WRF experiments usually involve more than just a single simulation. Hindcast, multi-parametric or long (climate) experiments involve a number of model runs with dependencies. These experiments are usually managed by the user in an ad-hoc manner, with specific scripts that account for the workflow required. On the other hand, the user has usually access to multiple computer resources and a given experiment is manually sent to one of them. WRF4G is a wrapper of the WRF workflow that simplifies the design of experiments and distributes them across the computational resources at hand. As a large-scale example of access to multiple sites, Grid computing enables the use of geographically distributed computer resources in a transparent way. WRF4G can send the experiments to Grid resources. Moreover, interrupted simulations are restarted in an unattended way until the experiment is finished.
Tweets by SantanderMeteo