适用于Hadoop云平台的洪水淹没分析并行算法

数字高程模型(Digital Elevation Model, DEM)特殊的数据结构有利于数据在计算机中进行计算,但其数据量大,受计算机性能的限制,无法满足大规模洪水淹没分析的需要。针对单机计算能力有限以及有源淹没分析递归算法占用计算机资源较多的问题,提出了一种利用Hadoop云平台进行洪水淹没分析的并行算法。该算法对研究区域DEM按规则格网的思想进行划分,将大规模的研究区分割成多个规则的小区域。在指定水位下,对每个小区域进行数据压缩,计算每个栅格单元的淹没值,并将同一行中相连的淹没栅格序列记为淹没块,每个淹没块包含该栅格序列所属的区块号、行号、起始列号、终止列号以及淹没标记;当不同行或不同区域中的淹没块相互连通时,需对其淹没标记进行一致性设计。最终,根据淹没点所在淹没块的淹没标记提取出淹没点所在的淹没区,即为实际的淹没范围。实验结果表明,在指定淹没水位下,通过该算法可以快速提取淹没范围以及实现淹没水深的计算。

A parallel algorithm of flood submergence analysis for Hadoop cloud platform

The special data structure of Digital Elevation Model (DEM) is beneficial to the calculation of data, but due to the large data pool and limitation of the computer performance, it fails to meet the needs of large-scale flood inundation analysis. A parallel algorithm for flood inundation analysis by Hadoop cloud platform is proposed to solve the problems including limited computing power of a single computer, and active submergence analysis recursion algorithm taking up too much computer resources. The algorithm divides the research area DEM according to the rule grid and further divides the large-scale research area into several regular small areas. At the specified water level, the data of each cell is compressed, the submergence value of each grid unit is calculated, and the submerged grid sequence connected in the same row is recorded as a submerged block. Each submerged block contains the block number, line number, starting column number, stop column number and inundation mark of the raster sequence. When the submerged blocks in different regions or regions that are interconnected, consistency design of their drowning signs is carried out. Finally, according to the inundation mark of the submerged block, the inundated area of the submerged point is extracted, which is the actual inundated area. Experiments show that under the specified submerged water level, the algorithm can quickly extracts the submerged area and achieves the submerged depth.