Modern data assimilation schemes produce a "best" analysis of the
current state by combining a model forecast and observations.
Information from the observation is spread to the model state variables
using the forecast error statistics. Whereas most operational data
assimilation schemes assume these error statistics, the ensemble Kalman
filter (EnKF) scheme provides flow-dependent error estimates, and in
theory, more optimal usage of observations. In addition, the EnKF
generates a sample of equally likely analyses that can be used to
estimate the analysis error and can be used for ensemble forecasting.
Since December 2004, Ryan Torn
maintained a real-time psuedo-operational Weather Research and
Forecasting (WRF) model EnKF for a domain over the eastern Pacific and
western North America. This system was build as a proof-of-concept test
of the performance of an EnKF in a region of sparse in-situ data and to
do basic research on predictability and observation network design (real time EnKF
Although the WRF EnKF system assimilates approximately only ~2% of the
observations used by other operational systems, the forecasts are
comparable in skill.
In addition to providing probabilistic analyses and forecasts, data
from this system are being used to understand forecast sensitivity and
the dynamics of weather systems.
Figure: Analysis ensemble mean SLP (contours) and spread (shading; an
estimate of analysis error) from the UW WRF EnKF system valid 00 UTC 28
Ancell, B., and G. J. Hakim, 2006: Comparing adjoint and ensemble sensitivity analysis. Mon. Wea. Rev., 133, submitted. (pdf)
Dirren, S., R. D. Torn, and G. J. Hakim, 2006: A data assimilation case-study using a limited-area ensemble Kalman filter. Mon. Wea. Rev., 133, accepted. (pdf)
Hakim, G. J., and R. D. Torn, 2006: Ensemble Synoptic Analysis. Fred Sanders Monograph, American Meteorological Society, accepted. (pdf)
Torn, R., D., G. J. Hakim, and C. Snyder, 2006: Boundary conditions for limited-area ensemble Kalman filters. Mon. Wea. Rev., 134, 2490--2502. (pdf)
Dirren, S., and G. J. Hakim, 2005: Toward the assimilation of time-averaged observations. Geophys. Res. Lett., 32, L04804, doi:10.1029/2004GL021444. (pdf)