Implementation and Research on the Operational Use of the Mesoscale Prediction Model COAMPS in Poland

摘要

Our long-term goal is to implement an operational high-resolution atmospheric data assimilation and prediction system and to use it for daily weather forecasting. In this project, we have worked on several operational and scientific aspects of the problem using the Coupled Ocean/Atmosphere Mesoscale Prediction System (COAMPS): (a) setting up stable large-scale data transfer capability to facilitate COAMPS runs 2-4 times per day in an operational manner for different geographical regions and using different nested grid configurations; (b) perform research on the MPI scalability of COAMPS on selected computer architectures and to optimize the code to take advantage of the vector capabilities of the Cray X1 and the massively parallel features of our Linux cluster; (c) identify and understand the uncertainties in high- resolution NWP forecast and their impact on severe weather, such as extreme rainfall, and to develop model metrics appropriate to mesoscale weather phenomena; and (d) improve our knowledge of observational error characteristics for spatially-correlated data and develop the numerical schemes capable of assimilating these types of observations.