Understanding the Impacts of Assimilating the Atmospheric Motion Vectors on the Analyses and Forecasts of Tropical Cyclone Track and Structure

Atmospheric Motion Vectors (AMVs) are derived using a sequence of three satellite images to track targets. Special high-resolution (space and time) AMVs datasets were prepared hourly by CIMSS for the duration of Typhoon Sinlaku (2008) and Hurricane Ike (2008). Rapid-Scan AMVs, if available, were also processed for these two cases. Six parallel data-denial cycles are prepared, in which AMVs covering specified tropospheric layers and radial distances from the TC center are intentionally withheld from assimilation. These subsets of AMVs are assimilated into the Weather Research and Forecasting (WRF) model with a resolution of 27 km on the analysis grid (and 9 km in the forward forecast model) using the Ensemble Kalman Filter (EnKF) with 84 members. To identify the relative contributions of the different tropospheric layers of AMVs to the analyses and forecasts of the TC and its environment, three data-denial cycles are designed: eliminate all AMVs between 150-350 mb; eliminate all AMVs between 350-700 mb; and eliminate all AMVs between 700-999 mb respectively. Two further data-denial cycles are prepared, with AMVs withheld within (outside) 10 degrees from the TC center, in order to investigate the contribution of near-storm versus environmental AMVs on the TC structure and track forecasts.

While there is some variance between the results of the two TC cases, they both suggest the interior, lower- and upper-layer AMVs are crucial at maintaining the model structure of the vortex primary and secondary circulations. The environmental upper-layer AMVs are most important at reducing track errors. It is also evident, especially in the case of Sinlaku, that the forecasts initialized from the analyses in which the Rapid-Scan AMVs are assimilated capture the track re-curvature while the control runs have the storm making landfall in southeastern China.