Error Characterization of Atmospheric Motion Vectors through Intercomparison with ADM-Aeolus, NWP, and In-situ Observations

Atmospheric motion vectors (AMVs) are wind observations derived from tracking clouds and water vapor features in satellite images. Their assimilation in numerical weather prediction (NWP) systems has shown to positively impact operational forecasts. However, error characterization continues to be an issue as the derivation process can introduce uncertainties. In particular, height-assignment error tends to account for a large portion of AMV uncertainty and the total error. This study aims to improve AMV error characterization by leveraging an existing assessment tool, the NOAA Products Validation System (NPROVS), for data evaluation of AMVs relative to ADM-Aeolus reference wind observations, where ADM-Aeolus observations are satellite-derived horizontal line-of-sight (HLOS) global wind profiles.

The focus of this work is the integration of wind observations into NPROVS in order to evaluate characteristics of AMVs and eventually 3D-winds derived from hyperspectral IR-based water vapor soundings. NPROVS is comprised of a suite of applications used to compare consistently collocated data from global radiosonde networks, satellite-derived atmospheric soundings, and NWP products. At present, NPROVS has the capability to calculate statistics and display a variety of visualizations for collocated temperature and water vapor profiles, including global images, skew-T diagrams, scatter plots, and seasonal trends. For this study, we develop a component of NPROVS that integrates the AMVs and ADM-Aeolus data into the system. A triple collocation technique is implemented to provide consistent collocations across the wind products. Using the assessment tool, relevant statistics including height assignment biases of AMV data are evaluated relative to ADM-Aeolus, and the performance of each satellite-derived wind observation is intercompared with the other references including NWP, radiosonde, and aircraft flight level winds. Statistical analysis results as well as the potential improvement/better utilization of AMVs and 3D-Winds from assimilating ADM-Aeolus HLOS data in NWP are discussed.