Intercomparison of Cloud Heights for AMV Generation

Atmospheric Motion Vectors (AMVs) are an important source of satellite data for assimilation into Numerical Weather Prediction (NWP) models.  A large driver of the accuracy of AMVs are the heights assigned to each vector.  In recent years, AMV applications are relying more and more on the cloud heights generated from external cloud height algorithms, including those participating in the International Cloud Working Group (ICWG).  AMV applications provide a direct link to the cloud height products and the NWP community.

While cloud height comparisons have been a common activity in the past Cloud Retrieval and Evaluation Workshops (CREW), the AMV application only uses a sub-set of cloud heights.  This subset usually consists of cloud edges and other features that are identifiable targets in the AMV tracking schemes.  To make a relevant analysis, we will use an AMV algorithm to select pixels which would serve as AMV targets and restrict our analysis to those pixels.  We will use the NOAA/CIMSS GOES-R AMV software and use 1DVAR heights from NOAA (ACHA) and EUMETSAT (OCA).  Our goal will be to compare the cloud height performance for the AMV-relevant pixels. Another aspect of this work is to compare the cloud height uncertainty estimates and other meta-data. We intend to collaborate with the International Winds Working Group (IWWG) and work towards standard and relevant definitions of our products and their meta-data.