QLC: An Automated Forecast Verification Suite for CAMS and AI-Integrated Weather Prediction Systems

The Quick Look Content (QLC) suite provides automated forecast verification and analysis capabilities optimized for the Copernicus Atmosphere Monitoring Service (CAMS) and emerging AI-integrated forecasting systems. QLC addresses the growing need for systematic, reproducible evaluation of atmospheric composition forecasts through an end-to-end workflow from data retrieval to publication-quality visualizations.

The system integrates direct access to ECMWF's MARS archive with currently 16 observation networks including EBAS, AirNow, and GHOST-harmonized datasets, covering 7,855 atmospheric variables. Native GRIB support preserves forecast step information critical for analyzing temporal forecast evolution. QLC handles the complete verification workflow: automated MARS data retrieval, model-observation collocation with configurable spatial and temporal matching, statistical analysis including bias, RMSE, and correlation metrics, and generation of comprehensive visualizations (e.g., maps, time series, scatter plots, Taylor diagrams).

Recent development of an AIFS-specific workflow enables systematic comparison of AI-integrated forecasts against traditional IFS-COMPO runs and observational data. QLC supports multiple evaluation modes: single experiment validation, multi-experiment intercomparison, and observation-only analysis. Processing scales from single-station quick looks to continental-scale multi-variable assessments. Integration with the evaltools package (CNRM/Météo-France) provides advanced statistical diagnostics including diurnal cycles, station score maps, and exceedance analysis.

Here we introduce QLC and demonstrate its capabilities through verification examples comparing IFS-COMPO and AIFS-COMPO forecasts for ozone, nitrogen oxides, and particulate matter against multi-network observations. Results highlight the tool's utility for operational forecast monitoring, model development support, and scientific analysis. The open-source package (PyPI: rc-qlc) is designed for use on both HPC systems and workstations, with one-command installation and comprehensive documentation at docs.researchconcepts.io/qlc.

This work is supported by CAMS2_35_bis_KNMI: "Developments on reactive gases and aerosol in the global system" (https://atmosphere.copernicus.eu/).