Abstract
This is a real-data Four Dimensional Data Assimilation (FDDA) study using MM5 in conjunction with West Texas Mesonet surface observations and ACARS (Aircraft Communications Addressing and Reporting System) profile data collected by commercial aircraft during both en route and ascent/descent phases of their flights. The high-frequency mesonet data and ACARS wind and temperature profiles are ideal for testing the effects of FDDA on short-term mesoscale numerical weather prediction. The mesonet experiments involved 35 sites with an average horizontal spacing of about 30 km, while in the ACARS case ninety five profiles were used. Results indicated that nudging the MM5 model with the surface-based data over the relatively small area of the mesonet domain had limited impact on the model’s performance. In the ACARS runs, FDDA had long-lasting impact throughout the entire model atmosphere. FDDA appeared to improve the quantitative precipitation forecasting skill of MM5 and reduce slightly the model’s warm bias at the surface. The study suggests that ACARS has potential to significantly enhance our expertise in short-term mesoscale modeling and to support the need to rapidly and accurately adjust high-resolution meteorological model forecasts to real-time observations.
Highlights
The focus of this paper is on mesoscale modeling and Four Dimensional Data Assimilation (FDDA)
The West Texas Mesonet (WTM) surface observations reported at five minute frequency, along with the NOAA ACARS observations taken by commercial aircraft at about every ten minutes while crisscrossing the nation [2], offer valuable data for mesoscale FDDA
The work was carried out based on MM5 real-data simulation experiments in conjunction with WTM surface data and ACARS profile data
Summary
The focus of this paper is on mesoscale modeling and FDDA. Because of its small spatial size mesoscale weather is strongly influenced by fast-changing local conditions such as cloud, friction, and surface heating. The MM5 simulations included a 24-h control run (CNTR) without invoking the WTM data along with a similar 24-h FDDA run with observational nudging between 6 h and 12 h model time.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have