You are here:
Adapting MPAS FDDA to use analysis nudging with 3-hour updates
Citation:
Bullock, Orren. Adapting MPAS FDDA to use analysis nudging with 3-hour updates. Joint WRF and MPAS Users' Workshop 2019, Boulder, CO, June 10 - 14, 2019.
Impact/Purpose:
This work improves meteorological simulations of the Model for Prediction Across Scales (MPAS) by shortening the update period for Four-Dimensional Data Assimilation (FDDA) guidance from 6 hours to 3 hours. MPAS will be used in the Next Generation air-quality modeling system. Any of EPA's program offices, regional offices, and public partners performing air-quality assessment would be interested.
Description:
Four-dimensional data assimilation (FDDA) was previously implemented in the MPAS - Atmosphere model using an analysis nudging technique with guidance based on the National Center for Environmental Prediction (NCEP) FNL (Final) meteorological analyses available every 6 hours. This frequency is generally considered the minimum necessary to resolve diurnal meteorological cycles. It stands to reason that improved MPAS model accuracy could be achieved with more frequent updates to FDDA guidance. Besides the FNL product, which considers all information that is eventually fed into NCEP's Meteorological Analysis and Data Ingest System (MADIS), initialization fields for the Global Forecast System (GFS) and 3-hour simulation results are also available. GFS initialization fields are produced every 6 hours just like FNL fields, but they are subject to model simulation scheduling and are rarely developed from a complete data ingest. Nonetheless, combined with 3-hour simulation results, these GFS analyses offer the ability to update FDDA guidance every 3 hours. MPAS has been applied with 3-hour FDDA updates facilitated two ways; 1) using GFS initial and 3-hour forecast fields, and 2) using FNL analyses and GFS 3-hour forecast fields. These MPAS simulations are evaluated against surface and upper air observations and compared to results previously obtained with 6-hour FDDA updates. The results show improved accuracy using GFS initial and 3-hour forecast fields, and best accuracy using a blend of FNL analyses and 3-hour GFS forecast fields.
