CDI-Type I: Data Rods: Enabling Time-Series Analysis of Massive Multi-Modality Cryospheric Data

点击翻译按钮获取中文摘要

This award is funded under the American Recovery and Reinvestment Act of 2009 (Public Law 111-5). The goal of most climate research is to detect change over time. Much of the data are in the form of spatiallyarranged remote sensing data or model output. These data sets are voluminous and tend to be organized as separate geo-referenced files for a single time, which is a function of how the data were collected. Subsetting these data files for change detection over select regions a very common goal of climate data analysis is time consuming. The goal of this project is to design and prototype a process for addressing time-series data as pure objects that will enable time-centric change analysis of massive multi-modality cryospheric data. This project will create ?Data Rods? as a spatiotemporal data construct in a pure object database. Rather than storing data at a fixed time and a variable spatial extent, this project will store the data at fixed spatial extents with a variable time component, thus the data will be optimized for space-time queries and searches. A Data Rod is a logical object that integrates all the information that is known about a point on the earth (the pixel/grid cell) through time. The introduction of time to the discrete pixel allows changes over time to be represented by items in the database, as opposed to timestamped spatial grids. The result is a very efficient storage and representation system. Queries to this database can span millions of data rods simultaneously across both time and space. This project will also develop time-series change analysis techniques, which exploit the spatiotemporal aspects of the Data Rods to efficiently query for features, patterns and anomalies directly and quickly. The initial focus will be on gridded data maintained by the National Snow and Ice Data Center (NSIDC), but it will be extended to other remote sensing and point data. The initial science focus will be to evaluate the contribution of the Greenland ice sheet to sea level rise. The approach will be demonstrated by using real data for Greenland to answer questions such as: At what time interval did the albedo over Greenland change the fastest? Or, was 2007 an anomalous melt year? What local factors contributed to this anomaly? These questions involve huge data sets and statistical queries that are poorly served by existing tools. The Data Rod construct and new search algorithms will be tested for feasibility, speed and performance of this technology to solve critical space-time questions. In terms of broader impacts, the set of technologies developed in this project could transform how science researchers access and analysis massive volumes of data in many scientific domains such as life and social science working with spatiotemporal data. The expected science results and techniques will be disseminated through journal articles, standards bodies, cyberlearning resources, curriculum design, and outreach activities to (future) IT and science professionals including underrepresented groups. Students will be trained in a cyber-intensive, interdisciplinary setting. CDI themes: ?From Data to Knowledge?, ?Understanding Complexity.?