Spatial factor models for high-dimensional and large spatial data: An application in forest variable mapping

Daniel Taylor-Rodriguez; Andrew O. Finley; Abhirup Datta; Chad Babcock; Hans Erik Andersen; Bruce D. Cook; Douglas C. Morton; Sudipto Banerjee

doi:10.5705/ss.202018.0005

Spatial factor models for high-dimensional and large spatial data: An application in forest variable mapping

Daniel Taylor-Rodriguez, Andrew O. Finley, Abhirup Datta, Chad Babcock, Hans Erik Andersen, Bruce D. Cook, Douglas C. Morton, Sudipto Banerjee

Forest Resources

Research output: Contribution to journal › Article › peer-review

14 Scopus citations

Abstract

Gathering information about forest variables is an expensive and arduous activity. Therefore, directly collecting the data required to produce high-resolution maps over large spatial domains is infeasible. Next-generation collection initiatives for remotely sensed light detection and ranging (LiDAR) data are specifically aimed at producing complete-coverage maps over large spatial domains. Given that LiDAR data and forest characteristics are often strongly correlated, it is possible to use the former to model, predict, and map forest variables over regions of interest. This entails dealing with high-dimensional (∼10²) spatially dependent LiDAR outcomes over a large number of locations (∼10⁵ − 10⁶). With this in mind, we develop the spatial factor nearest neighbor Gaussian process (SF-NNGP) model, which we embed in a two-stage approach that connects the spatial structure found in LiDAR signals with forest variables. We provide a simulation experiment that demonstrates the inferential and predictive performance of the SF-NNGP, and use the two-stage modeling strategy to generate complete-coverage maps of the forest variables, with associated uncertainty, over a large region of boreal forests in interior Alaska.

Original language	English (US)
Pages (from-to)	1155-1180
Number of pages	26
Journal	Statistica Sinica
Volume	29
Issue number	3
DOIs	https://doi.org/10.5705/ss.202018.0005
State	Published - 2019

Bibliographical note

Funding Information:
The research presented in this study was partially supported by NASA’s Arctic-Boreal Vulnerability Experiment (ABoVE) and Carbon Monitoring System (CMS) programs. Additional support was provided by the United States Forest Service Pacific Northwest Research Station. Finley was supported by National Science Foundation (NSF) DMS-1513481, EF-1137309, and EF-1241874, and Finley and Taylor-Rodriguez were supported by EF-1253225. Banerjee was supported by NSF DMS-1513654, NSF IIS-1562303, and NIH/NIEHS 1R01ES027027-01.

Publisher Copyright:
© 2019 Institute of Statistical Science. All rights reserved.

Keywords

Forest outcomes
LiDAR data
Nearest neighbor Gaussian processes
Spatial prediction

Publisher link

10.5705/ss.202018.0005

Find It

Find It at U of M Libraries

Cite this

@article{302f5eaf53f44f8ba9032317e3ffb503,

title = "Spatial factor models for high-dimensional and large spatial data: An application in forest variable mapping",

abstract = "Gathering information about forest variables is an expensive and arduous activity. Therefore, directly collecting the data required to produce high-resolution maps over large spatial domains is infeasible. Next-generation collection initiatives for remotely sensed light detection and ranging (LiDAR) data are specifically aimed at producing complete-coverage maps over large spatial domains. Given that LiDAR data and forest characteristics are often strongly correlated, it is possible to use the former to model, predict, and map forest variables over regions of interest. This entails dealing with high-dimensional (∼102) spatially dependent LiDAR outcomes over a large number of locations (∼105 − 106). With this in mind, we develop the spatial factor nearest neighbor Gaussian process (SF-NNGP) model, which we embed in a two-stage approach that connects the spatial structure found in LiDAR signals with forest variables. We provide a simulation experiment that demonstrates the inferential and predictive performance of the SF-NNGP, and use the two-stage modeling strategy to generate complete-coverage maps of the forest variables, with associated uncertainty, over a large region of boreal forests in interior Alaska.",

keywords = "Forest outcomes, LiDAR data, Nearest neighbor Gaussian processes, Spatial prediction",

author = "Daniel Taylor-Rodriguez and Finley, {Andrew O.} and Abhirup Datta and Chad Babcock and Andersen, {Hans Erik} and Cook, {Bruce D.} and Morton, {Douglas C.} and Sudipto Banerjee",

