Geoscience Reference
In-Depth Information
1 Introduction
The water cycle plays a crucial role in Earth's climate and environment, yet there are still
large gaps in our understanding of its components, particularly at the land surface (Lahoz
and De Lannoy
2013
; Trenberth and Asrar
2013
). Over the past decade, there has been a
steady increase in the number and types of satellite observations (or retrievals) related to
land surface hydrological conditions, including soil moisture, snow, and terrestrial water
storage (TWS; Bartalis et al.
2007
; Bruinsma et al.
2010
; Clifford
2010
; de Jeu et al.
2008
;
Entekhabi et al.
2010
; Foster et al.
2005
,
2011
; Gao et al.
2010
; Hall and Riggs
2007
; Hall
et al.
2010
; Horwath et al.
2011
; Kelly
2009
; Kerr et al.
2010
; Li et al.
2007
; Liu et al.
2011b
; Njoku et al.
2003
; Parinussa et al.
2012
; Pulliainen
2006
; Rowlands et al.
2005
,
2010
; Swenson and Wahr
2006
; Tedesco and Narvekar
2010
; Tedesco et al.
2010
; Wahr
et al.
2004
).
These observations can be assimilated into land surface models to provide land surface
hydrological estimates that are generally superior to the satellite observations or model
estimates alone (Andreadis and Lettenmaier
2006
; Crow and Wood
2003
; De Lannoy et al.
2012
; de Rosnay et al.
2012a
,
b
; Draper et al.
2012
; Drusch
2007
; Dunne and Entekhabi
2006
; Durand and Margulis
2008
; Forman et al.
2012
; Houborg et al.
2012
; Li et al.
2012
;
Liu et al.
2011a
; Margulis et al.
2002
; Pan and Wood
2006
; Pan et al.
2008
; Reichle and
Koster
2005
; Reichle et al.
2007
,
2009
; Sahoo et al.
2012
; Su et al.
2008
,
2010
; Zaitchik
et al.
2008
).
However, land data assimilation systems must be designed carefully such that a number
of conceptual problems can be overcome and the potential improvements from data
assimilation can be realized. Earlier work addressed the bias between the satellite obser-
vations and model estimates within the assimilation system (De Lannoy et al.
2007
; Drusch
et al.
2005
; Kumar et al.
2012
; Reichle and Koster
2004
). Moreover, approaches to effi-
cient error modeling within the assimilation system, including adaptive methods, needed
to be developed (Crow and Reichle
2008
; Crow and van den Berg
2010
; Reichle et al.
2008a
,
b
). An overview of some relevant earlier literature in the context of the ensemble-
based Goddard Earth Observing System Model, Version 5 (GEOS-5) land data assimilation
system (LDAS)
developed at
the
NASA
Global
Modeling and
Assimilation
Office
(GMAO) is provided by Reichle et al. (
2009
).
Despite the early successes, the design and application of land data assimilation systems
still face additional conceptual problems. While land surface models are flexible in the
design and choice of model variables, satellite observations do not necessarily correspond
directly to the water cycle variables of interest. For example, space-borne microwave
observations can be converted into estimates of snow amount or surface soil moisture, but
the spatial resolution of such microwave-based retrievals is usually much coarser than
desired. Moreover, satellites typically observe electromagnetic properties such as back-
scatter and/or radiances (or brightness temperatures) that are only indirectly related to
snow amounts or soil moisture levels. Furthermore, satellite-observed backscatter and
radiances are at best sensitive to moisture in the top few centimeters of the soil. Infor-
mation on important water cycle components such as root zone soil moisture must
therefore be gained through even more indirect pathways in the land data assimilation
system.
The present paper addresses several major challenges that all relate to a seeming
mismatch between the assimilated observations and the water cycle variables of interest.
This mismatch can be overcome through the careful design of the land data assimilation
system. The conceptual challenges discussed here can be summarized as follows:
