High Performance Computing for Geospatial Applications: A Retrospective View

TL;DR

This paper reviews the evolution of high performance computing in geospatial applications, highlighting past limitations and recent advances with distributed and cloud-based systems to handle complex, large-scale geospatial data.

Contribution

It provides a comprehensive retrospective analysis of HPC approaches in geospatial analysis, emphasizing the shift towards distributed and cloud computing solutions.

Findings

Early HPC systems were hardware-specific and not portable.

Recent approaches leverage distributed, cyberinfrastructure, cloud, and fog computing.

These advances improve scalability and efficiency in geospatial analysis.

Abstract

Many types of geospatial analyses are computationally complex, involving, for example, solution processes that require numerous iterations or combinatorial comparisons. This complexity has motivated the application of high performance computing (HPC) to a variety of geospatial problems. In many instances, HPC assumes even greater importance because complexity interacts with rapidly growing volumes of geospatial information to further impede analysis and display. This chapter briefly reviews the underlying need for HPC in geospatial applications and describes different approaches to past implementations. Many of these applications were developed using hardware systems that had a relatively short life-span and were implemented in software that was not easily portable. More promising recent approaches have turned to the use of distributed resources that includes cyberinfrastructure as well as cloud and fog computing.