HIRLAM activities on the use of satellite data
Ad Stoffelen, KNMI
Several members of the HIRLAM group are conducting research and development on
the assimilation of satellite data in the HIRLAM model and these are organised
in a HIRLAM subproject. DNMI invited this group for a meeting on 6 November
1997 where the progress and plans in the area were discussed. Here follows a
brief account of the activities.
At the meeting Harald Schyberg presented the work done at DNMI to utilise
scatterometer data in the HIRLAM model. An error analysis method based on
division of observation errors into physical and observation space is used to
derive a general form of the observation error contribution to the cost
function. The method is utilised operationally for ambiguous scatterometer wind
observations at DNMI.
Ad Stoffelen presented results from an experiment on the use of scatterometer
data in the HIRLAM analysis/forecast system in a period where both ERS-1 and
ERS-2 data were available. The results showed a positive scatterometer impact
on the first guess in all experiments, and on the forecasts when data from both
satellites were used, but insignificant forecast impact with data from only one
satellite. With the increased data coverage from the NASA QuikSCAT (1998
launch), ADEOS-II SeaWinds (2000), and METOP ASCAT (2003) this is promising.
KNMI will develop and implement the scatterometer sea surface wind product of
the EUMETSAT Ocean & Sea Ice SAF, and be responsible for pre-processing, bias
tuning etc. In 1998 ambiguous wind vectors will become available through the
SAF. It was agreed that the processing system (PRESCAT) and assimilation system
for scatterometer data in OI at KNMI should be made operational and a part of
the HIRLAM reference system. KNMI will also work on the use of scatterometer
data from other satellites such as the planned NASA QuikSCAT and ADEOS-II
SeaWinds. QuikSCAT will be on a dedicated polar orbitter and planned for
real-time delivery at the beginning of 1999. It has a 1800 km broad swath.
QuikSCAT bridges a gap between ERS-2 and ASCAT. Due to its scanning geometry,
research and development is required in order to achieve an appropriate data
As 3dvar becomes available in HIRLAM, DNMI will implement variational
analysis of ERS scatterometer winds based on their operational version of
the cost function in 1998.
Ad Stoffelen presented the status and plans for the NWP SAF where KNMI plans to
be a partner together with the U.K. Met. Office, ECMWF and Meteo France. The
SAF will co-ordinate the interface between satellite data processing and data
assimilation. It was agreed that the SAF will be a useful tool for the HIRLAM
group as the results will be available for all EUMETSAT member states. In 1998
Ad Stoffelen will make an inventory of the HIRLAM requirements and priorities
as input to the NWP SAF work programme.
(A) TOVS work and plans.
There has been a problem in accomplishing the work at FMI as key-persons
have left FMI this year. However, Niko Sokka started in August, and he
presented the status of the TOVS work at FMI. 1D-var is implemented and
tested in the HIRLAM analysis with neutral impact. Work is ongoing with
cloud clearing, bias correction and better utilisation of the locally
received data on 40 km resolution. In the HIRLAM group FMI can usefully
contribute with their experience on 1D-var and with work on bias
correction and impact studies.
At SMHI Per Unden started working on the utilisation of TOVS data. The work
will be both on 1D-var and 3D-var, and, as SMHI is responsible for the high
latitude part of the nowcasting SAF, on cloud clearing problems.
DNMI started a 4-year project on the use of ATOVS. DNMI will especially look
into the observation error statistics and the problems of utilising the data at
high resolution. As DNMI has responsibilities for sea ice and SST in the Ocean
& Sea Ice SAF it will also be natural to focus on the effects of ocean and ice
KNMI is working on the implementation of the processing package AAPP, and also
assisted other HIRLAM members (DNMI) with this. The expected extensions of AAPP
(e.g., 1D-var) will also be implemented by KNMI.
(B) Other Satellite data
KNMI and Meteo France participate in the EU SODA project on assimilation
of ozone data. The work is on extraction of dynamic information around the
tropopause from the data, and on effects from ozone on the radiative
transfer model. The HIRLAM project awaits the success of this approach.
SMHI starts a project on 4D-var assimilation of ground-based GPS. In the Nordic
countries there are 50 stations working with 5 minutes time resolution giving
possibilities for indirect moisture observations. It was agreed that this
should be brought into the HIRLAM plans. Co-ordination should be considered
with GPS activities at DMI and KNMI.