Observing system experiments with
the DMI HIRLAM Optimum Interpolation analysis/three-dimensional
variational analysis and
Kristian S. Mogensen
Danish Meteorological Institute
October 25, 2000
In the past few years the numbers of ACARS (Aircraft Communication
Addressing and Reporting System) and AMDAR (Aircraft Meteorological Data
Reporting) aircraft observation data received at DMI have increased.
We have made two sets of observational system experiments (OSE) with these
observation types. One set has used the DMI HIRLAM Optimum Interpolation
(OI) data assimilation system in the (latest DMI standard) test
period (00 UTC 1 November to 18 UTC 29 December 1999).
Three parallel experiments have been run with OI,
one without and two including
the AMDAR/ACARS data. The runs including the data had two different
``time windows'' in which the data were used in the analyses,
±1800 s and ±3600 s as opposed to
the ±3 h used for other data types. The second set of OSEs
used HIRVDA, the HIRLAM 3D-Var data assimilation system, in the last
part (00 UTC 20 October to 18 UTC 15 November) of the EUCOS (EUMETNET
Composite Observing System) special period with an increased number
of AMDAR data. A revised redundancy check for aircraft observations
was used in some of the 3D-Var OSEs.
The results show that there is a clear positive impact when using
these data in the OI data assimilation cycle, and small
differences when increasing the time window from ±1800 s to
±3600 s with a slight tendency for the larger time window
to give better scores. For the HIRLAM 3D-Var data assimilation system the
results are mixed. Some runs with AMDAR/ACARS give better scores than
the run without and some runs with AMDAR/ACARS give worse scores.
SATOB (atmospheric motion vectors from geostationary satellites)
observational data received at DMI (in ASCII format) via GTS
have been tested with
the DMI HIRLAM OI data assimilation system in the same period as that
The results show
an overall negative impact on the DMI HIRLAM system from SATOBs, in
particular in the northern half of the model area.