A New Technique for Composite Maps of Ocean Winds
Time averaged (3 days, weekly, monthly) composite maps of wind speed derived from LEO satellite microwave radiometers display clear artifacts, typically visible along the edges of successive orbital swaths. These spatial discontinuities result from the temporal variability of the geophysical variable, calibration errors or uncertainties in Level 2 retrieval algorithms, and can have a major impact on downstream applications. Using computer vision and machine learning techniques, ORBTY has recently developed the Metocean Multi-Dimentional Dynamic Data Fusion Processor (M3DFP), an innovative system that seamlessly ingests multitemporal Level 2 datasets from a single or multiple instruments to produce Level 3 maps with significantly improved spatio-temporal representation of the geophysical parameter of interest.
Initially designed for Sea Surface Temperature (SST) observations, the M3DFP has been generalized to other satellite-derived metocean variables.
In this work, we show the applicability of the M3DFP system for winds derived from AMSR-2, GMI and Windsat and highlight the improvement achieved in the homogeneity of wind speed’s spatial distribution.