| 1. |
- Aksoy, Samet, et al.
(författare)
-
Forest Biophysical Parameter Estimation via Machine Learning and Neural Network Approaches
- 2023
-
Ingår i: IGARSS 2023 - 2023 IEEE International Geoscience and Remote Sensing Symposium. - : IEEE. - 9798350320107 - 9798350320091 - 9798350331745 ; , s. 2661-2664
-
Konferensbidrag (refereegranskat)abstract
- This paper presents the first results of the ongoing development of new forest mapping methods for the Swedish national forest mapping case using Airborne Laser Scanning (ALS) data, utilizing the recent findings in machine learning (ML) and Artificial Intelligence (AI) techniques. We used Random Forest (RF) and eXtreme Gradient Boosting (XGBoost) as ML models. In addition, Neural networks (NN) based approaches were utilized in this study. ALS derived features were used to estimate the stem volume (V), above-ground biomass (AGB), basal area (B), tree height (H), stem diameter (D), and forest stand age (A). XGBoost ML algorithm outperformed RF 1 % to 3 % in the R² metric. NN model performed similar to ML model, however it is superior in the estimation of V, AGB, and B parameters.
|
|
| 2. |
- Bohlin, Jonas, et al.
(författare)
-
A comparison of forest inventories based on aerial image matching and Airborne Laser Scanning data
- 2014
-
Konferensbidrag (övrigt vetenskapligt/konstnärligt)abstract
- Forest inventories are now commonly done by Airborne Laser Scanning (ALS), especially because many countries are collecting ALS data nation-wide to produce high quality elevation data. With an accurate digital elevation model, 3D data from aerial image matching could be a more cost-effective alternative to repeated ALS acquisitions for providing updated data to forest management planning in the future. This study aims at comparing the quality of forest inventory data obtained by aerial image matching and ALS. In the study area, a mixed boreal forest situated in central Sweden, aerial images from the national acquisition program with a ground sampling distance of 0.5 m and ALS data with a point density of 1.5-2 pulses per m2 from the national ALS production were available. The aerial images were matched with three different algorithms to assess possible differences in forest information content. Two hundred field plots, located within the study area, were utilized for non-parametric prediction of forest variables using random forest. Accuracy assessment was made by leave-one-out cross-validation at plot level. The results show similar accuracy of ALS and image matching-based predictions, with ALS slightly superior. Accuracy, in terms of root mean square errors in percent of surveyed plot mean, of ALS were: 6.4% for tree height; 12.5% for tree diameter; 18.2% for basal area and 20.0% for stem volume, and of image matching: 9.5% for tree height; 15.3% for tree diameter; 21.8% for basal area and 24.8% for stem volume. Among the image matching algorithms used, SURE was found to estimate the forest variables with best accuracies. However, the other algorithms produced similar results. These results indicate that inventory data acquired by matching of aerial images have a large potential for operational use in forest management planning as a cost-effective alternative to new ALS campaigns.
|
|
| 3. |
- Bohlin, Jonas, et al.
(författare)
-
Deciduous forest mapping using change detection of multi-temporal canopy height models from aerial images acquired at leaf-on and leaf-off conditions
- 2016
-
Ingår i: Scandinavian Journal of Forest Research. - : Informa UK Limited. - 0282-7581 .- 1651-1891. ; 31, s. 517-525
-
Tidskriftsartikel (refereegranskat)abstract
- Discrimination of deciduous trees using spectral information from aerial images has only been partly successfully due to the complexity of the reflectance at different view angles, times of acquisition, phenology of the trees and inter-tree radiance. Therefore, the objective was to evaluate the accuracy of estimating the proportion of deciduous stem volume (P) utilizing change detection between canopy height models (CHMs) generated by digital photogrammetry from leaf-on and leaf-off aerial images instead of using spectral information. The study was conducted at a hemiboreal study area in Sweden. Using aerial images from three seasons, CHMs with a resolution of approximately 0.5 m were generated using semi-global matching. For training plots, metrics describing the change between leaf-on and leaf-off conditions were calculated and used to model the continuous variable P, using the Random Forest approach. Validated at sub-stands, the estimation accuracy of P in terms of root mean square error and bias was found to be 18% and −6%, respectively. The overall classification accuracy, using four equally wide classes, was 83% with a kappa value of 0.68. The validation plots in classes of high proportion of coniferous or deciduous stem volume were well classified, whereas the mixed forest classes showed lower classification accuracies.
|
|
| 4. |
|
|
| 5. |
- Bohlin, Jonas, et al.
