the spatial scale of in situ platforms ranges from tens of metres (individual stations)
to thousands of kilometres (regional-scale networks). Along with the availability of dense
