Data assimilation into NWP systems
Operational weather centres have been innovative in the development of
data assimilation for application to operational weather forecasting.
By the mid-1970's, the major weather agencies were using optimal interpolation
(OI), which is a minimum variance estimation method. Subsequent developments
include 3d-var and 4d-var techniques.
Despite these developments, there remains ample scope for further development
of the theory and practice of data assimilation. Efforts at the weather
centres have concentrated on the use of data that are available in near
real time, and so has been restricted chiefly to data from operational
weather satellites, such as the NOAA series of polar orbiters carrying
nadir-viewing instruments. Comparatively little work has been done until
recently to assimilate data from research satellites, mainly because the
data are not often available in real time, or may have limited lifetimes,
and therefore the weather centres have little incentive to develop techniques
to cater for such data. The advantages of data assimilation have therefore
not been exploited to reap full benefit from hugely expensive research
satellite programmes.
A number of initiatives have recently been taken to remedy this serious
shortcoming. In the USA, a Global Modeling and
Assimilation Office (GMAO) has been set up to advance the state of
the art of data assimilation, and to produce high quality assimilated
datasets that are needed to address questions concerning the Earth system
and global change. In Europe there are several collaborative projects
involving the weather agencies and the academic community, including the
Data Assimilation Research
Centre (DARC) in the UK, and the SACADA (Synoptic Analysis of Chemical
Constituents by Advanced Data Assimilation) initiative in Germany.
The ASSET research programmes aim to assimilate in the immediate term
temperature, ozone, water vapour and a comprehensive suite of photochemical
species from instruments aboard Envisat. In the longer term, these programmes
are seeking to build the framework for a European capability for the effective
exploitation of data from future research satellites (as well as from
future operational satellites), including feedback on the design of future
observing systems (including the ground segment).
Expected results from these programmes (involving NWP systems, as well
as models with sophisticated photochemistry -- see below) would be:
(a) a European capability for the effective exploitation of current,
past and future EO data,
(b) a European capability for chemical and UV forecasting, and
(c) a European capability for coupled climate/chemistry modelling.
In ASSET, the following NWP assimilation configurations will be used:
(1) an NWP system with parametrized photochemistry (UREADMY/METO),
(2) an NWP system coupled to a CTM (GAME-MF/GAME-CNRS/CERFACS),
and
(3) an NWP system adapted to assimilate limb radiances (ECMWF).
Configurations (1) and (2) will assimilate retrieved profiles of temperature,
ozone, water vapour and other chemical species. Configuration (3) will
focus on radiance data for temperature and ozone.
Data assimilation into models with sophisticated photochemistry
Assimilation of research satellite data (e.g., ozone profiles and column
amounts) into models with sophisticated photochemistry, e.g, Chemistry-transport
Models (CTMs), is increasingly taking place (SODA
1999). The relatively simpler configuration of a CTM compared to an
NWP system allows one to include a large number of chemical species and
provides a tool for investigating the distribution and variability of
atmospheric photochemical species, testing photochemical theories, and
producing climatologies of observed species and of unobserved chemical
species (using the model photochemical relations).
The chemical assimilation techniques include the simplified Kalman filter,
OI and 4d-var, with most European groups using 4d-var.
Many of these groups are involved in programmes to assimilate research
satellite data from NASA and ESA missions. In ASSET, the following chemical
assimilation configurations will be used:
(1) trajectory (i.e. Lagrangian) models using 4d-var (KNMI,
UPMC), and
(2) Eulerian models using 4d-var (BIRA-IASB,
FRIUUK)
Other configurations such as the Kalman filter may also be used for short
experiments. These configurations will be used to assimilate a comprehensive
suite of photochemical species.
|
= ASSET members and
PAG; others may request access from the coordinators