Improve Surface Analysis for 3D-RTMA

A development project is underway for a three-dimensional Real-Time Mesoscale Analysis (3D-RTMA). This project is supported by the Department of Commerce/National Oceanic and Atmospheric Administration Joint Technology Transfer Initiative (JTTI). The goal of the 3D-RTMA is to extend the operational 2D-RTMA to three dimensions for “full atmosphere” situational awareness and for a 3-D analysis of record. Besides providing full-column representation of standard meteorological fields such as temperature, water vapor, and wind, as well as hydrometeors (i.e., clouds, precipitation of all forms), and eventually aerosols, the 3D-RTMA will also include land-surface diagnostics (e.g., soil moisture, snow state from multi-level land-surface fields), and convective (e.g., hail size, supercell rotation tracks) fields. The 3-D cloud hydrometeor analysis in 3D-RTMA has great potential to provide critical “nowcast” information, merging all available observations with a prior forecast background, for identifying hazards to aviation (e.g., in-flight icing, and ceiling and visibility restrictions) and ground transportation (e.g., highway snow and ice accumulation in complex terrain). With all these components, the NOAA 3D-RTMA will 1) improve tools for situational awareness and nowcasting, 2) provide a three-dimensional analysis of record (AOR) suitable for verification and bias-correction, and 3) accelerate improvement of numerical weather prediction (NWP) models. It will ultimately improve forecast guidance provided to the public by NWS forecast offices and national centers.

Experimental real-time 3D-RTMA products have been evaluated at the 2019 Spring Forecast Experiment in the Hazardous Weather Testbed in May 2019. Generally, the 3D-RTMA surface analysis was well-received by forecasters so far. This presentation will focus on efforts to improve the surface analysis component of the 3D-RTMA surface, including for 2-m temperature, 2-m dew point and 10-m wind fields. The impacts of observation error, length scale and hybrid ensemble weighting will be illustrated. The benefits of using convective-scale ensemble background error covariance vs. that from the global GDAS ensemble will also be demonstrated.