Postdoctoral Research Fellowships in Biology for FY 2009

该行动根据2009年美国复苏和再投资法案（公法111-5）为2009财年的NSF生物学博士后研究奖学金提供资金。 该研究金支持丹尼尔谢尔顿的一项研究和培训计划，题为“改变物种分布的算法：跟踪世界上的树燕”。这项研究的主办机构是俄勒冈州州立大学，赞助科学家是托马斯迪特里希。了解物种的地理分布以及它们如何变化对保护，生态科学和全球气候变化研究至关重要。这项研究开发了计算方法来量化树燕（Tachycineta bicolor）分布的变化模式，这是一种迁徙的鸣禽，在其为期八个月的非繁殖季节形成密集的公共栖息地。清晨离开栖息地的鸟类创造了一个独特的“栖息环”模式，这是在NEXRAD天气雷达图像可见。该项目采用计算机视觉、机器学习和概率建模等技术，从雷达中自动检测和提取栖息环，收集美国各地树燕分布的每日估计值，然后推断迁移模式和其他事件，这些事件会导致分布随时间发生变化，包括短期和长期变化。NEXRAD雷达最大的优点之一是能够利用存档数据查看过去15年的情况。本项目旨在解决燕子栖息生物学和种群趋势的重要问题，推进雷达鸟类学技术，并为识别物种分布变化模式奠定算法基础。培训目标包括加强计算机视觉，雷达鸟类学和物种分布建模的技能。这项研究吸引了生物学家、计算机科学家、自然资源保护主义者和公众的广泛参与，提供了一个完整而准确的大规模鸣禽分布模型，并提供了持续监测过去15年和未来变化的规定。

This action funds an NSF Postdoctoral Research Fellowship in Biology for FY 2009 under the American Recovery and Reinvestment Act of 2009 (Public Law 111-5). The fellowship supports a research and training plan entitled "Algorithms for changing species distributions: tracking the world's tree swallows" for Daniel Sheldon. The host institution for this research is Oregon State University, and the sponsoring scientist is Thomas Dietterich.Understanding the geographical distributions of species and how they change is critical to conservation, ecological science, and global climate change research. This research develops computational methods to quantify patterns of change in the distribution of the Tree Swallow (Tachycineta bicolor), a migratory songbird that forms dense communal roosts during its eight-month non-breeding season. The early-morning departure of birds from the roost creates a distinctive "roost ring" pattern that is visible in NEXRAD weather radar images. The project employs techniques from computer vision, machine learning and probabilistic modeling to automatically detect and extract roost rings from radar, assembles daily estimates of the distribution of Tree Swallows throughout the United States, and then infers the patterns of migration and other events that cause distribution changes over time, both short-term and long-term. One of the great assets of NEXRAD radar is the ability to look fifteen years into the past with archived data. This project addresses significant questions about the roosting biology and population trends of swallows, advances techniques in radar ornithology, and lays an algorithmic foundation for identifying patterns of change in species distributions.Training objectives include strengthening skills in computer vision, radar ornithology, and species distribution modeling. This research engages a broad audience of biologists, computer scientists, conservationists, and the general public by providing a complete and accurate model of songbird distribution on a large scale, with provisions to continuously monitor changes for the past fifteen years and into the future.