Characteristics of Ensemble Transform Kalman Filter adaptive sampling guidance for tropical cyclones

The Ensemble Transform Kalman Filter (ETKF) is one candidate strategy for targeting observations around tropical cyclones. The characteristics of target areas identified by a revised version of the ETKF are investigated using the THORPEX Interactive Grand Global Ensemble. In order to emphasize sensitivity to the environmental flow, the axisymmetric circulation associated with the tropical cyclone is removed from each ensemble member. The guidance is found to differ markedly from the ensemble variance, and the guidance products based on ensembles from different operational centres often disagree. For the pre-recurvature stage of typhoon Sinlaku (2008), the targets correspond to asymmetries close to the typhoon, neighbouring features including the adjacent subtropical ridge and a nearby upper-level trough, and far-field features including the monsoon trough and areas of flow convergence. During and after recurvature, targets exist in the midlatitude jet upstream and downstream of the typhoon. The midlatitude targets are discernible for forecasts beyond 3 days. The sensitivity to hypothetical observations of wind is more pronounced than those of temperature or specific humidity. The targets based on upper-level observations correspond to the outflow of the tropical cyclone and nearby and remote jets, while the lower-level targets correspond to confluent flow near the cyclone. Multiple remote targets are evident, particularly for longer forecast times, and some of these may be spurious. When ensemble forecasts initialized at different times are used for the same case, the characteristics of the guidance differ, although not dramatically. Future priorities include the evaluation of the effects of assimilating observations sampled in target areas versus those in non-target areas, and the improvement of the data assimilation methodology assumed in the ETKF.