Geoscience Reference
In-Depth Information
Draper et al. ( 2012 ) recently investigated combined data assimilation, using an Ensemble
Kalman Filter, of active and passive soil moisture satellite data. They evaluated the impact
of soil moisture products data assimilation on the analysed soil moisture. They showed
that, although correlation with ground data was better for the LSM than for the satellite
data, data assimilation still has a positive impact on the analysed soil moisture. Their study
confirms the potential of satellite-based soil moisture data for NWP applications.
4 Conclusion
This paper presented the current status of data assimilation systems used to initialise land
surface variables for Numerical Weather Prediction. Different approaches used to analyse
soil moisture and snow depth in Numerical Weather Prediction (NWP) systems were
reviewed. Based on ECMWF experiments, analysis results and atmospheric forecast
impact were presented for different land surface data assimilation approaches.
Snow processes strongly influence the hydrological cycle, and they have a large impact
on the energy budget. So, accurate initialisation of snow depth for NWP applications is
highly relevant. Snow analysis schemes currently used for operational NWP rely on simple
approaches. Using the NOAA IMS snow cover product, a simple update approach is used
at the UKMO. The NCEP latest reanalysis and the GLDAS also use a simple update
approach using combined IMS/SNODEP products and the MODIS snow cover product,
respectively. The German meteorological service relies on a Cressman interpolation and
used the SYNOP snow depth reports. Since 1999 the Canadian Meteorological Center uses
a 2D Optimal Interpolation to assimilate SYNOP snow depth observations. At ECMWF
both SYNOP observations and the IMS snow cover data are assimilated. A Cressman
Interpolation was used for more than 20 years before it was recently replaced by an
Optimal Interpolation in 2010.
A qualitative comparison between snow depth fields illustrated differences between the
Cressman and the Optimal Interpolation snow depth analyses. In contrast to Cressman, the
Optimal Interpolation accounts for the model background and the observations errors,
which allows to optimally combine model background and observations. So, by improving
the structure functions, the Optimal Interpolation makes a better use of the observations
than the Cressman interpolation. The quantitative impact of the snow analysis scheme on
the atmospheric forecast was illustrated using ECMWF results. The revised ECMWF snow
analysis, using an Optimal Interpolation and SYNOP observations combined with the 4-km
IMS snow cover data and improved observation pre-processing and quality control, was
compared to the old ECMWF snow analysis based on a Cressman interpolation that uses
SYNOP and the 24-km IMS snow cover products without quality control. Results showed
that the root mean square error forecast of the 1,000 hPa geopotential height in the northern
hemisphere is reduced by 1-4 % in the short range (until forecast day 4) for the winter
2009-2010. This significant and large-scale impact of the snow analysis on the atmo-
spheric forecast illustrates the major importance of the snow analysis for Numerical
Weather Prediction applications. SWE products from satellite sensors are not yet used in
NWP although they have a potential to provide reliable and near-real-time information on
snow mass. It is expected that Snow Water Equivalent products quality will be improved in
the next few years. Potential future satellite missions such as the proposed ESA CoReH2O
mission are expected to provide SWE estimates from space with an improved accuracy
compared to current products.
Search WWH ::




Custom Search