Cornell University has been granted an award to address a fundamental problem in population biology: How can we estimate true abundance of wild bird populations across North America? They approach this problem from three directions. First, they will use one of the largest and longest-running resources of environmental time-series data sets in existence by organizing the distributed resources of observation-based bird monitoring projects. Second, they will develop novel data analysis methods targeted for analyzing abundance of wild bird populations across North America. Their data mining approaches include ensemble learning, statistical smoothing, multi-task machine learning, and the estimation of change in abundance over time to quantify variation in spatio-temporal variation in abundance and to estimate the impact of environmental conditions. Third, they will make all data and analysis tools available online to allow browsing of bird-monitoring data. Fast response times are guaranteed through novel methods for approximate results to data-exploration queries. In addition to advancing research both in computer science and population biology, they will expose advances in high performance computing and data analysis to new audiences, from biologists, conservation agencies, and land planners to school classrooms and through their website to the literally millions of people who watch and appreciate wild birds. They will enable online interpretation of the results of their analyses and computations through web-based data visualizations and active dissemination and use of this information through collaborative education and conservation programs.
