2019
DOI: 10.3390/w11061158
|View full text |Cite
|
Sign up to set email alerts
|

Using Random Forest Classification and Nationally Available Geospatial Data to Screen for Wetlands over Large Geographic Regions

Abstract: Wetland impact assessments are an integral part of infrastructure projects aimed at protecting the important services wetlands provide for water resources and ecosystems. However, wetland surveys with the level of accuracy required by federal regulators can be time-consuming and costly. Streamlining this process by using already available geospatial data and classification algorithms to target more detailed wetland mapping efforts may support environmental planning efforts. The objective of this study was to c… Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
1
1
1
1

Citation Types

0
7
0

Year Published

2020
2020
2024
2024

Publication Types

Select...
8
2

Relationship

0
10

Authors

Journals

citations
Cited by 14 publications
(7 citation statements)
references
References 40 publications
0
7
0
Order By: Relevance
“…However, some machine learning models, in particular deep neural networks (DNNs), cannot be very well understood (Räz and Beisbart, 2022;Sullivan, 2022); the performance will vary depending on the architectures and hyperparameter selection of the chosen algorithm (Shrestha and Mahmood, 2019;Yang and Shami, 2020); applying them globally is extremely hard (Shanthamallu and Spanias, 2021). Therefore, this study only selected the random forest (RF) algorithm for comparison on variable importance ranking because it is currently one of the most widely used algorithms for large-scale wetland remote sensing (Felton et al, 2019;Millard and Richardson, 2013;Tian et al, 2016). This work calculated JM distance, ED, SAD, and RF variable importance (VI) ranking models for importance evaluation to calculate separability.…”
Section: B Methodsmentioning
confidence: 99%
“…However, some machine learning models, in particular deep neural networks (DNNs), cannot be very well understood (Räz and Beisbart, 2022;Sullivan, 2022); the performance will vary depending on the architectures and hyperparameter selection of the chosen algorithm (Shrestha and Mahmood, 2019;Yang and Shami, 2020); applying them globally is extremely hard (Shanthamallu and Spanias, 2021). Therefore, this study only selected the random forest (RF) algorithm for comparison on variable importance ranking because it is currently one of the most widely used algorithms for large-scale wetland remote sensing (Felton et al, 2019;Millard and Richardson, 2013;Tian et al, 2016). This work calculated JM distance, ED, SAD, and RF variable importance (VI) ranking models for importance evaluation to calculate separability.…”
Section: B Methodsmentioning
confidence: 99%
“…The second approach (RFM) is a non-parametric random forest (RF) ensemble classifier that produces classification trees from a training data set (Breiman, 2001;Cutler et al, 2007). RF can well handle high-dimensional data and has been widely used for wetland identification (Felton et al, 2019;Xue et al, 2018). In this study, RFM was trained to simulate current wetland distribution at 15″ from eight predictors, including five climate variables (average temperature, precipitation, solar radiation, wind speed, and water vapor pressure), two hydrologic variables (drainage basin and river network), and the topographic variable TI.…”
Section: Predictive Models For Wetland Distributionmentioning
confidence: 99%
“…Decision Tree (DT) is one of the classification techniques that uses the branches method to illustrate decision-making in each possible outcome (Felton et al, 2019). Structurally, DT comprises three kinds of nodes that frame an established tree, which a tree required to have 'root node,' 'internal node,' and 'leaf.'…”
Section: Random Forest Algorithm For Classification Document Subject mentioning
confidence: 99%