A Framework for the Abstraction of Mesoscale Modeling for Weather Simulation

Widely disseminated weather forecast results (e. g. from various national centers and private companies) are useful for typical users in gauging future atmospheric disturbances. However, these canonical forecasts may not adequately meet the needs of end-users in the various scientific fields since a predetermined model, as structured by the model administrator, produces these forecasts. To perform his/her own successful forecasts, a user faces a steep learning curve involving the collection of initial condition data (e.g. radar, satellite, and reanalyses) and operation of a suitable model (and associated software/computing). In this project, we develop an intermediate (prototypical) software framework and a web-based front-end interface that allow for the abstraction of an advanced weather model upon which the end-user can perform customizable forecasts and analyses. Having such an accessible, front-end interface for a weather model can benefit educational programs at the secondary school and undergraduate level, scientific research in the fields like fluid dynamics and meteorology, and the general public. In all cases, our project allows the user to generate a localized domain of choice, run the desired forecast on a remote high-performance computer cluster, and visually see the results. For instance, an undergraduate science curriculum could incorporate the resulting weather forecast performed under this project in laboratory exercises. Scientific researchers and graduate students would be able to readily adjust key prognostic variables in the simulation within this project’s framework. The general public within the contiguous United States could also run a simplified version of the project’s software with adjustments in forecast clarity (spatial resolution) and region size (domain). Special cases of general interests, in which a detailed forecast may be required, would be over areas of possible strong weather activities.