Handling Massive and Dynamic Terrain Data

Research output: Book/anthology/dissertation/reportPh.D. thesisResearch

Documents

  • Morten Revsbæk, Denmark
Recent technological advances have greatly increased our ability to collect and
store detailed information about the topography of the Earth, both below and
above the ocean. This information is often stored as digital terrain models.
These terrain models have a large range of applications from analyzing flood
risk and visibility, to producing nautical charts.
Many algorithms and much commercial software have been developed to
help analyze terrain models. However, significant algorithmic challenges arise
from the increasing detail (and therefore size) of modern terrain models. Furthermore,
terrain models are increasingly being updated as a result of new data
collection or editing done by users. This transforms a terrain model into a dynamic
object, and presents a set of new algorithmic challenges. In this thesis
we consider some of the above challenges.
To enable analysis of massive terrain data we develop so-called I/O-efficient
algorithms for a set of well-known terrain analysis problems. First, we present
an I/O-efficient algorithm for terrain model simplification. This algorithm can
be used in connection with terrain analysis to reduce the topological complexity
of a detailed terrain model before performing the actual analysis. Then we
present an I/O-efficient algorithm for extracting simplified yet precise contour
maps from detailed terrain models. Finally, we present an I/O-efficient algorithm
for estimating the flood risk from water collecting in basins of a terrain
model during a rain event
Considering the terrain as a dynamic object is a relatively new research
challenge. To meet this challenge, we consider the terrain as a continuously
deforming object and present a data structure that can maintain certain topological
attributes of the terrain as it deforms. These topological attributes are
central to a range of other attributes derived from the terrain.
v
Original languageEnglish
PublisherDepartment of Computer Science, Aarhus University
Number of pages126
Publication statusPublished - 2014

See relations at Aarhus University Citationformats

Download statistics

No data available

ID: 81761643