Deep Learning Classification of Cheatgrass Invasion in the Western United States Using Biophysical and Remote Sensing Data
Cheatgrass (<i>Bromus tectorum</i>) invasion is driving an emerging cycle of increased fire frequency and irreversible loss of wildlife habitat in the western US. Yet, detailed spatial information about its occurrence is still lacking for much of its presumably invaded range. Deep learni...
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2021-03-01
|
| Series: | Remote Sensing |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2072-4292/13/7/1246 |
| _version_ | 1797540115920191488 |
|---|---|
| author | Kyle B. Larson Aaron R. Tuor |
| author_facet | Kyle B. Larson Aaron R. Tuor |
| author_sort | Kyle B. Larson |
| collection | DOAJ |
| description | Cheatgrass (<i>Bromus tectorum</i>) invasion is driving an emerging cycle of increased fire frequency and irreversible loss of wildlife habitat in the western US. Yet, detailed spatial information about its occurrence is still lacking for much of its presumably invaded range. Deep learning (DL) has demonstrated success for remote sensing applications but is less tested on more challenging tasks like identifying biological invasions using sub-pixel phenomena. We compare two DL architectures and the more conventional Random Forest and Logistic Regression methods to improve upon a previous effort to map cheatgrass occurrence at >2% canopy cover. High-dimensional sets of biophysical, MODIS, and Landsat-7 ETM+ predictor variables are also compared to evaluate different multi-modal data strategies. All model configurations improved results relative to the case study and accuracy generally improved by combining data from both sensors with biophysical data. Cheatgrass occurrence is mapped at 30 m ground sample distance (GSD) with an estimated 78.1% accuracy, compared to 250-m GSD and 71% map accuracy in the case study. Furthermore, DL is shown to be competitive with well-established machine learning methods in a limited data regime, suggesting it can be an effective tool for mapping biological invasions and more broadly for multi-modal remote sensing applications. |
| first_indexed | 2024-03-10T12:55:31Z |
| format | Article |
| id | doaj.art-cbffa9fff9094585bfca1cadd2edd784 |
| institution | Directory Open Access Journal |
| issn | 2072-4292 |
| language | English |
| last_indexed | 2024-03-10T12:55:31Z |
| publishDate | 2021-03-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Remote Sensing |
| spelling | doaj.art-cbffa9fff9094585bfca1cadd2edd7842023-11-21T11:58:27ZengMDPI AGRemote Sensing2072-42922021-03-01137124610.3390/rs13071246Deep Learning Classification of Cheatgrass Invasion in the Western United States Using Biophysical and Remote Sensing DataKyle B. Larson0Aaron R. Tuor1Pacific Northwest National Laboratory, Richland, WA 99352, USAPacific Northwest National Laboratory, Richland, WA 99352, USACheatgrass (<i>Bromus tectorum</i>) invasion is driving an emerging cycle of increased fire frequency and irreversible loss of wildlife habitat in the western US. Yet, detailed spatial information about its occurrence is still lacking for much of its presumably invaded range. Deep learning (DL) has demonstrated success for remote sensing applications but is less tested on more challenging tasks like identifying biological invasions using sub-pixel phenomena. We compare two DL architectures and the more conventional Random Forest and Logistic Regression methods to improve upon a previous effort to map cheatgrass occurrence at >2% canopy cover. High-dimensional sets of biophysical, MODIS, and Landsat-7 ETM+ predictor variables are also compared to evaluate different multi-modal data strategies. All model configurations improved results relative to the case study and accuracy generally improved by combining data from both sensors with biophysical data. Cheatgrass occurrence is mapped at 30 m ground sample distance (GSD) with an estimated 78.1% accuracy, compared to 250-m GSD and 71% map accuracy in the case study. Furthermore, DL is shown to be competitive with well-established machine learning methods in a limited data regime, suggesting it can be an effective tool for mapping biological invasions and more broadly for multi-modal remote sensing applications.https://www.mdpi.com/2072-4292/13/7/1246deep learningmachine learningRandom Forestsupervised classificationlogistic regressionland cover classification |
| spellingShingle | Kyle B. Larson Aaron R. Tuor Deep Learning Classification of Cheatgrass Invasion in the Western United States Using Biophysical and Remote Sensing Data Remote Sensing deep learning machine learning Random Forest supervised classification logistic regression land cover classification |
| title | Deep Learning Classification of Cheatgrass Invasion in the Western United States Using Biophysical and Remote Sensing Data |
| title_full | Deep Learning Classification of Cheatgrass Invasion in the Western United States Using Biophysical and Remote Sensing Data |
| title_fullStr | Deep Learning Classification of Cheatgrass Invasion in the Western United States Using Biophysical and Remote Sensing Data |
| title_full_unstemmed | Deep Learning Classification of Cheatgrass Invasion in the Western United States Using Biophysical and Remote Sensing Data |
| title_short | Deep Learning Classification of Cheatgrass Invasion in the Western United States Using Biophysical and Remote Sensing Data |
| title_sort | deep learning classification of cheatgrass invasion in the western united states using biophysical and remote sensing data |
| topic | deep learning machine learning Random Forest supervised classification logistic regression land cover classification |
| url | https://www.mdpi.com/2072-4292/13/7/1246 |
| work_keys_str_mv | AT kyleblarson deeplearningclassificationofcheatgrassinvasioninthewesternunitedstatesusingbiophysicalandremotesensingdata AT aaronrtuor deeplearningclassificationofcheatgrassinvasioninthewesternunitedstatesusingbiophysicalandremotesensingdata |