| Home > Research > Ifsdocs > OBSERVATIONS > |
IFS documentation Front Page
Chapter 2. Observations: Types, variables and error statistics
Chapter 3. CMA creation (MAKEMA)
Chapter 4. The FEEBACK task
Chapter 5. The TOOLS task
Chapter 6. Central-memory array (CMA) structure/format
Chapter 7. BUFR feedback data structure/format
Chapter 8. SIMULATED-observations data structure/format
Chapter 9. NAMELISTS
Chapter 10. Processing of scatterometer data
REFERENCES
Previous Section
10.4 Bias corrections
Two separate bias corrections are included in ERS1IF to improve the accuracy of the winds retrieved with CMOD4.
A
bias correction is first performed before the wind retrieval,
by substracting constant bias estimates from the raw backscatter measurements
as a function of their antenna and node numbers. These bias estimates, derived
from a routine comparison between the 
s measured by the scatterometer and
the 
s simulated by CMOD4 from the first-guess
winds of the ECMWF model, are supposed to account both for the variations
that may occur in the instrumental calibration in time and for the residual
defaults affecting the fit of the transfer function in the backscatter space.A wind-speed bias correction is then added following the wind retrieval, in the form of a cubic spline function applied to the retrieved wind speeds that is dependent on the measurement node number. The purpose is now to match the scatterometer and model wind speeds over the whole wind-speed range (especially at high winds where CMOD4 tends to be biased low) so as to avoid introducing any speed-up or slow-down tendency in the assimilation process. Like the
bias correction, this wind-speed-dependent
bias correction relies on a direct comparison between scatterometer and
model data, in which the wind speeds retrieved with the 
bias correction are fitted as a function
of those deduced from the model first guess according to a Maximum Likelihood
Estimation (MLE) procedure. However, conventional observations from ships
and buoys are also taken into account, to first assess the respective errors
of both systems through a triple-collocation analysis. Furthermore, no time
variations are considered here, since these are assumed to be already described
by the 
bias term.The
and wind-speed bias corrections are defined by two dedicated
files read in the initialization subroutine INIERSCA, and containing appropriate
coefficients both for ERS-1 and ERS-2. The 
bias file is normally updated on a monthly
basis, the bias applied over a given month being computed from the data
from the month before, whilst the wind-speed-bias file is kept constant
as a rule. More information about these bias corrections and their derivation
can be found in Le Meur et al. (1997).It should be noted that the corrections made are not kept explicitly in the scatterometer CMA reports, where the main outputs are limited to the retrieved wind components as well as to the distances to the cone and the associated quality-control flags. Moreover, the original
measurements are also stored, together
with the ESA-retrieved wind speeds and directions, to allow subsequent data
monitoring from the analysis-feedback file.Next Section
Previous Section
 |
18.04.2002 |
 |
© ECMWF |
