Observation Influence Diagnostic of a Data Assimilation System - Data Assimilation for Atmospheric, Oceanic and Hydrologic Applications - page 99

GOES-Ra O3

MTSAT-Rad

MET-Rad

AMSU-B

MERIS
MHS

TMI-1

SSMIS

AMSR-E

GPS-RO

IASI

AIRS

AMSU-A

HIRS

SCAT

MODIS-AMV

MTSAT-AMV
Meteosat-AMV

GOES-AMV

PROFILER

PILOT

DROP

TEMP

DRIBU

Aircraft

SYNOP

0 2 4 6 8 0 2 4 6 8 0 2 4

DFS %

Fig. 4.3 Degree of Freedom for Signal ( DFS ) of all observations assimilated in the ECMWF
4DVar system in October 2011. Observation types are described in Table 4.1

The difference between OI and DFS comes from the number of observation
assimilated. Therefore, despite the generally low observation influence of satellite
measurements, they show quite large DFS because of the large number assimilated.
A large discrepancy between OI and DFS points on those observation types where
a revision of the assigned covariance matrices R and B will be beneficial: more
information extracted from e.g. satellite measurements.
Another index of interest is the partial Observation Influence ( OI m /

for any

selected subset of data

P
i 2 I S ii
m I

OI m D (4.33)

where can represent a

specific observation type, a specific vertical or horizontal domain or a particular
meteorological variable. In Fig. 4.4 the OI of Aircraft data (

m I is the number of data in subset I . The subset I

) is plotted as a
function of pressure layers and for all observed parameters: temperature (t), zonal
(u) and meridional (v) component of the wind. The largest OI is provided by
temperature observations ( 0:4

I

) similar distributed on the different pressure layers.
Wind observations have larger influence (0.4) on the top of the atmosphere (above
400 hPa) than on the bottom one (0.2) due to the fact that there are very few wind
observations on the troposphere and lower stratosphere mainly over the oceans.
At those levels, temperature information is also provided by different satellite
platforms (in terms of brightness temperature or radiance). In Fig. 4.5 the Aircraft
DFS with respect to different pressure levels and observed parameters is shown. The
largest DFS in the lower troposphere (below 700 hPa) for temperature measurements
( 10 % with respect to the total Aircraft DFS ) with respect to wind ones is due to