Title: Diagnosing spatial and inter channel observation error statistics for Doppler radar radial wind and SEVIRI observations

Author: Joanne A. Waller (University of Reading)
S. L. Dance (University of Reading)
N. K. Nichols (University of Reading)
D. Simonin (Met Office)
S. P. Ballard (Met Office)
G. Kelly (Met Office)

With the development of convection permitting numerical weather prediction the efficient use of high resolution observations, such as Doppler radar radial winds and SEVIRI radiances, in data assimilation is becoming increasingly important. These observations are now routinely assimilated in operational systems, though to avoid violating the assumption of uncorrelated observation errors it is necessary to reduce the density of the observations. This is achieved both by the use of superobservations and observation thinning. Taking into account the full, potentially correlated, error statistics will allow the quantity of observations used to be increased and may improve the impact that the observations have in the assimilation. In this work we use a diagnostic that makes use of statistical averages of background and analysis innovations to calculate observation error statistics (for further detail see abstract of Dance et al.). Spatial error covariances are calculated for the Doppler radar radial winds and spatial and inter-channel error statistics for SEVIRI radiances that are assimilated into the Met Office 1.5km model. Results for the radar data show that the error standard deviations are similar to those used operationally and increase as the height of the observation increases. The observation correlation length scales found using the diagnostic are larger than operational thinning distance of 6km and are dependent on both the height of the observation and on the distance of the observation away from the radar. Additional tests show that the long correlation length scale is unlikely to be attributed to the background error covariance matrix used in the assimilation or to the use of superobservations. It is possible, however, that the use of a simplified observation operator results in some of the large length scale horizontal correlations.. For the SEVIRI radiances the error standard deviations are much smaller than those used operationally. The observations exhibit no horizontal correlation as they are thinned to 24km. However, there is significant inter-channel correlation.

GMAO Head: Steven Pawson Global Modeling and Assimilation Office NASA Goddard Space Flight Center

Curator: Nikki Privé Last Updated: Feb 9 2015