(författare)
-
Extraction of Spectral Information from Airborne 3D Data for Assessment of Tree Species Proportions
- 2021
-
Ingår i: Remote Sensing. - : MDPI AG. - 2072-4292. ; 13
-
Tidskriftsartikel (refereegranskat)abstract
- With the rapid development of photogrammetric software and accessible camera technology, land surveys and other mapping organizations now provide various point cloud and digital surface model products from aerial images, often including spectral information. In this study, methods for colouring the point cloud and the importance of different metrics were compared for tree species-specific estimates at a coniferous hemi-boreal test site in southern Sweden. A total of three different data sets of aerial image-based products and one multi-spectral lidar data set were used to estimate tree species-specific proportion and stem volume using an area-based approach. Metrics were calculated for 156 field plots (10 m radius) from point cloud data and used in a Random Forest analysis. Plot level accuracy was evaluated using leave-one-out cross-validation. The results showed small differences in estimation accuracy of species-specific variables between the colouring methods. Simple averages of the spectral metrics had the highest importance and using spectral data from two seasons improved species prediction, especially deciduous proportion. Best tree species-specific proportion was estimated using multi-spectral lidar with 0.22 root mean square error (RMSE) for pine, 0.22 for spruce and 0.16 for deciduous. Corresponding RMSE for aerial images was 0.24, 0.23 and 0.20 for pine, spruce and deciduous, respectively. For the species-specific stem volume at plot level using image data, the RMSE in percent of surveyed mean was 129% for pine, 60% for spruce and 118% for deciduous.
|
|
| 6. |
- Bohlin, Jonas, et al.
(författare)
-
Forest variable estimation using photogrammetric matching of digital aerial images in combination with a high-resolution DEM
- 2012
-
Ingår i: Scandinavian Journal of Forest Research. - : Informa UK Limited. - 0282-7581 .- 1651-1891. ; 27, s. 692-699
-
Tidskriftsartikel (refereegranskat)abstract
- The rapid development in aerial digital cameras in combination with the increased availability of high-resolution Digital Elevation Models (DEMs) provides a renaissance for photogrammetry in forest management planning. Tree height, stem volume, and basal area were estimated for forest stands using canopy height, density, and texture metrics derived from photogrammetric matching of digital aerial images and a high-resolution DEM. The study was conducted at a coniferous hemi-boreal site in southern Sweden. Three different data-sets of digital aerial images were used to test the effects of flight altitude and stereo overlap on an area-based estimation of forest variables. Metrics were calculated for 344 field plots (10 m radius) from point cloud data and used in regression analysis. Stand level accuracy was evaluated using leave-one-out cross validation of 24 stands. For these stands the tree height ranged from 4.8 to 26.9 m (17.8 m mean), stem volume 13.3 to 455 m3 ha-1 (250 m3 ha-1 mean), and basal area from 4.1 to 42.9 m2 ha-1 (27.1 m2 ha-1 mean) with mean stand size of 2.8 ha. The results showed small differences in estimation accuracy of forest variables between the data-sets. The data-set of digital aerial images corresponding to the standard acquisition of the Swedish National Land Survey (Lantma¨teriet), showed Root Mean Square Errors (in percent of the surveyed stand mean) of 8.8% for tree height, 13.1% for stem volume and 14.9% for basal area. The results imply that photogrammetric matching of digital aerial images has significant potential for operational use in forestry.
|
|
| 7. |
- Bohlin, Jonas, et al.
