Der deutsche Education Blog

November, 2010

Microsoft Research Connections Blog

The Microsoft Research Connections blog shares stories of collaborations with computer scientists at academic and scientific institutions to advance technical innovations in computing, as well as related events, scholarships, and fellowships.

November, 2010

  • Microsoft Research Connections Blog

    MSR-FAPESP Workshop Focuses on Multidisciplinary, Data-Driven Environmental Science

    • 0 Comments

    On November 11 and 12, 2010, an international group of distinguished researchers met in São Paulo, Brazil, to expand ongoing environmental research on the Brazilian tropical rainforest ecosystem. The workshop built upon the Sensor Nets in Tropical Forests, a pilot project that deployed a sensor net in a 1 km2 area of forest in Serra do Mar, Ubatuba, southeast of Brazil. The workshop was sponsored by Microsoft Research-FAPESP Institute for IT Research, a joint venture of the External Research division of Microsoft Research and FAPESP (the São Paulo Research Foundation).

    Dr. Carlos NobreThe gathering brought together scientists from the United States, Chile, Canada, and Brazil, representing such diverse disciplines as phenology, climate science, soil ecology, successional forestry, micrometeorology, and earth system modeling. Among them was Dr. Carlos Nobre, one of the world's foremost researchers on climate change, whose talk, "The Fate of the Amazon Forest in the 21st Century," discussed the impact of global warming, climate change, and fires to the balance of the Amazon forest.

    The distinguished attendees identified and tied together a multidisciplinary set of research problems with technology development and collaboration requirements, creating an experiment plan that broadly charts several years of environmental experiments and technology development. In the course of developing the plan, the researchers spent considerable time reflecting on how technology can support them in the ongoing research, pondering the implications of using new methods and technology support to conduct their science. There was broad consensus that environmental science is faced with enormous data-driven challenges-from data acquisition to data management, cleaning, provenance, annotation, visualization, sharing, publication, discovery, and archival. These challenges are accompanied by acceleration in problem complexity and the need for new, integrated multidisciplinary approaches to research.

    In response to these important and daunting trends, it is essential to develop new technologies and particularly new computational solutions. In the collaborative rainforest project, we find just such interdisciplinary and data complexity problems, and we see this as a tremendous opportunity to solve specific problems by flexible, extensible, and adoptable means. By so doing, we will build solutions with applicability beyond their immediate corresponding problems.

    —Juliana Salles, senior research program manager in the External Research division of Microsoft Research

  • Microsoft Research Connections Blog

    Climate Prediction Leverages Home Computers

    • 0 Comments

    A mind is a terrible thing to waste, and so, apparently, is an idle computer. On November 18, scientists from Oregon State University (OSU) will begin harnessing the idle time of thousands of personal computers to study climate modeling and the resulting weather effects on the Pacific Northwest and California. This project illustrates the power of "citizen scientists," in this case, the thousands of individuals whose personal computers will be leveraged during idle times. By lashing together the computing power of these privately owned PCs, the OSU team will create a digital network with the computational chops of a supercomputer.

    The OSU participation in this worldwide initiative, under the leadership of Philip Mote, director of the Oregon Climate Change Research Institute, will study the reliability of various climate simulation models in the western United States. Microsoft Research provided primary funding for this initiative, with additional support coming from the Bureau of Land Management, the California Energy Commission, and the U.S. Geological Survey.

    The work at OSU will join that already underway in South Africa and the United Kingdom. Since the project's inception in 2003, hundreds of thousands of citizen scientists have donated their spare computer processing power to run global climate change models. During idle times, the personal computers collect data on regional climate information, including temperature, winds, and humidity, which is then compiled and fed into climate model simulations.

    Experiments in the OSU-led initiative include predicting regional weather events-including droughts, floods, and temperature extremes-during the 2020s and 2030s, and forecasting the possible climate impact if global temperatures rise two, three, or four degrees by the end of the century.

    Surface temperature in the global model - you can watch this progressing as your model runs.

    Surface temperature in the global model - you can watch this progressing as your model runs.

    Mote notes that the rationale behind these climate modeling studies is elegantly simple. "In less than two months, we can run 40,000 different year-long climate simulation models with our network of volunteers. A dedicated supercomputer, during that time, could simulate a couple hundred years worth of data."

    "It's exciting that both climate modeling and computer technology have advanced to the point that people at home can contribute to the effort to study climate change," he added.

    Individuals in the western United States who wish to donate their computer's spare processing power are encouraged to visit www.climateprediction.net/weatherathome, where they can register to have the climate model application downloaded. Much like a screensaver, it will run only when your computer is idle.

    Dan Fay, director of the External Research division of Microsoft Research

Page 2 of 2 (5 items) 12