Proceedings - C-SRNWP Project

3D-FGAT
Tests with the 3D-FGAT method were started at the Hungarian Meteorological Service.
Originally the 3D-FGAT coded in ARPEGE/ALADIN adds the analysis increment to the
background at the beginning of the assimilation window, as it is a simplification of 4DVAR.
For better comparison with 3DVAR this feature was changed by adding the increment to the
background at the middle of the assimilation window. For the comparison of 3D-FGAT with
3DVAR several assimilation experiments were run using as many observations as possible.
Results show, that the FGAT method can improve the quality of the forecast at the beginning
of the integration for geopotential, wind and slightly for temperature but degrades the scores
for humidity in the PBL. On Fig 2. the impact of the FGAT is shown for geopotential as an
example.
Fig 2. Impact of the 3D-FGAT compared to 3DVAR. Red colors correspond to improvement, blue colors to
degradation due to FGAT
Further studies were dedicated to examine the impact of SYNOP observations in FGAT in
different ways. Possible ways are to use the SYNOPs in every timeslot (there are 7 slots in the
6 hour assimilation window) or just the one observed at analysis time. The results reflect that
a better analysis and forecast is provided if only that SYNOP report is assimilated which was
observed at the analysis time.
Observation use
During the first half of the year the main emphasis was put to the use of AMV (Atmospheric
Motion Vectors) data from MSG. The goal was to find the optimal use of these data with two
degrees of freedom given by the quality index of the data and the usability over land and sea.
The quality indexes (QI) are provided according to the actual quality of each observation. As
the number of available observation is decreasing with an increasing QI, it is the user’s choice
to set the optimal QI to be used. Concerning the usage of the data over land and sea, we
should mention that in the global ARPEGE model, AMV data are used uniquely over sea.
However as the European domain used at the Hungarian Meteorological Service (HMS)
includes mainly land surface, experiments were run using these high-resolution data over land
as well. Altogether 3 experiments were run within the Hungarian 3DVAR system (WDEF: QI
> 85%, data only over sea, W80P: QI>80%, data only over sea, WLAN: QI>85% data over
sea + land). The reference was the operational 3DVAR suite (NAM2). Results show that the
AMV data have a good impact on the analysis and the forecast especially for geopotential and
wind but also for humidity through the multivariate Jb (Fig 3). It can also be concluded, that
adding the AMV data over land can further improve the forecast (Fig. 4).
204