The Continued Development of an Automated Observation Quality Control Package for 3D-RTMA

Work is underway at NOAA’s Environmental Modeling Center (EMC) and Global Systems Laboratory (GSL) to transition the operational 2DVar-based Real-Time Mesoscale Analysis (RTMA) to a hybrid 3DEnVar-based system (3D-RTMA). The first 3D-RTMA implementation is currently expected in late 2025.

Observation quality control remains a considerable challenge for the operational RTMA and future 3D-RTMA suites, as these systems are designed to provide a close fit to the observations. This presentation will provide an overview of the automated quality control (AutoQC) package that is being developed for the 3D-RTMA suite. The AutoQC package generates accept lists for surface fields (temperature, pressure, moisture, and wind), which are derived using standard deviation, bias, and stuck instrument checks; diurnal reject lists for temperature and moisture will also be added prior to implementation. These lists will largely replace the fixed lists used in the operational RTMA suite, allowing for poor-quality observations to more efficiently be removed from the analyses in real-time and minimize the need for manual QC in some circumstances. The AutoQC package will also include a bias correction algorithm for mesonet wind observations, structured as a Kalman-Bucy filter, in an attempt to address a long-standing low wind speed bias that is driven by the assimilation of mesonet wind observations. Although initial development efforts have focused on surface observations, aircraft observations will also be assimilated in 3D-RTMA, and the QC of these observations will be a necessary addition to the AutoQC package. Additionally, artificial intelligence and machine learning techniques for observation QC are beginning to be explored for inclusion in the AutoQC package.

This presentation will include results on the performance of the AutoQC package and mesonet wind bias correction algorithm using comparisons against control experiments that employ traditional QC techniques. The performance will also be assessed for recent cases in which the operational RTMA performed poorly based on feedback received from NWS forecast offices. In order to foster stakeholder engagement, which remains a critical aspect of development on the RTMA suites, a prototype version of 3D-RTMA that incorporates the AutoQC package will be shared with stakeholders in late 2024.