Improving Ku-band Scatterometer Ocean Surface Wind Direction Retrievals in Tropical Cyclones

Tropical cyclones are regions of very strong rain and very high winds, both of which present major challenges to surface wind vector retrieval from Ku-band scatterometers. Wind speed and wind direction retrievals can incur severe errors in regions of high rain rates. One particular signature of rain contamination is wind directions in the across-swath direction, which often leads to displaced circulation centers. Recently, Stiles et al. (2014) developed a method for retrieving QuikSCAT tropical cyclone wind speeds using a neural network approach that was tuned using H*WIND surface wind analyses and passive microwave-estimated rain rates from satellites. We are developing a scene-wide methodology by which a set of dynamically-consistent wind directions can be estimated from these wind speeds. The method is based on an iterative use of a tropical cyclone-specific sea-level pressure retrieval technique that we developed. The sea-level pressure analysis uses a boundary layer model that includes the dynamical shallowing of the tropical cyclone boundary layer toward the storm center, a roll-off in surface drag at high wind speeds, and, storm motion-corrected nonlinear mean flow advection effects. Scene-wide consistency is enforced by the integral nature (with respect to the surface wind vector field) of the derived surface pressure pattern and a constraint that the geostrophic contribution to the total flow is non-divergent. We are currently developing methods to evaluate the retrieved wind directions based on HRD aircraft observations and a limited-domain wind vector partitioning of the retrieved wind vectors into irrotational, non-divergent, and, background flow deformation contributions.