An ensemble approach to space-time interpolation

Elizabeth Wentz, Donna J. Peuquet, Sharolyn Anderson

Research output: Contribution to journalArticle

9 Citations (Scopus)

Abstract

The availability of spatial data on an unprecedented scale as well as advancements in analytical and visualization techniques gives researchers the opportunity to study complex problems over large urban and regional areas. Nevertheless, few individual data sets exist that provide both the requisite spatial and/or temporal observational frequency to truly facilitate detailed investigations. Some data are collected frequently over time but only at a few geographic locations (e.g., weather stations). Similarly, other data are collected with a high level of spatial resolution but not at regular or frequent time intervals (e.g., satellite data). The purpose of this article is to present an interpolation approach that leverages the relative temporal richness of one data set with the relative spatial richness of another to fill in the gaps. Because different interpolation techniques are more appropriate than others for specific types of data, we propose a space-time interpolation approach whereby two interpolation methods - one for the temporal and one for the spatial dimension - are used in tandem to increase the accuracy results. We call our ensemble approach the space-time interpolation environment (STIE). The primary steps within this environment include a spatial interpolation processor, a temporal interpolation processor, and a calibration processor, which enforces phenomenonrelated behavioral constraints. The specific interpolation techniques used within the STIE can be chosen on the basis of suitability for the data and application at hand. In this article, we first describe STIE conceptually including the data input requirements, output structure, details of the primary steps, and the mechanism for coordinating the data within those steps. We then describe a case study focusing on urban land cover in Phoenix, Arizona, using our working implementation. Our empirical results show that our approach increased the accuracy for estimating urban land cover better than a single interpolation technique.

Original languageEnglish (US)
Pages (from-to)1309-1325
Number of pages17
JournalInternational Journal of Geographical Information Science
Volume24
Issue number9
DOIs
StatePublished - 2010

Fingerprint

interpolation
Interpolation
land cover
time
weather station
visualization
spatial data
satellite data
spatial resolution
Visualization
Availability
Satellites
Calibration
calibration

Keywords

  • Arizona
  • Phoenix
  • Space-time interpolation
  • Urban growth

ASJC Scopus subject areas

  • Information Systems
  • Geography, Planning and Development
  • Library and Information Sciences

Cite this

An ensemble approach to space-time interpolation. / Wentz, Elizabeth; Peuquet, Donna J.; Anderson, Sharolyn.

In: International Journal of Geographical Information Science, Vol. 24, No. 9, 2010, p. 1309-1325.

Research output: Contribution to journalArticle

Wentz, Elizabeth ; Peuquet, Donna J. ; Anderson, Sharolyn. / An ensemble approach to space-time interpolation. In: International Journal of Geographical Information Science. 2010 ; Vol. 24, No. 9. pp. 1309-1325.
@article{b4f22e7b1c78472a85e03584fa139d16,
title = "An ensemble approach to space-time interpolation",
abstract = "The availability of spatial data on an unprecedented scale as well as advancements in analytical and visualization techniques gives researchers the opportunity to study complex problems over large urban and regional areas. Nevertheless, few individual data sets exist that provide both the requisite spatial and/or temporal observational frequency to truly facilitate detailed investigations. Some data are collected frequently over time but only at a few geographic locations (e.g., weather stations). Similarly, other data are collected with a high level of spatial resolution but not at regular or frequent time intervals (e.g., satellite data). The purpose of this article is to present an interpolation approach that leverages the relative temporal richness of one data set with the relative spatial richness of another to fill in the gaps. Because different interpolation techniques are more appropriate than others for specific types of data, we propose a space-time interpolation approach whereby two interpolation methods - one for the temporal and one for the spatial dimension - are used in tandem to increase the accuracy results. We call our ensemble approach the space-time interpolation environment (STIE). The primary steps within this environment include a spatial interpolation processor, a temporal interpolation processor, and a calibration processor, which enforces phenomenonrelated behavioral constraints. The specific interpolation techniques used within the STIE can be chosen on the basis of suitability for the data and application at hand. In this article, we first describe STIE conceptually including the data input requirements, output structure, details of the primary steps, and the mechanism for coordinating the data within those steps. We then describe a case study focusing on urban land cover in Phoenix, Arizona, using our working implementation. Our empirical results show that our approach increased the accuracy for estimating urban land cover better than a single interpolation technique.",
keywords = "Arizona, Phoenix, Space-time interpolation, Urban growth",
author = "Elizabeth Wentz and Peuquet, {Donna J.} and Sharolyn Anderson",
year = "2010",
doi = "10.1080/13658816.2010.488238",
language = "English (US)",
volume = "24",
pages = "1309--1325",
journal = "International Journal of Geographical Information Science",
issn = "1365-8816",
publisher = "Taylor and Francis Ltd.",
number = "9",

