Geoscience Reference
In-Depth Information
•
There is a need to exploit the simultaneous use of multiple sensors (Pan et al.
2008
;
Draper et al.
2012
) and explore the capabilities of new sensors (Andreadis et al.
2007
;
Durand et al.
2008
);
•
There is a need to combine state and input (forcing) information with parameter
updates (Moradkhani et al.
2005b
; Liu et al.
2011
; Vrugt et al.
2012
);
•
There is a need to explore advanced filtering techniques, for example, the use of the
particle filter to account for non-Gaussian errors (Plaza et al.
2012
);
•
There is a need to improve the representation of observation and forecast errors, and to
specify biases in observational and model information (De Lannoy et al.
2007b
; Crow
and Reichle
2008
; Reichle et al.
2008
; De Lannoy et al.
2009
; Crow and van den Berg
2010
);
•
There is a need to preserve water balance in the land system (Pan and Wood
2006
;
Yilmaz et al.
2011
) and draw lessons from the information in the assimilation
increments;
•
There is a need to have access to adequate computational resources.
5.5 Validation
•
Needs ground observations with substantial spatial and temporal coverage;
•
Needs tools to address scaling and representativeness errors (Crow et al.
2012
);
•
Needs appropriate and effective validation metrics (Entekhabi et al.
2010b
).
6 Conclusions
To understand the hydrological cycle over land, we need to make observations and develop
models that encapsulate our understanding. These models have a basis on the information
gathered from observations, as well as on previous experience, and are used to project our
understanding into the future by making predictions. A crucial element in this procedure is
confronting models with observations. Data assimilation, which combines observational
and model information, provides an objective method to confront models against obser-
vations and add value to both the model and the observations. Data assimilation adds value
to observations by filling the gaps between them and adds value to models by constraining
them with observations. In this paper, we touch on the main conceptual problems that limit
a full integration of land surface models and observations by reviewing progress in land
surface data assimilation research over the last decade.
Collectively, the advent of new satellite missions, the increasing attention to forecast
uncertainty due to errors in the land surface model structure, parameters and input, and the
development of advanced assimilation techniques will eventually close the largest gaps in
our understanding of the hydrological cycle over land.
Acknowledgments This paper arose from the International Space Science Institute (ISSI) workshop
''The Earth's Hydrological Cycle'', held at ISSI, Bern, Switzerland, on 6-10 February 2012. A NILU
internal project supported WAL. Thanks to Alexandra Griesfeller for providing Fig.
2
.
Open Access This article is distributed under the terms of the Creative Commons Attribution License
which permits any use, distribution, and reproduction in any medium, provided the original author(s) and the
source are credited.
