Thursday, 16 January 2020: 8:30 AM
258B (Boston Convention and Exhibition Center)
Mesoscale modeling, particularly with respect to precipitation and atmospheric temperature and moisture forecasting, remains a challenge as physical parameterizations have inherent biases. Forecast accuracy is especially important in regions of the world where severe weather frequently occurs, such as North and South America. This study compares two WRF model runs, with 20 km horizontal resolution and identical physical parameterizations except for the cumulus scheme, to observational radiosonde data in North and South America during the 2016 and 2018/19 warm seasons, respectfully. Both have significant and similar forecast biases in both continents during the warm season: in general, errors are greater at longer forecast leads, over South America, and with Grell-Freitas cumulus as opposed to Kain-Fritsch. In particular, they feature a slight mid-level warm bias and a much larger mid-level dry bias in each continent on average. Analysis of parcel backward trajectories and meridional wind errors within the models suggests that this dry bias is likely caused by 1) a poor representation of the strength and depth of a low-level jet, and/or 2) an overestimate in subsidence from the high terrain in the west. While data is limited to the South American warm season of 2018/19, this dry bias shifts to a moist bias during North American cool seasons. There are likely many more factors are play which will be addressed in future work with new observational data from the recent RELAMPAGO field campaign in Argentina.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner