Lessons learned from implementing science algorithms into an operational production system

Science teams for space-based remote sensing instruments are increasingly faced with the need to produce large data sets by processing high volume data streams in an operational mode. Experience with several instruments and spacecraft has shown that both scientists and program managers underestimate the time, effort and money required to go from a scientific algorithm concept to an implemented production system. This has a serious impact on the plans and budgets for individual scientists, instrument teams, and spacecraft projects. This talk is based on 8 years of experience. We discuss some of the problems encountered by the MODIS Science Data Support Team. We describe some approaches that may help future managers and science teams faced with similar tasks. The focus is on activities that occur after development of the algorithmic concept used to convert instrument measurements into geophysical quantities and before application of resulting geophysical data in scientific investigations.