During the past few decades, the academic meteorology enterprise has supported a national collegiate forecast contest that seeks to engage mostly undergraduate students with some graduate students and faculty in practical forecasting under a variety of geographical and phenomenological circumstances. Known today as Weather Challenge (WxChallenge) and sponsored by the University of Oklahoma, each individual participant forecasts the maximum and minimum temperature, precipitation amount, and maximum sustained wind speed for select North American cities. WxChallenge provides students an opportunity to compete against their peers and faculty mentors at other institutions (64 nationwide in 2006-2007) for honors as the top weather forecaster in the nation.

In spring 2007 the NSF-funded ITR project, Linked Environments for Atmospheric Discovery (LEAD) engaged a manageable subset of the WxChallenge community and provided access to the LEAD Gateway portal and its underlying services. The goal of this so-called “LEAD-WxChallenge Pilot Project” was to provide students with the ability to generate, run, analyze, and visualize their own WRF forecasts using the web-enabled services developed by LEAD researchers. For seven weeks, 75 students and faculty from 10 institutions were invited to join the pilot project. For the participants it provided an unprecedented opportunity for enhancing their understanding of numerical modeling, high performance computing, physical parameterization schemes, data assimilation, and workflow orchestration. It also demonstrated a key motivation for LEAD: to democratize and empower students by lowering the barrier for access to complex, integrated services, thereby allowing users the freedom to select inputs such as initialization fields, set model domains, and run WRF at a time and location determined by the user, and not constrained as in static, fixed and prescribed operational environments. In return, LEAD developers benefited from user feedback that exposed strengths and weaknesses, and provided a better sense of the challenges and resource requirements associated with maintaining a reliable and persistent system aimed at enabling a larger community.

Participants were given authorization to the LEAD Gateway portal to access data, build experiments, and compose workflows with sufficient computing resources to run WRF and save the WRF output in myLEAD workspace. LEAD also provided tools for visualizing the output and user support. Upon completion of a 42-hour, high-resolution WRF run, students integrated user-generated output products into their personal schema for preparing a forecast for stations previously selected by WxChallenge.

Over the seven week pilot project participants launched a total of 279 forecast workflows and generated 0.6 TBytes of data. Over 160 processors were reserved on a multi-processor server five days each week from 10am to 8pm EDT. For the NAM initialized WRF forecast, 78 percent of the workflows submitted were successful, 22 percent failed. The ADAS initialed WRF forecast was less successful with 36 percent successful and 64 percent failing. When the WxChallenge ended and the pilot project concluded, the participants were asked to complete a survey, the results of which will be used to refine the production release and prepare for an expanded release to a much larger population of WxChallenge participants in fall 2007.

This paper will present an overview of the LEAD-WxChallenge pilot project along with a more detailed description of the emergent problems and challenges. In addition, preliminary results of the fall 2007 LEAD-WxChallenge project will be reported.