Geoscience Reference
In-Depth Information
1. How can coarse-scale satellite observations increase our knowledge of land surface
conditions at finer scales (horizontal downscaling), and how can unobserved areas be
updated using information from neighboring observations?
2. How can vertically integrated measurements (such as TWS) be partitioned into their
component variables within the assimilation system?
3. How can satellite radiances (rather than geophysical retrievals) be assimilated to
improve estimates of land surface hydrological conditions (e.g., soil moisture and
snow)?
4. How can the most relevant types of observations be selected for the analysis of a water
cycle component that is not observed (such as root zone soil moisture)?
The present paper illustrates each of these conceptual problems based on recent progress
using the GEOS-5 system for land surface hydrological data assimilation. The examples
use satellite observations of land surface water cycle components from the Advanced
Microwave Scanning Radiometer for EOS (AMSR-E), the Moderate Resolution Imaging
Spectroradiometer (MODIS), the Gravity Recovery and Climate Experiment (GRACE)
mission, the Advanced Scatterometer (ASCAT), and the Soil Moisture Ocean Salinity
(SMOS) mission for the analysis of soil moisture (AMSR-E, ASCAT, SMOS, GRACE),
snow (AMSR-E, MODIS, GRACE), and TWS (GRACE). After a brief discussion of the
GEOS-5 LDAS, Sect.
2
provides details and references for the various satellite observa-
tions used in the examples. Section
3
addresses each of the above-mentioned challenge
questions in a separate subsection. Results are discussed and summarized in Sect.
4
.
Finally, Sect.
5
provides conclusions and a brief outlook on future research directions.
2 Data and Methods
2.1 GEOS-5 Land Data Assimilation System
The GEOS-5 LDAS consists of the NASA Catchment land surface model and an imple-
mentation of the ensemble Kalman filter (EnKF; Evensen
2003
). The GEOS-5 EnKF has
also been included in the NASA Land Information System, a comprehensive land surface
modeling and assimilation software framework, so that it can be used with a variety of land
surface models (Kumar et al.
2008a
,
b
). A brief summary of the key characteristics of the
system is provided below. For a more comprehensive discussion, see Reichle et al. (
2009
)
and references therein.
The Catchment land surface model (hereinafter Catchment model; Ducharne et al.
2000
;
Koster et al.
2000
) differs from traditional, layer-based land surface models by including
an explicit treatment of the spatial variation within each hydrological catchment (or
computational element) of the soil water and water table depth, as well as its effect on
runoff and evaporation. Within each element, the vertical profile of soil water down to the
bedrock is given by the equilibrium soil moisture profile and the deviations from the
equilibrium profile. The deviations are described by excess and deficit variables for a
0-2 cm (or 0-5 cm) surface layer and for a ''root zone'' layer that extends from the surface
to a depth z
R
of 75 cm B z
R
B 100 cm depending on local soil conditions. The spatial
variability of soil moisture is diagnosed at each time step from the bulk water prognostic
variables and the statistics of the catchment topography. One key feature of the Catchment
model is the groundwater component implicit in the modeling of the water table depth
(through the modeling of the subsurface water profile down to the bedrock). This
