Geoscience Reference
In-Depth Information
cover information in all weather conditions. It provides a binary information on snow
cover. In other words, it indicates if there is snow or not on the ground, but if there is snow,
it does not indicate the snow quantity on the ground. The IMS product is available daily for
the northern hemisphere (Helfrich et al.
2007
; Brubaker et al.
2009
; Ramsay
1998
). It is
available from 1997 at a resolution of 24 km (Ramsay
1998
) and from 2004 at 4 km
resolution (Helfrich et al.
2007
). The NOAA NESDIS IMS snow cover product has been
used to analyse snow in NWP systems at ECMWF and the UKMO since 2004 and 2008,
respectively (Pullen et al.
2011
; de Rosnay et al.
2011b
; Drusch et al.
2004
). It is also used
in the latest National Centers for Environmental Prediction (NCEP) Climate Forecast
System Reanalysis (CFSR) (Meng et al.
2012
).
2.3 Snow Analysis Methods
A number of NWP centres recently developed snow analysis approaches to improve the
initialisation of snow variables, with expected impacts on the near surface weather
parameters.
The UKMO snow analysis was implemented in operations 2008. It entirely relies on the
NOAA NESDIS IMS 4 km Snow Cover information (Pullen et al.
2011
). As part of the
IMS pre-processing, the 4-km product is interpolated on the Unified Model grid, and a
snow cover fraction is computed for each model grid point. To correct the model snow
depth prognostic variable, a simple update approach is used, as described by Pullen et al.
(
2011
). If the IMS product indicates snow-free conditions, the analysed snow depth is set to
zero. Otherwise, the IMS snow cover is compared to the model background. If IMS
indicates a region is snow covered and the model background agrees, the model is simply
cycled, that is, the analysed snow depth is set to the background snow depth. If IMS
indicates a region is covered by snow while the model background is snow free, the
analysed snow depth is computed as a function of the observed snow cover using a
logarithmic depletion curve.
The NASA/NOAA GLDAS snow analysis follows a similar approach along the same
line, using the MODIS snow cover product (Rodell and Houser
2004
). The NCEP CFSR
reanalysis also relies on a simple update approach, with input data resulting from combined
IMS and Air Force Weather Agency's snow depth analysis (SNODEP), as described by
Meng et al. (
2012
).
Most of other NWP services use SYNOP snow depth reports available on the GTS. The
snow analysis is a spatial interpolation of weighted background and observed snow depth.
The DWD (Deutscher Wetterdienst) assimilates SYNOP reports of snow depth using the
Cressman (
1959
) interpolation. The Cressman analysis accounts for weighting functions of
vertical and horizontal distances between observations and model grid points. The Cana-
dian Meteorological Center (CMC) uses a 2D Optimal Interpolation (OI) scheme
developed by Brasnett (
1999
). Similarly to the Cressman interpolation, the OI expresses
the observations weighting functions from vertical and horizontal structure functions. In
addition, it accounts for covariance matrices of background and observations errors which
enable to optimally combine model background and observations. At ECMWF, the latest
ECMWF Re-Analysis, ERA-Interim, uses a Cressman interpolation for the snow analysis
(Dee et al.
2011
). The operational snow analysis relied on a Cressman interpolation for
more than 20 years until it was replaced by a 2D Optimal Interpolation in November 2010
(de Rosnay et al.
2011b
). The ECMWF operational snow analysis is a two-step algorithm.
In the first step, a simple update scheme similar to the one used at NCEP or at the UKMO
is used to account for the IMS snow cover information. Grid boxes, which are snow free in
