3.6 Python for Radiance Diagnostics in the Global Forecast Dropout Prediction Tool

Monday, 8 January 2018: 11:45 AM
Room 8 ABC (ACC) (Austin, Texas)
Andrew Eichmann, NOAA/NESDIS, College Park, MD; and K. Kumar and J. C. Alpert

The Global Forecast Dropout Prediction Tool (GFDPT) is a collaborative project of National Environmental Satellite, Data, and Information Service (NESDIS) and National Centers for Environmental Prediction (NCEP), Environmental Modeling Center (EMC) to provide operational forecasters with the means to predict and diagnose poor skill forecasts (aka dropouts) in the GFS with focus on assimilated satellite radiance data. Python was chosen for new development in the project due to its combination of flexible plotting capabilities, numerical computing libraries, and licensing that allows sharing, portability, and extendibility. The GFDPT is applied to find probable model forecast dropouts by correlating GFS forecasts with those of other models, particularly ECMWF. Then the analyses used as the model initial condition for the respective model forecasts are compared to find the largest volumetric differences in their physical fields, with the aim of examining their contribution to divergent forecasts. Python scripts are used to process the relevant GSI radiance diagnostics (using numpy and read_diag) and generate geographical plots highlighting anomalies in the radiance fields, and web pages to display them. Python scripts are also used to statistically analyze the radiance observations to determine which observations are to be masked for observation denial experiments with the aim of improving quality control of satellite radiance data.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner