Adaptive observation strategies for lidar observations

Doppler Wind Lidar (DWL) measurements require considerable energy, and it would desirable to be able to “observe only 10% of the time and obtain 90% of the potential impact”. Through Observing System Simulation Experiments (OSSE), we study the effectiveness of using the ensemble spread from the Local Ensemble Transform Kalman Filter (LETKF) to determine adaptively the optimal observation locations, since the ensemble spread is a measure of the uncertainty in the forecast. We compare several strategies choosing 10% of the possible locations (maximum ensemble spread, uniform distribution, random location and ensemble spread climatology), with the impact that would be obtained observing at 100% of the locations. Results show that maximum ensemble spread is the best strategy, and that 10% observations provide 90% of the impact when using LETKF ensemble spread even when these observations are assimilated within a 3D-Var data assimilation system, indicating that the ensemble spread does indeed provide information about where the large “errors of the day” are located.