Geoscience Reference
In-Depth Information
consistency between land surface fluxes and soil moisture (Best et al.
2011
; Balsamo et al.
2009
; Krinner et al.
2005
; de Rosnay et al.
2002
).
Land surface initialisation is of crucial importance for NWP. A number of studies have
shown a significant impact of soil moisture conditions on weather forecast skill at short and
medium range (van den Hurk et al.
2008
; Drusch and Viterbo
2007
; Douville et al.
2000
;
Mahfouf et al.
2000
; Beljaars et al.
1996
) as well as at seasonal range (Weisheimer et al.
2011
; Koster et al.
2011
,
2004
). Cold processes are also a key component of the land-
surface interactions. Snow is characterised by a very high albedo and a low thermal
conductivity, and the snowpack constitutes a substantial water storage reservoir (De
Lannoy et al.
2012
; Brown and Mote
2009
; Barnett et al.
2005
). Snow has a strong
influence on the summer water supply, and it affects the energy balance at the surface and
the surface-atmosphere interactions (Gong et al.
2004
; Walland and Simmonds
1997
). So,
initialisation of snow conditions has a large impact on the atmospheric forecast accuracy
(Drusch et al
2004
; Brasnett
1999
).
In this paper, methods used in operational NWP models to analyse LSMs' prognostic
variables are reviewed. Section
2
describes current snow analysis approaches used in NWP
centres. It presents ground and satellite observations of snow that are relevant to operational
applications and shows results of snow data assimilation experiments. Based on results from
the European Centre for Medium-Range Weather Forecasts (ECMWF), the impact on the
atmospheric forecasts is presented and compared for different snow data assimilation
approaches. Section
3
reviews soil moisture analysis systems used for NWP applications. It
includes a discussion on the use of satellite data to analyse soil moisture. ECMWF results
are shown to illustrate the influence of different soil moisture analysis approaches on surface
and low-level atmospheric fields. Concluding remarks are given in Sect.
4
.
2 Snow Analysis
2.1 Snow Forecast Models
Snow processes are parameterised in LSMs to account for a range of processes, including
snow accumulation on the ground, snow melting and snow compaction. The LSM used at
ECMWF is H-TESSEL (Hydrology Tiled ECMWF Scheme for Surface Exchange over
Land) (ECMWF
2012
; Balsamo et al.
2009
; Viterbo and Beljaars
1995
). H-TESSEL snow
parameterisation was revised in 2009 (Dutra et al.
2010
). It now accounts for liquid water
content in the snowpack, and it includes a new snow density formulation that expresses the
fresh snow density as a function of wind speed and air temperature. Snow Cover Fraction
(SCF) and Snow Water Equivalent (SWE) are related by a depletion curve which depends
on snow density. So, H-TESSEL represents the SCF hysteresis between accumulation and
depletion periods (Dutra et al.
2010
).
H-TESSEL has an explicit treatment of the snowpack evolution, and it uses a single-
layer snow model, in contrast to LSMs used at the United Kingdom Meteorological Office
(UKMO) or at M
´
t
´
o France, which include a multi-layered snow scheme (Best et al.
2011
; Dutra et al.
2010
; Boone et al.
2004
). Like most other LSMs, H-TESSEL represents
the effects of snow on the surface roughness length and for sub-grid scale processes.
Current LSMs represent well the duration of snow cover; however, they still have large
uncertainties in terms of snow accumulation, due to inaccuracies in the meteorological
forcing and to imperfect model parameterisations (Essery et al.
2009
; Boone et al.
2004
).
Data assimilation approaches, by optimally combining models and observations, are
