skip to main content
OSTI.GOV title logo U.S. Department of Energy
Office of Scientific and Technical Information

Title: Integration of historical map and aerial imagery to characterize long-term land-use change and landscape dynamics: An object-based analysis via Random Forests

Abstract

Tracking Earth’s past helps us to move from hindsight to foresight in seeking landscape sustainability, a pursuit aided by modern mapping capabilities but hindered by a dearth of historical landscape information. To fill the data gap and exemplify the use of old map for land-use change sciences, we combined a paper-based US civil war map and aerial photos to derive land-use history and landscape dynamics at fine scales for a region near Chancellorsville, VA, USA, from 1867 to 2014. We also tested how advanced algorithms—object-based analysis and random forests (RF)—could aid in data processing. Automatic classification of the 1867 map proved difficult, but its manual digitization could benefit from object-based image segmentation. Classifying digital aerial images was more accurate via the object-based than pixel-based method, but only if the images were segmented appropriately. In the object-based classification, spectral-based features were much more important than shape/geometry features, as ranked by RF. During the 147 years, 32% of the region changed in land type. Settlement and roads increased in extent by 1850% and 691%, respectively, and woodland decreased by 19%. These changes fragmented the landscape, altered the hydrological regime, and affected river morphology. The utility of old maps exemplified here provides anmore » impetus for exploring and leveraging extant old maps or historical records to support land-use and global change research. Our study also connotes the importance of geotagging current data, such as photos and videos, that may serve as a baseline to document future landscape change.« less

Authors:
; ; ; ; ; ;
Publication Date:
Research Org.:
Pacific Northwest National Lab. (PNNL), Richland, WA (United States)
Sponsoring Org.:
USDOE
OSTI Identifier:
1501837
Report Number(s):
PNNL-SA-129615
Journal ID: ISSN 1470-160X
DOE Contract Number:  
AC05-76RL01830
Resource Type:
Journal Article
Journal Name:
Ecological Indicators
Additional Journal Information:
Journal Volume: 95; Journal Issue: P1; Journal ID: ISSN 1470-160X
Publisher:
Elsevier
Country of Publication:
United States
Language:
English
Subject:
Historical map, object-based, Random Forests, old map

Citation Formats

Liu, Dan, Toman, Elizabeth, Fuller, Zane, Chen, Gang, Londo, Alexis, Zhang, Xuesong, and Zhao, Kaiguang. Integration of historical map and aerial imagery to characterize long-term land-use change and landscape dynamics: An object-based analysis via Random Forests. United States: N. p., 2018. Web. doi:10.1016/j.ecolind.2018.08.004.
Liu, Dan, Toman, Elizabeth, Fuller, Zane, Chen, Gang, Londo, Alexis, Zhang, Xuesong, & Zhao, Kaiguang. Integration of historical map and aerial imagery to characterize long-term land-use change and landscape dynamics: An object-based analysis via Random Forests. United States. doi:10.1016/j.ecolind.2018.08.004.
Liu, Dan, Toman, Elizabeth, Fuller, Zane, Chen, Gang, Londo, Alexis, Zhang, Xuesong, and Zhao, Kaiguang. Sat . "Integration of historical map and aerial imagery to characterize long-term land-use change and landscape dynamics: An object-based analysis via Random Forests". United States. doi:10.1016/j.ecolind.2018.08.004.
@article{osti_1501837,
title = {Integration of historical map and aerial imagery to characterize long-term land-use change and landscape dynamics: An object-based analysis via Random Forests},
author = {Liu, Dan and Toman, Elizabeth and Fuller, Zane and Chen, Gang and Londo, Alexis and Zhang, Xuesong and Zhao, Kaiguang},
abstractNote = {Tracking Earth’s past helps us to move from hindsight to foresight in seeking landscape sustainability, a pursuit aided by modern mapping capabilities but hindered by a dearth of historical landscape information. To fill the data gap and exemplify the use of old map for land-use change sciences, we combined a paper-based US civil war map and aerial photos to derive land-use history and landscape dynamics at fine scales for a region near Chancellorsville, VA, USA, from 1867 to 2014. We also tested how advanced algorithms—object-based analysis and random forests (RF)—could aid in data processing. Automatic classification of the 1867 map proved difficult, but its manual digitization could benefit from object-based image segmentation. Classifying digital aerial images was more accurate via the object-based than pixel-based method, but only if the images were segmented appropriately. In the object-based classification, spectral-based features were much more important than shape/geometry features, as ranked by RF. During the 147 years, 32% of the region changed in land type. Settlement and roads increased in extent by 1850% and 691%, respectively, and woodland decreased by 19%. These changes fragmented the landscape, altered the hydrological regime, and affected river morphology. The utility of old maps exemplified here provides an impetus for exploring and leveraging extant old maps or historical records to support land-use and global change research. Our study also connotes the importance of geotagging current data, such as photos and videos, that may serve as a baseline to document future landscape change.},
doi = {10.1016/j.ecolind.2018.08.004},
journal = {Ecological Indicators},
issn = {1470-160X},
number = P1,
volume = 95,
place = {United States},
year = {2018},
month = {12}
}