A complete observation error correlation matrix from CYGNSS

The Cyclone Global Navigation Satellite System (CYGNSS) observes global ocean surface windspeed from a constellation of seven small satellites that measure GPS signals that scatter from ocean surfaces. Not only are these measurements related in space in time, but so are the errors from these measurements. Traditional mechanisms of data assimilation frequently thin or ‘super-ob’ datasets in order to maintain the characteristics of independent and uncorrelated errors between samples. These techniques have a strongly deleterious impact to CYGNSS utility due to the sparse nature of CYGNSS sampling on ocean surfaces.

Based on prior study, an extensive model of instrument-derived correlated error has been conceived and propagated through the Level 2 windspeed retrieval. Further, fundamental geophysical errors and their decorrelation behaviors in space and time have been evaluated for surface windspeed measurements. Together, a complete observation error correlation matrix can be defined for data assimilation purposes, which can optimally weight the values of samples for maximum novel information content and enhance forecast skill.