亿万文档 免费下载

侵权投诉# 1.3.2_abstract_Parallel Flow-Accumulation

### Word文档免费下载：1.3.2_abstract_Parallel Flow-Accumulation

（下载1-4页，共4页）

Parallel Flow-Accumulation Calculation for Hydrology Analysis in Large Digital

Elevation Models

Ling Jing1, Guo-An Tang1,2*, Xiaodong Song1

Key Laboratory of Virtual Geographic Environment, Ministry of Education,

Nanjing Normal University, Nanjing, Jiangsu, 210046, China

2

Hunan State Engineering Laboratory of Geospatial Information, Xiangtan, 411201, China

*Corresponding author, e-mail: tangguoan@http://www.wendangxiazai.com

Keywords: Parallelism; Flow-Accumulation; Large Datasets; DEM;

1

is based on a model for representation of flow across the surface and between grid cells. The flow model follows the procedure [4] of (1) filling sinks, (2) computing flow direction, and (3) computing the flow accumulation to each grid cell. With the increase of scope and resolution of DEMs, calculation flow accumulation on a single machine requires too much memory and often results in computer thrashing – excessive swapping of data between memory and the hard disk – resulting in unacceptably slow performance[5-8]. A solution

Objective:

Digital elevation model (DEM for short) is data structure representing rectangular grids of terrain data composed of cells arranged as a raster, where each cell is composed of a floating point value equivalent to the elevation of that geographic point above some base value [1]. It has been widely used for surface hydrology analysis including the automatic extraction of catchment areas [2] or drainage network [3]. All these

computations are linked to the calculation of flow accumulation. But its computational efficiency is very low for large datasets. This paper we propose a method able to compute in parallel the global flow accumulation from a large DEM.

Background:

is to parallelize the computation. This paper

presents a parallel algorithm for calculating flow accumulation. Data:

As remote sensing precisions and

accuracies have improved, the resolutions of DEMs have gone from 30-100 meter to 1-5 mete. Moreover, there is lower than 1 meter. Thus, DEMs have been becoming larger and larger. As a result, many of the geo-science analysis techniques for lower resolutions and smaller DEMs become prohibitively time consuming when being applied to high resolution data.

Hydrologic information derived from DEMs

The data set of Mount Lu is used for the experiments. It consists of a much larger 4903 cells by 6469 cells grid and has 5m resolution. The time taken to complete the task was measured using the serial algorithm for each data set.

Methodology:

For this paper, the processes of filling sinks and computing flow direction (the D8 method) are assumed to have already been completed. Assumed to be a constant of 1 each cell in the DEM, the formula of calculating flow

accumulation is this:

(1)

## 我要评论