POWRE: Artificial Neural Networks for Spatial Aggregation and Disaggregation Problems

POWRE：用于空间聚合和分解问题的人工神经网络

• 批准号: 9973474
• 负责人: Sucharita Gopal
• 金额: $ 7.36万
• 依托单位: Trustees of Boston University
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 1999
• 资助国家: 美国
• 起止时间: 1999-08-15 至 2001-09-30
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=9973474&HistoricalAwards=false
• 关键词: POWRE; Artificial; Neural; Networks; Spatial

Advances in satellite remote sensing (RS) and geographic information systems (GIS) technologies are contributing to rapid expansion in the availability and use of spatial information about Earth's environment at local to global scales. They have also led to the use of new methods in spatial analysis based on computational intelligence (CI). Artificial Neural Networks (ANN) are one type of CI that have been widely used in geography and RS for classification, change detection, optimization and spatial modeling. The main objective of this research is to apply ANN for spatial scaling, particularly aggregation and disaggregation resulting from the use of multiscale and multiresolution data in GIS and RS. Three specific areas of geographical enquiry will be addressed: first, characterization of land cover at the global scale, including improved estimates of land cover area as well as changes in land cover area; second, developing models integrating socio-economic and physical data collected at different spatial scales (e.g., Mediterranean Desertification and Landuse or MEDALUS); and third, the population census zone design problem, in which data are aggregated from fine to coarser spatial scales. This type of spatial analysis and modeling has a number of practical applications including census geography, global change research, land surface characterization, strategic landuse and transportation planning, environmental analysis and planning, and allocation of education, health, and other services. This research, which will focus on analytical and methodological issues centered around ANN applications in geography and remote sensing, will involve collaboration between the principal investigator and two leading European researchers. The POWRE award will enhance the principal investigator's research and educational experience, as well as contribute to her academic advancement and contribute to her leadership potential in the field in which women are significantly underrepresented.

卫星遥感（RS）和地理信息系统（GIS）技术的进步有助于迅速扩大在地方到全球范围内提供和使用关于地球环境的空间信息。它们还导致了基于计算智能（CI）的空间分析新方法的使用。人工神经网络（Artificial Neural Networks， ANN）是一种广泛应用于地理和遥感的CI，用于分类、变化检测、优化和空间建模。本研究的主要目的是将人工神经网络应用于空间尺度，特别是在GIS和RS中使用多尺度和多分辨率数据所产生的聚集和分解。地理调查的三个具体领域将得到解决：第一，全球尺度上的土地覆盖特征，包括改进土地覆盖面积的估计以及土地覆盖面积的变化；第二，建立综合在不同空间尺度收集的社会经济和自然数据的模型（例如，地中海荒漠化和土地利用或MEDALUS）；第三，人口普查区设计问题，数据从细尺度到粗尺度进行汇总。这种类型的空间分析和建模有许多实际应用，包括人口普查地理学、全球变化研究、陆地表面特征、战略性土地利用和交通规划、环境分析和规划，以及教育、卫生和其他服务的分配。这项研究将集中于围绕人工神经网络在地理和遥感中的应用的分析和方法问题，将涉及首席研究员和两位领先的欧洲研究人员之间的合作。POWRE奖将提高首席研究员的研究和教育经验，并有助于她的学术进步，并有助于她在女性代表性明显不足的领域发挥领导潜力。