Group seminar on 12. July, 14:15 CET

Information content of the mass-field and wind-field information in a perfect-model global ensemble Kalman filter data assimilation system

Lanqin Li

 Previous research on analysis uncertainties and short-term forecast errors in the state-of-the-art numerical weather prediction (NWP) models has shown that both analysis and forecast errors are largest in the tropics. This is not surprising given challenges of tropical data assimilation, complex tropical dynamics and wind observation availability. At the same time, modern data assimilation systems are capable to extract information about unobserved variables (for example, wind and moisture) from available observations (for example, temperature), especially in 4D-Var. Given this fact, we ask to what extent the classical argument about a greater value of wind observations than mass observations in the tropics and at small scales still applies. This question is answered by comparing the outputs from three observing system simulation experiments (OSSEs) that were conducted by the DART team at NCAR using the ensemble Kalman filter (EnKF) data assimilation and a perfect-model framework. The three experiments assimilated only wind observations, only temperature observations, and both types of observations.
In the seminar, I will introduce the OSSE methodology and present a preliminary comparison of the three OSSEs. The analysis will focus on the comparison of ensemble reliability and forecast uncertainties in physical space in the three experiments. Preliminary results suggest that the assimilation of wind observations play a more important role for analysis accuracy than temperature data. The same applies to the temperature and humidity fields. In the tropics, the reduction of uncertainties in short-term forecasts by the assimilation of wind observations is greatest in the upper troposphere for all variables.