摘要
讨论了传统MODIS数据处理方法中的局限性,提出使用MapReduce计算模型来进行MODIS数据的多节点并行处理的方法,并基于Hadoop系统进行具体实现。该方法把数据处理逻辑和数据存储作为统一的整体部署在各个计算节点上,减少了存储节点到运算节点的数据同步时间,提高了处理效率。对基于时间序列和地理空间分布的方式划分的MODIS数据,有着普遍的适用性。
We discuss the limitation of traditional MODIS data processing, and present a new method which uses the MapReduce computation model to carry out multiple computing nodes parallel processing of the MODIS data, and the concrete implementation is achieved based on Hadoop system. This method deploys the data processing logic code and the stored data themselves onto every computing node as a unified entirety, this reduces the data synchronising time between the storage nodes and the computing nodes, and improves the processing efficiency as well. The method has universal availability to MODIS data divided in time series-based and geographical space-distributed modes.
出处
《计算机应用与软件》
CSCD
北大核心
2013年第7期262-264,289,共4页
Computer Applications and Software
