Converging Remote Sensing and Data Assimilation Techniques: A Data Fusion Approach for Near-real-time 4D Analyses

A new approach to data fusion is under development as part of a pilot project at the National Oceanic and Atmospheric Administration’s (NOAA) Center for Satellite Applications and Research (STAR). This approach is, in a sense, data fusion at the radiometric level: leveraging NOAA’s current data assimilation (DA) systems and incorporating satellite remote sensing algorithms in order to create and observation-driven, 4D, global analysis. The system presented here utilizes a comprehensive set of observations from the current global observing system, including from both satellite (radiances) and conventional (in-situ) sources. The data fusion system has three main components: a pre-processor that combines multiple satellite retrieval algorithms for the purpose of performing adjustment to DA background fields and providing observation quality control and other boundary conditions, the 3DVAR DA using the Gridpoint Statistical Interpolation (GSI), and a post-processor which includes diagnosed products based on the analyzed fields. The data fusion tool is able to provide hourly, observation-driven analyses at 30km resolution (with the potential of increasing to sub-hourly, 15km grids) over the global domain. The analysis includes the traditional suite of remote sensing products (e.g. soundings, precipitable water, rainfall rate, cryospheric parameters, trace gases), as well as a number of added-value products (e.g. vorticity, geopotential height, stability parameters). In addition to producing hourly analyses, the data fusion tool is also capable of providing relevant information regarding satellite data quality, such as the data age, coverage, and its convergence (i.e. the analysis fit to the satellite observations), to better inform the operational user about the analysis quality.

The availability of a tool that fuses the data from multiple satellite and conventional sources and provides consistent geophysical parameters that fit those observations, on a global scale, should prove beneficial for situational awareness and short-term forecast guidance. We will present overviews of the data fusion processing system and its components, as well as preliminary validation results through comparisons with numerical weather prediction analysis fields and ground truth. Additionally, we will present case studies from high impact weather events, to demonstrate the value-added benefits offered to operational forecasters for enhanced situational awareness and nowcasting and short-term forecasting capabilities.