Airborne Lidar Sampling Strategies to Enhance Forest Aboveground Biomass Estimation from Landsat Imagery
Accurately estimating aboveground biomass (AGB) is important in many applications, including monitoring carbon stocks, investigating deforestation and forest degradation, and designing sustainable forest management strategies. Although lidar provides critical three-dimensional forest structure infor...
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2019-08-01
|
| Series: | Remote Sensing |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2072-4292/11/16/1906 |
| _version_ | 1818959276755910656 |
|---|---|
| author | Siqi Li Lindi J. Quackenbush Jungho Im |
| author_facet | Siqi Li Lindi J. Quackenbush Jungho Im |
| author_sort | Siqi Li |
| collection | DOAJ |
| description | Accurately estimating aboveground biomass (AGB) is important in many applications, including monitoring carbon stocks, investigating deforestation and forest degradation, and designing sustainable forest management strategies. Although lidar provides critical three-dimensional forest structure information for estimating AGB, acquiring comprehensive lidar coverage is often cost prohibitive. This research focused on developing a lidar sampling framework to support AGB estimation from Landsat images. Two sampling strategies, systematic and classification-based, were tested and compared. The proposed strategies were implemented over a temperate forest study site in northern New York State and the processes were then validated at a similar site located in central New York State. Our results demonstrated that while the inclusion of lidar data using systematic or classification-based sampling supports AGB estimation, the systematic sampling selection method was highly dependent on site conditions and had higher accuracy variability. Of the 12 systematic sampling plans, R<sup>2</sup> values ranged from 0.14 to 0.41 and plot root mean square error (RMSE) ranged from 84.2 to 93.9 Mg ha<sup>−1</sup>. The classification-based sampling outperformed 75% of the systematic sampling strategies at the primary site with R<sup>2</sup> of 0.26 and RMSE of 70.1 Mg ha<sup>−1</sup>. The classification-based lidar sampling strategy was relatively easy to apply and was readily transferable to a new study site. Adopting this method at the validation site, the classification-based sampling also worked effectively, with an R<sup>2</sup> of 0.40 and an RMSE of 108.2 Mg ha<sup>−1</sup> compared to the full lidar coverage model with an R<sup>2</sup> of 0.58 and an RMSE of 96.0 Mg ha<sup>−1</sup>. This study evaluated different lidar sample selection methods to identify an efficient and effective approach to reduce the volume and cost of lidar acquisitions. The forest type classification-based sampling method described in this study could facilitate cost-effective lidar data collection in future studies. |
| first_indexed | 2024-12-20T11:39:04Z |
| format | Article |
| id | doaj.art-227d57b0729743ff9a8c9f184de71beb |
| institution | Directory Open Access Journal |
| issn | 2072-4292 |
| language | English |
| last_indexed | 2024-12-20T11:39:04Z |
| publishDate | 2019-08-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Remote Sensing |
| spelling | doaj.art-227d57b0729743ff9a8c9f184de71beb2022-12-21T19:42:02ZengMDPI AGRemote Sensing2072-42922019-08-011116190610.3390/rs11161906rs11161906Airborne Lidar Sampling Strategies to Enhance Forest Aboveground Biomass Estimation from Landsat ImagerySiqi Li0Lindi J. Quackenbush1Jungho Im2Department of Environmental Resources Engineering, State University of New York College of Environmental Science and Forestry, Syracuse, NY 13210, USADepartment of Environmental Resources Engineering, State University of New York College of Environmental Science and Forestry, Syracuse, NY 13210, USASchool of Urban and Environmental Engineering, Ulsan National Institute of Science and Technology, Ulsan 44919, KoreaAccurately estimating aboveground biomass (AGB) is important in many applications, including monitoring carbon stocks, investigating deforestation and forest degradation, and designing sustainable forest management strategies. Although lidar provides critical three-dimensional forest structure information for estimating AGB, acquiring comprehensive lidar coverage is often cost prohibitive. This research focused on developing a lidar sampling framework to support AGB estimation from Landsat images. Two sampling strategies, systematic and classification-based, were tested and compared. The proposed strategies were implemented over a temperate forest study site in northern New York State and the processes were then validated at a similar site located in central New York State. Our results demonstrated that while the inclusion of lidar data using systematic or classification-based sampling supports AGB estimation, the systematic sampling selection method was highly dependent on site conditions and had higher accuracy variability. Of the 12 systematic sampling plans, R<sup>2</sup> values ranged from 0.14 to 0.41 and plot root mean square error (RMSE) ranged from 84.2 to 93.9 Mg ha<sup>−1</sup>. The classification-based sampling outperformed 75% of the systematic sampling strategies at the primary site with R<sup>2</sup> of 0.26 and RMSE of 70.1 Mg ha<sup>−1</sup>. The classification-based lidar sampling strategy was relatively easy to apply and was readily transferable to a new study site. Adopting this method at the validation site, the classification-based sampling also worked effectively, with an R<sup>2</sup> of 0.40 and an RMSE of 108.2 Mg ha<sup>−1</sup> compared to the full lidar coverage model with an R<sup>2</sup> of 0.58 and an RMSE of 96.0 Mg ha<sup>−1</sup>. This study evaluated different lidar sample selection methods to identify an efficient and effective approach to reduce the volume and cost of lidar acquisitions. The forest type classification-based sampling method described in this study could facilitate cost-effective lidar data collection in future studies.https://www.mdpi.com/2072-4292/11/16/1906systematic samplingclassification-based samplingforest typesdata fusionregressionrandom forest |
| spellingShingle | Siqi Li Lindi J. Quackenbush Jungho Im Airborne Lidar Sampling Strategies to Enhance Forest Aboveground Biomass Estimation from Landsat Imagery Remote Sensing systematic sampling classification-based sampling forest types data fusion regression random forest |
| title | Airborne Lidar Sampling Strategies to Enhance Forest Aboveground Biomass Estimation from Landsat Imagery |
| title_full | Airborne Lidar Sampling Strategies to Enhance Forest Aboveground Biomass Estimation from Landsat Imagery |
| title_fullStr | Airborne Lidar Sampling Strategies to Enhance Forest Aboveground Biomass Estimation from Landsat Imagery |
| title_full_unstemmed | Airborne Lidar Sampling Strategies to Enhance Forest Aboveground Biomass Estimation from Landsat Imagery |
| title_short | Airborne Lidar Sampling Strategies to Enhance Forest Aboveground Biomass Estimation from Landsat Imagery |
| title_sort | airborne lidar sampling strategies to enhance forest aboveground biomass estimation from landsat imagery |
| topic | systematic sampling classification-based sampling forest types data fusion regression random forest |
| url | https://www.mdpi.com/2072-4292/11/16/1906 |
| work_keys_str_mv | AT siqili airbornelidarsamplingstrategiestoenhanceforestabovegroundbiomassestimationfromlandsatimagery AT lindijquackenbush airbornelidarsamplingstrategiestoenhanceforestabovegroundbiomassestimationfromlandsatimagery AT junghoim airbornelidarsamplingstrategiestoenhanceforestabovegroundbiomassestimationfromlandsatimagery |