F. Rabier et al., SENSITIVITY OF FORECAST ERRORS TO INITIAL CONDITIONS, Quarterly Journal of the Royal Meteorological Society, 122(529), 1996, pp. 121-150
The adjoint method has been used to calculate the sensitivity of short
-range forecast errors to the initial conditions. The gradient of the
energy of the day 2 forecast error with respect to the initial conditi
ons can be interpreted as a sum of rapidly growing components of the a
nalysis error. An analysis modified by subtracting an appropriately sc
aled vector, proportional to the gradient, provides initial conditions
for a 'sensitivity integration' that can be used to diagnose the effe
ct of initial-data errors on forecast errors. Statistics of sensitivit
y calculations for the month of April 1994 characterize the sensitivit
y patterns as small-scale, middle or lower tropospheric structures whi
ch are tilted in the vertical. The general pattern of these structures
is known to be associated with the fastest possible growth of forecas
t error When used as initial perturbations, they evolve rapidly into s
ynoptic-scale structures, propagating both downstream and to higher at
mospheric levels. On average, the sensitivity integration corrects for
about a tenth of the day 2 forecast error, which indicates that indee
d not all of the error is in the fastest-amplifying modes. But the fra
ction of the error corrected at day 2 is important for an improvement
in the medium-range, as this fraction continues to grow substantially
in the non-linear regime. These results have proved that there is stil
l scope for great improvement in the medium-range forecast, particular
ly over Europe, by a better description of the initial conditions. The
sensitivity experimentation suggests that many cases of major forecas
t-errors may be explained by defects in the analysis. A small but well
-chosen change in the analysis can frequently improve the forecast qua
lity.