Visitor Information: The Visitors Center and entrance to the Boulder Department of Commerce facilities are located on Broadway at Rayleigh Road. All visiting seminar attendees, including pedestrians and bike riders, are required to check in at the Visitors Center at the Security Checkpoint to receive a visitor badge. Seminar attendees need to present a valid photo ID and mention the seminar title or the speaker's name to obtain a visitor badge. .
Upcoming Seminars
| Title: | Unraveling observation indicators (SURFRAD, aircraft, radiosonde) to identify NWP model moisture biases and remedies
|
| Speaker: |
Stan Benjamin
Stan Benjamin is a senior research associate at CIRES affiliated with NOAA GSL. He has an extensive background in data assimilation and earth-system model development. He received a PhD from Penn State University and has worked primarily in NOAA Research. He has led important NWP development efforts over the years working with key colleagues at GSL, NWS, NCAR, other labs and universities and continues to contribute to ongoing efforts.
|
|---|
| Date/Time: |
Thursday, February 19, 2026 01:30 PM MST (-0700)
   |
| Location: |
David Skaggs Research Center, Room GC402
Google Meet |
Abstract
Accurate representation of clouds, precipitation, and convective storms within numerical weather prediction (NWP) forecast models depends strongly on evolution of moisture especially in the atmospheric boundary layer. Moisture errors in the model can be introduced in the data assimilation initialization process or in errors in representing physical processes. Moisture validation is critical for NWP model evaluation.
Evaluating current NOAA/NWS operational NWP models (HRRR and RAP) over the US using radiosonde (raob) relative humidity (RH) observations has suggested that these models have a moist bias in the lowest 200-400 hPa. However, a recent study using SURFRAD observations confirmed widespread excessive downward shortwave radiation in HRRR and RAP related to insufficient clouds and an apparent dry bias, an opposite signal. Moreover, evaluating those same models using aircraft (AMDAR WVSS-II) RH observations now available over the last 10 years agree with the signal from SURFRAD, that the models have a dry bias. In a new 2-year collocation study, a dry bias of raobs (compared to AMDAR) is found to be 4% RH overall and 8-10% nearing saturation. Raobs show saturated conditions five times less frequently than AMDAR.
The larger multi-year detective investigation using different observation types for model evaluation will be described. The new assessments using SURFRAD and AMDAR are consistent in indicating that raob RH obs have given a misleading signal for model development. Moreover, assimilation of raob RH data has itself added to the dry bias of models. A summary of causes for dry biases in NOAA hourly updated models will be presented. Recommendations are made for RH model verification and for data assimilation of RH observations. Implications of this raob dry-bias finding are described including for research and climate assessment.
|
Archives
