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

Title: Wildfire Mapping in Interior Alaska Using Deep Neural Networks on Imbalanced Datasets

Abstract

Wildfires are the dominant disturbance impacting many regions in Alaska and is expected to intensify due to climate change. Accurate tracking and quantification of wildfires are important for climate modeling and ecological studies in this region. Remote sensing platforms (e.g., MODIS, Landsat) are an important tool for mapping wildfires in Alaska at varying spatial and temporal scales. Deep neural networks (DNN) have shown superior performance in many classification problems such as high-dimensional remote sensing data. Detection of wildfires is an imbalanced classification problem where one class containsa much smaller or larger sample size and DNNs performance can decline. We take a known weight-selection strategy during DNN training and apply them to MODIS variables (e.g., NDVI, surface reflectance) for a binary classification (i.e., wildfire or no-wildfire) across Alaska during the 2004 wildfire year, which is one of the largest on record. The method splits the input training data into sets, one for training the DNN to update weights and the other that splits into a validation set that is equally balanced between the wildfire and no-wildfire class. The performance is monitored on the validation set to select the weights based on the best validation-loss score. This approach was applied to twomore » sampled datasets, such as where the no-wildfire class can significantly outweigh the wildfire class. The normal DNN training strategy was unable to map wildfires for the highly imbalanced dataset, however, the weight-selection strategy was able to map wildfires very accurately (0.96 recall score for 78,702 wildfire pixels (500×500 m)).« less

Authors:
 [1]; ORCiD logo [1]; ORCiD logo [1]
  1. ORNL
Publication Date:
Research Org.:
Oak Ridge National Laboratory (ORNL), Oak Ridge, TN (United States)
Sponsoring Org.:
USDOE Office of Science (SC), Biological and Environmental Research (BER)
OSTI Identifier:
1491321
DOE Contract Number:  
AC05-00OR22725
Resource Type:
Conference
Resource Relation:
Conference: IEEE International Conference on Data Mining Workshops - Singapore, , Singapore - 11/17/2018 10:00:00 AM-11/20/2018 10:00:00 AM
Country of Publication:
United States
Language:
English

Citation Formats

Langford, Zach, Kumar, Jitendra, and Hoffman, Forrest. Wildfire Mapping in Interior Alaska Using Deep Neural Networks on Imbalanced Datasets. United States: N. p., 2018. Web. doi:10.1109/ICDMW.2018.00116.
Langford, Zach, Kumar, Jitendra, & Hoffman, Forrest. Wildfire Mapping in Interior Alaska Using Deep Neural Networks on Imbalanced Datasets. United States. https://doi.org/10.1109/ICDMW.2018.00116
Langford, Zach, Kumar, Jitendra, and Hoffman, Forrest. 2018. "Wildfire Mapping in Interior Alaska Using Deep Neural Networks on Imbalanced Datasets". United States. https://doi.org/10.1109/ICDMW.2018.00116. https://www.osti.gov/servlets/purl/1491321.
@article{osti_1491321,
title = {Wildfire Mapping in Interior Alaska Using Deep Neural Networks on Imbalanced Datasets},
author = {Langford, Zach and Kumar, Jitendra and Hoffman, Forrest},
abstractNote = {Wildfires are the dominant disturbance impacting many regions in Alaska and is expected to intensify due to climate change. Accurate tracking and quantification of wildfires are important for climate modeling and ecological studies in this region. Remote sensing platforms (e.g., MODIS, Landsat) are an important tool for mapping wildfires in Alaska at varying spatial and temporal scales. Deep neural networks (DNN) have shown superior performance in many classification problems such as high-dimensional remote sensing data. Detection of wildfires is an imbalanced classification problem where one class containsa much smaller or larger sample size and DNNs performance can decline. We take a known weight-selection strategy during DNN training and apply them to MODIS variables (e.g., NDVI, surface reflectance) for a binary classification (i.e., wildfire or no-wildfire) across Alaska during the 2004 wildfire year, which is one of the largest on record. The method splits the input training data into sets, one for training the DNN to update weights and the other that splits into a validation set that is equally balanced between the wildfire and no-wildfire class. The performance is monitored on the validation set to select the weights based on the best validation-loss score. This approach was applied to two sampled datasets, such as where the no-wildfire class can significantly outweigh the wildfire class. The normal DNN training strategy was unable to map wildfires for the highly imbalanced dataset, however, the weight-selection strategy was able to map wildfires very accurately (0.96 recall score for 78,702 wildfire pixels (500×500 m)).},
doi = {10.1109/ICDMW.2018.00116},
url = {https://www.osti.gov/biblio/1491321}, journal = {},
number = ,
volume = ,
place = {United States},
year = {2018},
month = {11}
}

Conference:
Other availability
Please see Document Availability for additional information on obtaining the full-text document. Library patrons may search WorldCat to identify libraries that hold this conference proceeding.

Save / Share:

Works referenced in this record:

Convolutional Neural Network Approach for Mapping Arctic Vegetation Using Multi-Sensor Remote Sensing Fusion
conference, November 2017


Relative importance of weather and climate on wildfire growth in interior Alaska
journal, January 2011


Deep Transfer Learning for Crop Yield Prediction with Remote Sensing Data
conference, June 2018

  • Wang, Anna X.; Tran, Caelin; Desai, Nikhil
  • COMPASS '18: ACM SIGCAS Conference on Computing and Sustainable Societies, Proceedings of the 1st ACM SIGCAS Conference on Computing and Sustainable Societies
  • https://doi.org/10.1145/3209811.3212707

Quantifying CO emissions from the 2004 Alaskan wildfires using MOPITT CO data
journal, January 2005


MODIS-informed greenness responses to daytime land surface temperature fluctuations and wildfire disturbances in the Alaskan Yukon River Basin
journal, November 2012


Deep learning
journal, May 2015


Training deep neural networks on imbalanced data sets
conference, July 2016


SMOTE: Synthetic Minority Over-sampling Technique
journal, January 2002


Representativeness-based sampling network design for the State of Alaska
journal, June 2013


Growing season and spatial variations of carbon fluxes of Arctic and boreal ecosystems in Alaska (USA)
journal, December 2013


XGBoost: A Scalable Tree Boosting System
conference, January 2016


Deep Multi-patch Aggregation Network for Image Style, Aesthetics, and Quality Estimation
conference, December 2015