Characterization of tracer height assignment errors: A recurring theme for satellite-derived atmospheric motion vectors

Atmospheric Motion Vectors (AMVs) derived from geostationary, and now polar satellites, provide invaluable information regarding the atmospheric wind field that has been successfully leveraged within operational Numerical Weather Prediction (NWP) data assimilation systems over the past ten years. AMVs derived from geostationary satellites continue to be indispensable over the world oceans and in the tropics. More recently, AMVs derived from the Terra and Aqua polar orbiting satellites, have been shown by the NWP community to be invaluable in improving analyses of the polar wind field, and at the same time, improve the quality of forecasts in the polar regions. The satellite wind community has worked together over the past ten to fifteen years to improve the quality of the AMVs and certainly can take a significant portion of the credit for the success that these observations continue to make on NWP. The NWP community can certainly take its share of this success given improved utilization of the AMV observations as well as improvements made to operational data assimilation schemes. However, despite these successes, the NWP community continues to ask the satellite wind community for some fundamental information regarding the quality of the AMVs. Specifically, what portion of the AMV error can be attributed to height assignment? Height assignment errors associated with the AMV data have long concerned the members of the wind producing and NWP communities alike. Tracer height assignment remains a key issue because, in many cases, it is the largest contributor to the total AMV error that can render them useless or cause significant problems for a NWP data assimilation system. The goal of this work is to characterize the errors of the height assignments assigned to the AMVs generated at NOAA/NESDIS. Results from a “Level-of-Best-Fit” analysis, that involves the utilization collocated rawinsonde observations, GOES AMVs, and MODIS AMVs, to characterize tracer height assignment errors as a function of height assignment method, latitude, and season, will be presented. Only after this analysis is completed can we begin to answer the fundamental questions being asked by the NWS community regarding what component of the AMV error is attributed to errors in tracer height assignment. Results of this analysis may also provide the necessary information that should lead to further improvements in the height assignment algorithms currently in use today.