Bayesian and Classical Machine Learning Methods: A Comparison for Tree Species Classification with LiDAR Waveform Signatures
A plethora of information contained in full-waveform (FW) Light Detection and Ranging (LiDAR) data offers prospects for characterizing vegetation structures. This study aims to investigate the capacity of FW LiDAR data alone for tree species identification through the integration of waveform metrics...
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2017-12-01
|
| Series: | Remote Sensing |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2072-4292/10/1/39 |
| _version_ | 1818914564564058112 |
|---|---|
| author | Tan Zhou Sorin C. Popescu A. Michelle Lawing Marian Eriksson Bogdan M. Strimbu Paul C. Bürkner |
| author_facet | Tan Zhou Sorin C. Popescu A. Michelle Lawing Marian Eriksson Bogdan M. Strimbu Paul C. Bürkner |
| author_sort | Tan Zhou |
| collection | DOAJ |
| description | A plethora of information contained in full-waveform (FW) Light Detection and Ranging (LiDAR) data offers prospects for characterizing vegetation structures. This study aims to investigate the capacity of FW LiDAR data alone for tree species identification through the integration of waveform metrics with machine learning methods and Bayesian inference. Specifically, we first conducted automatic tree segmentation based on the waveform-based canopy height model (CHM) using three approaches including TreeVaW, watershed algorithms and the combination of TreeVaW and watershed (TW) algorithms. Subsequently, the Random forests (RF) and Conditional inference forests (CF) models were employed to identify important tree-level waveform metrics derived from three distinct sources, such as raw waveforms, composite waveforms, the waveform-based point cloud and the combined variables from these three sources. Further, we discriminated tree (gray pine, blue oak, interior live oak) and shrub species through the RF, CF and Bayesian multinomial logistic regression (BMLR) using important waveform metrics identified in this study. Results of the tree segmentation demonstrated that the TW algorithms outperformed other algorithms for delineating individual tree crowns. The CF model overcomes waveform metrics selection bias caused by the RF model which favors correlated metrics and enhances the accuracy of subsequent classification. We also found that composite waveforms are more informative than raw waveforms and waveform-based point cloud for characterizing tree species in our study area. Both classical machine learning methods (the RF and CF) and the BMLR generated satisfactory average overall accuracy (74% for the RF, 77% for the CF and 81% for the BMLR) and the BMLR slightly outperformed the other two methods. However, these three methods suffered from low individual classification accuracy for the blue oak which is prone to being misclassified as the interior live oak due to the similar characteristics of blue oak and interior live oak. Uncertainty estimates from the BMLR method compensate for this downside by providing classification results in a probabilistic sense and rendering users with more confidence in interpreting and applying classification results to real-world tasks such as forest inventory. Overall, this study recommends the CF method for feature selection and suggests that BMLR could be a superior alternative to classical machining learning methods. |
| first_indexed | 2024-12-19T23:48:23Z |
| format | Article |
| id | doaj.art-3a36f44471ef42038647843e83aceea8 |
| institution | Directory Open Access Journal |
| issn | 2072-4292 |
| language | English |
| last_indexed | 2024-12-19T23:48:23Z |
| publishDate | 2017-12-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Remote Sensing |
| spelling | doaj.art-3a36f44471ef42038647843e83aceea82022-12-21T20:01:14ZengMDPI AGRemote Sensing2072-42922017-12-011013910.3390/rs10010039rs10010039Bayesian and Classical Machine Learning Methods: A Comparison for Tree Species Classification with LiDAR Waveform SignaturesTan Zhou0Sorin C. Popescu1A. Michelle Lawing2Marian Eriksson3Bogdan M. Strimbu4Paul C. Bürkner5LiDAR Applications for the Study of Ecosystems with Remote Sensing (LASERS) Laboratory, Department of Ecosystem Science and Management, Texas A&M University, College Station, TX 77843, USALiDAR Applications for the Study of Ecosystems with Remote Sensing (LASERS) Laboratory, Department of Ecosystem Science and Management, Texas A&M University, College Station, TX 77843, USADepartment of Ecosystem Science and Management, Texas A&M University, College Station, TX 77843, USADepartment of Ecosystem Science and Management, Texas A&M University, College Station, TX 77843, USADepartment of Forest Engineering, Resources and Management, Oregon State University, Corvallis, OR 97331, USAInstitute of Psychology, University of Münster, Münster 48149, GermanyA plethora of information contained in