Mapping of the Canopy Openings in Mixed Beech–Fir Forest at Sentinel-2 Subpixel Level Using UAV and Machine Learning Approach
The presented study demonstrates a bi-sensor approach suitable for rapid and precise up-to-date mapping of forest canopy gaps for the larger spatial extent. The approach makes use of Unmanned Aerial Vehicle (UAV) red, green and blue (RGB) images on smaller areas for highly precise forest canopy mask...
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2020-11-01
|
| Series: | Remote Sensing |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2072-4292/12/23/3925 |
| _version_ | 1797546148806787072 |
|---|---|
| author | Ivan Pilaš Mateo Gašparović Alan Novkinić Damir Klobučar |
| author_facet | Ivan Pilaš Mateo Gašparović Alan Novkinić Damir Klobučar |
| author_sort | Ivan Pilaš |
| collection | DOAJ |
| description | The presented study demonstrates a bi-sensor approach suitable for rapid and precise up-to-date mapping of forest canopy gaps for the larger spatial extent. The approach makes use of Unmanned Aerial Vehicle (UAV) red, green and blue (RGB) images on smaller areas for highly precise forest canopy mask creation. Sentinel-2 was used as a scaling platform for transferring information from the UAV to a wider spatial extent. Various approaches to an improvement in the predictive performance were examined: (I) the highest <i>R</i><sup>2</sup> of the single satellite index was 0.57, (II) the highest <i>R</i><sup>2</sup> using multiple features obtained from the single-date, S-2 image was 0.624, and (III) the highest <i>R</i><sup>2</sup> on the multitemporal set of S-2 images was 0.697. Satellite indices such as Atmospherically Resistant Vegetation Index (ARVI), Infrared Percentage Vegetation Index (IPVI), Normalized Difference Index (NDI45), Pigment-Specific Simple Ratio Index (PSSRa), Modified Chlorophyll Absorption Ratio Index (MCARI), Color Index (CI), Redness Index (RI), and Normalized Difference Turbidity Index (NDTI) were the dominant predictors in most of the Machine Learning (ML) algorithms. The more complex ML algorithms such as the Support Vector Machines (SVM), Random Forest (RF), Stochastic Gradient Boosting (GBM), Extreme Gradient Boosting (XGBoost), and Catboost that provided the best performance on the training set exhibited weaker generalization capabilities. Therefore, a simpler and more robust Elastic Net (ENET) algorithm was chosen for the final map creation. |
| first_indexed | 2024-03-10T14:25:58Z |
| format | Article |
| id | doaj.art-26f52d9668bc44cb94dbb10022c468a1 |
| institution | Directory Open Access Journal |
| issn | 2072-4292 |
| language | English |
| last_indexed | 2024-03-10T14:25:58Z |
| publishDate | 2020-11-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Remote Sensing |
| spelling | doaj.art-26f52d9668bc44cb94dbb10022c468a12023-11-20T22:56:40ZengMDPI AGRemote Sensing2072-42922020-11-011223392510.3390/rs12233925Mapping of the Canopy Openings in Mixed Beech–Fir Forest at Sentinel-2 Subpixel Level Using UAV and Machine Learning ApproachIvan Pilaš0Mateo Gašparović1Alan Novkinić2Damir Klobučar3Croatian Forest Research Institute, Division of Ecology, Cvjetno naselje 41, 10450 Jastrebarsko, CroatiaFaculty of Geodesy, University of Zagreb, Kačićeva 26, 10000 Zagreb, CroatiaCroatian Forests Ltd., Ivana Meštrovića 28, 48000 Koprivnica, CroatiaCroatian Forests Ltd., Ivana Meštrovića 28, 48000 Koprivnica, CroatiaThe presented study demonstrates a bi-sensor approach suitable for rapid and precise up-to-date mapping of forest canopy gaps for the larger spatial extent. The approach makes use of Unmanned Aerial Vehicle (UAV) red, green and blue (RGB) images on smaller areas for highly precise forest canopy mask creation. Sentinel-2 was used as a scaling platform for transferring information from the UAV to a wider spatial extent. Various approaches to an improvement in the predictive performance were examined: (I) the highest <i>R</i><sup>2</sup> of the single satellite index was 0.57, (II) the highest <i>R</i><sup>2</sup> using multiple features obtained from the single-date, S-2 image was 0.624, and (III) the highest <i>R</i><sup>2</sup> on the multitemporal set of S-2 images was 0.697. Satellite indices such as Atmospherically Resistant Vegetation Index (ARVI), Infrared Percentage Vegetation Index (IPVI), Normalized Difference Index (NDI45), Pigment-Specific Simple Ratio Index (PSSRa), Modified Chlorophyll Absorption Ratio Index (MCARI), Color Index (CI), Redness Index (RI), and Normalized Difference Turbidity Index (NDTI) were the dominant predictors in most of the Machine Learning (ML) algorithms. The more complex ML algorithms such as the Support Vector Machines (SVM), Random Forest (RF), Stochastic Gradient Boosting (GBM), Extreme Gradient Boosting (XGBoost), and Catboost that provided the best performance on the training set exhibited weaker generalization capabilities. Therefore, a simpler and more robust Elastic Net (ENET) algorithm was chosen for the final map creation.https://www.mdpi.com/2072-4292/12/23/3925Sentinel-2UAVDJI dronemachine learningforest canopycanopy gaps |
| spellingShingle | Ivan Pilaš Mateo Gašparović Alan Novkinić Damir Klobučar Mapping of the Canopy Openings in Mixed Beech–Fir Forest at Sentinel-2 Subpixel Level Using UAV and Machine Learning Approach Remote Sensing Sentinel-2 UAV DJI drone machine learning forest canopy canopy gaps |
| title | Mapping of the Canopy Openings in Mixed Beech–Fir Forest at Sentinel-2 Subpixel Level Using UAV and Machine Learning Approach |
| title_full | Mapping of the Canopy Openings in Mixed Beech–Fir Forest at Sentinel-2 Subpixel Level Using UAV and Machine Learning Approach |
| title_fullStr | Mapping of the Canopy Openings in Mixed Beech–Fir Forest at Sentinel-2 Subpixel Level Using UAV and Machine Learning Approach |
| title_full_unstemmed | Mapping of the Canopy Openings in Mixed Beech–Fir Forest at Sentinel-2 Subpixel Level Using UAV and Machine Learning Approach |
| title_short | Mapping of the Canopy Openings in Mixed Beech–Fir Forest at Sentinel-2 Subpixel Level Using UAV and Machine Learning Approach |
| title_sort | mapping of the canopy openings in mixed beech fir forest at sentinel 2 subpixel level using uav and machine learning approach |
| topic | Sentinel-2 UAV DJI drone machine learning forest canopy canopy gaps |
| url | https://www.mdpi.com/2072-4292/12/23/3925 |
| work_keys_str_mv | AT ivanpilas mappingofthecanopyopeningsinmixedbeechfirforestatsentinel2subpixellevelusinguavandmachinelearningapproach AT mateogasparovic mappingofthecanopyopeningsinmixedbeechfirforestatsentinel2subpixellevelusinguavandmachinelearningapproach AT alannovkinic mappingofthecanopyopeningsinmixedbeechfirforestatsentinel2subpixellevelusinguavandmachinelearningapproach AT damirklobucar mappingofthecanopyopeningsinmixedbeechfirforestatsentinel2subpixellevelusinguavandmachinelearningapproach |