}

TY - JOUR

T1 - An ensemble approach to space-time interpolation

AU - Wentz, Elizabeth

AU - Peuquet, Donna J.

AU - Anderson, Sharolyn

PY - 2010

Y1 - 2010

N2 - The availability of spatial data on an unprecedented scale as well as advancements in analytical and visualization techniques gives researchers the opportunity to study complex problems over large urban and regional areas. Nevertheless, few individual data sets exist that provide both the requisite spatial and/or temporal observational frequency to truly facilitate detailed investigations. Some data are collected frequently over time but only at a few geographic locations (e.g., weather stations). Similarly, other data are collected with a high level of spatial resolution but not at regular or frequent time intervals (e.g., satellite data). The purpose of this article is to present an interpolation approach that leverages the relative temporal richness of one data set with the relative spatial richness of another to fill in the gaps. Because different interpolation techniques are more appropriate than others for specific types of data, we propose a space-time interpolation approach whereby two interpolation methods - one for the temporal and one for the spatial dimension - are used in tandem to increase the accuracy results. We call our ensemble approach the space-time interpolation environment (STIE). The primary steps within this environment include a spatial interpolation processor, a temporal interpolation processor, and a calibration processor, which enforces phenomenonrelated behavioral constraints. The specific interpolation techniques used within the STIE can be chosen on the basis of suitability for the data and application at hand. In this article, we first describe STIE conceptually including the data input requirements, output structure, details of the primary steps, and the mechanism for coordinating the data within those steps. We then describe a case study focusing on urban land cover in Phoenix, Arizona, using our working implementation. Our empirical results show that our approach increased the accuracy for estimating urban land cover better than a single interpolation technique.

AB - The availability of spatial data on an unprecedented scale as well as advancements in analytical and visualization techniques gives researchers the opportunity to study complex problems over large urban and regional areas. Nevertheless, few individual data sets exist that provide both the requisite spatial and/or temporal observational frequency to truly facilitate detailed investigations. Some data are collected frequently over time but only at a few geographic locations (e.g., weather stations). Similarly, other data are collected with a high level of spatial resolution but not at regular or frequent time intervals (e.g., satellite data). The purpose of this article is to present an interpolation approach that leverages the relative temporal richness of one data set with the relative spatial richness of another to fill in the gaps. Because different interpolation techniques are more appropriate than others for specific types of data, we propose a space-time interpolation approach whereby two interpolation methods - one for the temporal and one for the spatial dimension - are used in tandem to increase the accuracy results. We call our ensemble approach the space-time interpolation environment (STIE). The primary steps within this environment include a spatial interpolation processor, a temporal interpolation processor, and a calibration processor, which enforces phenomenonrelated behavioral constraints. The specific interpolation techniques used within the STIE can be chosen on the basis of suitability for the data and application at hand. In this article, we first describe STIE conceptually including the data input requirements, output structure, details of the primary steps, and the mechanism for coordinating the data within those steps. We then describe a case study focusing on urban land cover in Phoenix, Arizona, using our working implementation. Our empirical results show that our approach increased the accuracy for estimating urban land cover better than a single interpolation technique.

KW - Arizona

KW - Phoenix

KW - Space-time interpolation

KW - Urban growth

UR - http://www.scopus.com/inward/record.url?scp=77956408646&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=77956408646&partnerID=8YFLogxK

U2 - 10.1080/13658816.2010.488238

DO - 10.1080/13658816.2010.488238

M3 - Article

AN - SCOPUS:77956408646

VL - 24

SP - 1309

EP - 1325

JO - International Journal of Geographical Information Science

JF - International Journal of Geographical Information Science

SN - 1365-8816

IS - 9

ER -