full-waveform (FW) Light Detection and Ranging (LiDAR) data offers prospects for characterizing vegetation structures. This study aims to investigate the capacity of FW LiDAR data alone for tree species identification through the integration of waveform metrics with machine learning methods and Bayesian inference. Specifically, we first conducted automatic tree segmentation based on the waveform-based canopy height model (CHM) using three approaches including TreeVaW, watershed algorithms and the combination of TreeVaW and watershed (TW) algorithms. Subsequently, the Random forests (RF) and Conditional inference forests (CF) models were employed to identify important tree-level waveform metrics derived from three distinct sources, such as raw waveforms, composite waveforms, the waveform-based point cloud and the combined variables from these three sources. Further, we discriminated tree (gray pine, blue oak, interior live oak) and shrub species through the RF, CF and Bayesian multinomial logistic regression (BMLR) using important waveform metrics identified in this study. Results of the tree segmentation demonstrated that the TW algorithms outperformed other algorithms for delineating individual tree crowns. The CF model overcomes waveform metrics selection bias caused by the RF model which favors correlated metrics and enhances the accuracy of subsequent classification. We also found that composite waveforms are more informative than raw waveforms and waveform-based point cloud for characterizing tree species in our study area. Both classical machine learning methods (the RF and CF) and the BMLR generated satisfactory average overall accuracy (74% for the RF, 77% for the CF and 81% for the BMLR) and the BMLR slightly outperformed the other two methods. However, these three methods suffered from low individual classification accuracy for the blue oak which is prone to being misclassified as the interior live oak due to the similar characteristics of blue oak and interior live oak. Uncertainty estimates from the BMLR method compensate for this downside by providing classification results in a probabilistic sense and rendering users with more confidence in interpreting and applying classification results to real-world tasks such as forest inventory. Overall, this study recommends the CF method for feature selection and suggests that BMLR could be a superior alternative to classical machining learning methods.https://www.mdpi.com/2072-4292/10/1/39Bayesian multinomial logistic regressionconditional inference forestsRandom forestswaveform signaturestree segmentationwatershedcomposite waveformdecomposition |
| spellingShingle | Tan Zhou Sorin C. Popescu A. Michelle Lawing Marian Eriksson Bogdan M. Strimbu Paul C. Bürkner Bayesian and Classical Machine Learning Methods: A Comparison for Tree Species Classification with LiDAR Waveform Signatures Remote Sensing Bayesian multinomial logistic regression conditional inference forests Random forests waveform signatures tree segmentation watershed composite waveform decomposition |
| title | Bayesian and Classical Machine Learning Methods: A Comparison for Tree Species Classification with LiDAR Waveform Signatures |
| title_full | Bayesian and Classical Machine Learning Methods: A Comparison for Tree Species Classification with LiDAR Waveform Signatures |
| title_fullStr | Bayesian and Classical Machine Learning Methods: A Comparison for Tree Species Classification with LiDAR Waveform Signatures |
| title_full_unstemmed | Bayesian and Classical Machine Learning Methods: A Comparison for Tree Species Classification with LiDAR Waveform Signatures |
| title_short | Bayesian and Classical Machine Learning Methods: A Comparison for Tree Species Classification with LiDAR Waveform Signatures |
| title_sort | bayesian and classical machine learning methods a comparison for tree species classification with lidar waveform signatures |
| topic | Bayesian multinomial logistic regression conditional inference forests Random forests waveform signatures tree segmentation watershed composite waveform decomposition |
| url | https://www.mdpi.com/2072-4292/10/1/39 |
| work_keys_str_mv | AT tanzhou bayesianandclassicalmachinelearningmethodsacomparisonfortreespeciesclassificationwithlidarwaveformsignatures AT sorincpopescu bayesianandclassicalmachinelearningmethodsacomparisonfortreespeciesclassificationwithlidarwaveformsignatures AT amichellelawing bayesianandclassicalmachinelearningmethodsacomparisonfortreespeciesclassificationwithlidarwaveformsignatures AT marianeriksson bayesianandclassicalmachinelearningmethodsacomparisonfortreespeciesclassificationwithlidarwaveformsignatures AT bogdanmstrimbu bayesianandclassicalmachinelearningmethodsacomparisonfortreespeciesclassificationwithlidarwaveformsignatures AT paulcburkner bayesianandclassicalmachinelearningmethodsacomparisonfortreespeciesclassificationwithlidarwaveformsignatures |