Numerical Weather Prediction on AWS Graviton2 with WRF provides a complete package to launch and manage WRF on AWS Graviton2. WRF is a mesoscopic model used by thousands of users worldwide, and able to generate forecasts for a wide range of atmospheric conditions and resolutions. These forecasts serve for operational functions and research purposes.
The AMI includes an integration of WRF with several tools for data processing and cluster performance & cost monitoring. It has dual functionality as it can be used either on single instances or by clusters launched with AWS ParallelCluster. The AMI also works with spot prices (usual conditions apply) for instances and cluster compute nodes, so substantial savings are available.
The package includes the following features:
-Implementation of WRF v4.3 optimized for AWS Graviton2 instances and clusters providing an optimal balance between cost and performance.
-Pre-compiled executables for ARW (standard nesting, nesting with a prescribed set of moves & nesting following a vortex), NMM and WRF-Chem.
-Excellent scalability for clusters with Elastic Fabric Adapter (EFA).
-Availability of NICE-DCV connection for high-performance remote display.
-Direct connection to the open data Community Earth System Model Large Ensemble and other meteorological datasets.
-Postprocessing tools include NCL, wrf-python and NCEP Unified Post Processor (UPP).
-Tools for monitoring cluster performance & cost via Grafana dashboards.