CYGNSS wind retrievals will be collocated in space and time with truth data. Typically truth data will include, but are not limited to, the NWS global forecast model analysis (GDAS) fields, buoys, ASCAT, OSCAT, WindSat, AMSR-2, and aircraft measurements during hurricane reconnaissance flights. The standard statistical analysis used for satellite microwave wind sensors will be utilized to characterize the CYGNSS wind speed retrievals. The global numerical weather prediction (NWP) models are a convenient source of truth data because they are available 4 times/day globally which results in the accumulation of a large number of collocations over a relatively short amount of time. The NWP model fields are not truth in the same way an actual observation would be, however, as long as there are no systematic errors in the NWP model output the collocations will converge in the mean for winds between approximately 3-20 m/s. The NWP models typically do not properly resolve the very low and high wind speeds in part due to limitations of the spatial scales they can account for. Buoy measurements, aircraft-based measurements and other satellite retrievals can be more directly compared on a point-by-point basis. Limited buoy locations, flight locations, and different satellite orbits tend to geographical restrict these collocations. Additionally, satellite microwave radiometers and scatterometers tend to have difficulty resolving what is happening within the tropical cyclone environment because of atmospheric contamination and spatial resolution issues. This is why a comprehensive validation requires utilizing all available truth data for comparisons, understanding and ultimately retrieval algorithm refinement.
With spatial resolution and measurement uncertainty being tunable parameters, additional analysis will be required to fully understand and characterize the CYGNSS retrievals within the tropical cyclone environment. Within tropical cyclones there are typically strong wind gradients occurring over relatively small spatial scales. Thus one important aspect that we will need to address is what is the proper balance between acceptable uncertainty in the wind speed retrieval and the spatial resolution of the retrieval, where finer spatial resolution will required to resolve the higher wind speeds in the strong wind gradient regions of the tropical cyclone. Another aspect of the CYGNSS retrievals that warrants further investigation is the impact of rain. While CYGNSS retrievals are relatively immune to the propagation path effects of liquid and cloud liquid water, the modification of the surface by rain may have an impact on the CYGNSS retrievals, particularly at the lower wind speeds found in tropical depressions and tropical waves. The improved understanding derived from these analyses will be fed back into the CYGNSS retrieval and quality flagging to achieve a more robust wind product. A summary of the analyses of these simulated data will be presented and discussed.