From the advent of the first hydrostatic mesoscale models in the early 1970s through the follow-on 3D non-hydrostatic cloud-resolving models that emerged in the late 1970s to today's Weather Research and Forecast (WRF) model, codes for mesoscale meteorology research have been developed and applied largely as stand-alone components. For example, early hurricane models were self-contained and initialized using idealized vortices and simple background atmospheric analyses. Likewise, early 3D cloud models – containing dynamics, turbulence and simple cloud physics parameterizations, utilized single soundings to define a horizontally homogeneous environment that was perturbed by a warm impulse to initiate convection. Although models have grown considerably in sophistication during the past three decades, and although various frameworks have been developed to couple model components or multiple models (e.g., Earth System Modeling Framework, or ESMF), less attention has been directed toward environments in which data feeds and repositories, data assimilation systems, models, and output post-processing and mining engines can easily be linked. Indeed, infrastructures in academia for doing so are quite limited, leading many researchers to operate models in overly simplified ways and forsake the use of special or new data sets.

Linked Environments for Atmospheric Discovery (LEAD), a 5-year NSF Large Information Technology Research grant, as created an integrated web service architecture, enabled for use in grid architectures, to support mesoscale meteorological data acquisition, analysis, assimilation, simulation modeling, prediction, mining and visualization. In this presentation we describe how LEAD has created a new paradigm for mesoscale meteorological research, the capabilities LEAD makes available to the broad community, the use of LEAD in realtime operational forecast experiments, and plans for LEAD now that it's five year run as an NSF grant is ending.