note = "Funding Information: The research presented in this study was partially supported by NASA{\textquoteright}s Arctic-Boreal Vulnerability Experiment (ABoVE) and Carbon Monitoring System (CMS) programs. Additional support was provided by the United States Forest Service Pacific Northwest Research Station. Finley was supported by National Science Foundation (NSF) DMS-1513481, EF-1137309, and EF-1241874, and Finley and Taylor-Rodriguez were supported by EF-1253225. Banerjee was supported by NSF DMS-1513654, NSF IIS-1562303, and NIH/NIEHS 1R01ES027027-01. Publisher Copyright: {\textcopyright} 2019 Institute of Statistical Science. All rights reserved.",

year = "2019",

doi = "10.5705/ss.202018.0005",

language = "English (US)",

volume = "29",

pages = "1155--1180",

journal = "Statistica Sinica",

issn = "1017-0405",

publisher = "Institute of Statistical Science",

number = "3",

}

TY - JOUR

T1 - Spatial factor models for high-dimensional and large spatial data

T2 - An application in forest variable mapping

AU - Taylor-Rodriguez, Daniel

AU - Finley, Andrew O.

AU - Datta, Abhirup

AU - Babcock, Chad

AU - Andersen, Hans Erik

AU - Cook, Bruce D.

AU - Morton, Douglas C.

AU - Banerjee, Sudipto

N1 - Funding Information: The research presented in this study was partially supported by NASA’s Arctic-Boreal Vulnerability Experiment (ABoVE) and Carbon Monitoring System (CMS) programs. Additional support was provided by the United States Forest Service Pacific Northwest Research Station. Finley was supported by National Science Foundation (NSF) DMS-1513481, EF-1137309, and EF-1241874, and Finley and Taylor-Rodriguez were supported by EF-1253225. Banerjee was supported by NSF DMS-1513654, NSF IIS-1562303, and NIH/NIEHS 1R01ES027027-01. Publisher Copyright: © 2019 Institute of Statistical Science. All rights reserved.

PY - 2019

Y1 - 2019

N2 - Gathering information about forest variables is an expensive and arduous activity. Therefore, directly collecting the data required to produce high-resolution maps over large spatial domains is infeasible. Next-generation collection initiatives for remotely sensed light detection and ranging (LiDAR) data are specifically aimed at producing complete-coverage maps over large spatial domains. Given that LiDAR data and forest characteristics are often strongly correlated, it is possible to use the former to model, predict, and map forest variables over regions of interest. This entails dealing with high-dimensional (∼102) spatially dependent LiDAR outcomes over a large number of locations (∼105 − 106). With this in mind, we develop the spatial factor nearest neighbor Gaussian process (SF-NNGP) model, which we embed in a two-stage approach that connects the spatial structure found in LiDAR signals with forest variables. We provide a simulation experiment that demonstrates the inferential and predictive performance of the SF-NNGP, and use the two-stage modeling strategy to generate complete-coverage maps of the forest variables, with associated uncertainty, over a large region of boreal forests in interior Alaska.

AB - Gathering information about forest variables is an expensive and arduous activity. Therefore, directly collecting the data required to produce high-resolution maps over large spatial domains is infeasible. Next-generation collection initiatives for remotely sensed light detection and ranging (LiDAR) data are specifically aimed at producing complete-coverage maps over large spatial domains. Given that LiDAR data and forest characteristics are often strongly correlated, it is possible to use the former to model, predict, and map forest variables over regions of interest. This entails dealing with high-dimensional (∼102) spatially dependent LiDAR outcomes over a large number of locations (∼105 − 106). With this in mind, we develop the spatial factor nearest neighbor Gaussian process (SF-NNGP) model, which we embed in a two-stage approach that connects the spatial structure found in LiDAR signals with forest variables. We provide a simulation experiment that demonstrates the inferential and predictive performance of the SF-NNGP, and use the two-stage modeling strategy to generate complete-coverage maps of the forest variables, with associated uncertainty, over a large region of boreal forests in interior Alaska.

KW - Forest outcomes

KW - LiDAR data

KW - Nearest neighbor Gaussian processes

KW - Spatial prediction

UR - http://www.scopus.com/inward/record.url?scp=85072089536&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85072089536&partnerID=8YFLogxK

U2 - 10.5705/ss.202018.0005

DO - 10.5705/ss.202018.0005

M3 - Article

AN - SCOPUS:85072089536

SN - 1017-0405

VL - 29

SP - 1155

EP - 1180

JO - Statistica Sinica

JF - Statistica Sinica

IS - 3

ER -

Spatial factor models for high-dimensional and large spatial data: An application in forest variable mapping

Abstract

Bibliographical note

Keywords

Publisher link

Find It

Other files and links

Fingerprint

Cite this