NSF Scientific Software Innovation Institutes to support environmental observatories

As we struggle with how to manage and curate increasing quantities of data, the NSF struggles with how to fund software development to support managing and curating increasing quantities of data. There are clearly not enough funds for each large research project to develop its own software system to support the complete data life-cycle. As part of its Cyberinfrastructure for the 21st Century Science and Engineering (CF21) vision, the National Science Foundation is exploring the creation of three to six Scientific Software Innovation Institutes (S2I2). I helped organize a workshop made up of participants from environmental observatories to give community feedback to this program. These environmental observatories span five NSF directorates: Geoscience [GEO]; Engineering [ENG]; Social, Behavioral, and Economic Sciences [SBE]; Biological Sciences [BIO]; and Polar Programs [OPP]. Vendor representatives were Red Hat, Microsoft Research, IBM, Environmental Systems Research Institute (ESRI), and Danish Hydraulic Institute (DHI). Other organizations represented included the National Center for Atmospheric Research (NCAR), Unidata, California Institute for Telecommunications and Information Technology (Calit2), San Diego Supercomputer Center (SDSC), National Snow and Ice Data Center (NSIDC), Stroud Water Research Center, and several universities where cyberinfrastructure research and development for environmental science is conducted. LTER PIs Scott Collins (SEV) and Corinna Gries (NTL) were participants as well as many faces familiar to LTER researchers. The uniqueness of this workshop among the many others held for this purpose by NSF was that it was probably the only one where a community of domain scientists were gathered to discuss the needs of the community vs the grand challenges of software engineering. In fact one of the major recommendations is that NSF recognize the importance of serving a role in expanding synthesis within and across these domain science communities and not just do major software development. The workshop report has been submitted to NSF and is now available here.