(författare)
-
Species-specific forest variable estimation using non-parametric modeling of multi-spectral photogrammetric point cloud data
- 2012
-
Konferensbidrag (övrigt vetenskapligt/konstnärligt)abstract
- The recent development in software for automatic photogrammetric processing of multispectral aerial imagery, and the growing nation-wide availability of Digital Elevation Model (DEM) data, are about to revolutionize data capture for forest management planning in Scandinavia. Using only already available aerial imagery and ALS-assessed DEM data, raster estimates of the forest variables mean tree height, basal area, total stem volume, and species-specific stem volumes were produced and evaluated. The study was conducted at a coniferous hemi-boreal test site in southern Sweden (lat. 58° N, long. 13° E). Digital aerial images from the Zeiss/Intergraph Digital Mapping Camera system were used to produce 3D point-cloud data with spectral information. Metrics were calculated for 696 field plots (10 m radius) from point-cloud data and used in k-MSN to estimate forest variables. For these stands, the tree height ranged from 1.4 to 33.0 m (18.1 m mean), stem volume from 0 to 829 m3 ha-1 (249 m3 ha-1 mean) and basal area from 0 to 62.2 m2 ha-1 (26.1 m2 ha-1 mean), with mean stand size of 2.8 ha. Estimates made using digital aerial images corresponding to the standard acquisition of the Swedish National Land Survey (Lantmäteriet) showed RMSEs (in percent of the surveyed stand mean) of 7.5% for tree height, 11.4% for basal area, 13.2% for total stem volume, 90.6% for pine stem volume, 26.4 for spruce stem volume, and 72.6% for deciduous stem volume. The results imply that photogrammetric matching of digital aerial images has significant potential for operational use in forestry.
|
|
| 8. |
|
|
| 9. |
- Chi, Jinshu, et al.
(författare)
-
The carbon balance of a managed boreal landscape measured from a tall tower in northern Sweden
- 2019
-
Ingår i: Agricultural and Forest Meteorology. - : Elsevier BV. - 0168-1923 .- 1873-2240. ; 274, s. 29-41
-
Tidskriftsartikel (refereegranskat)abstract
- Boreal forests exchange large amounts of carbon dioxide (CO2) with the atmosphere. A managed boreal landscape usually comprises various potential CO2 sinks and sources across forest stands of varying age classes, clear-cut areas, mires, and lakes. Due to this heterogeneity and complexity, large uncertainties exist regarding the net CO2 balance at the landscape scale. In this study, we present the first estimate of the net CO2 exchange over a managed boreal landscape (∼68 km2) in northern Sweden, based on tall tower eddy covariance measurements. Our results suggest that from March 1, 2016 to February 28, 2018, the heterogeneous landscape was a net CO2 sink with a 2-year mean uptake of −87 ± 6 g C m−2 yr−1. Due to an earlier and warmer spring and sunnier autumn, the landscape was a stronger CO2 sink during the first year (−122 ± 8 g C m−2) compared to the second year (−52 ± 9 g C m−2). Footprint analysis shows that 87% of the CO2 flux measurements originated from forests, whereas mires, clear-cuts, lakes, and grassland contributed 11%, 1%, 0.7%, and 0.2%, respectively. Altogether, the CO2 sink strength of the heterogeneous landscape was up to 38% lower compared to the sink strength of a mature stand surrounding the tower. Overall, this study suggests that the managed boreal landscape acted as a CO2 sink and advocates tall tower eddy covariance measurements to improve regional carbon budget estimates.
|
|
| 10. |
- Chi, Jinshu, et al.
(författare)
-
The Net Landscape Carbon Balance—Integrating terrestrial and aquatic carbon fluxes in a managed boreal forest landscape in Sweden
- 2020
-
Ingår i: Global Change Biology. - : Wiley. - 1354-1013 .- 1365-2486. ; 26:4, s. 2353-2367
-
Tidskriftsartikel (refereegranskat)abstract
- The boreal biome exchanges large amounts of carbon (C) and greenhouse gases (GHGs) with the atmosphere and thus significantly affects the global climate. A managed boreal landscape consists of various sinks and sources of carbon dioxide (CO2), methane (CH4), and dissolved organic and inorganic carbon (DOC and DIC) across forests, mires, lakes, and streams. Due to the spatial heterogeneity, large uncertainties exist regarding the net landscape carbon balance (NLCB). In this study, we compiled terrestrial and aquatic fluxes of CO2, CH4, DOC, DIC, and harvested C obtained from tall-tower eddy covariance measurements, stream monitoring, and remote sensing of biomass stocks for an entire boreal catchment (~68 km2) in Sweden to estimate the NLCB across the land–water–atmosphere continuum. Our results showed that this managed boreal forest landscape was a net C sink (NLCB = 39 g C m−2 year−1) with the landscape–atmosphere CO2 exchange being the dominant component, followed by the C export via harvest and streams. Accounting for the global warming potential of CH4, the landscape was a GHG sink of 237 g CO2-eq m−2 year−1, thus providing a climate-cooling effect. The CH4 flux contribution to the annual GHG budget increased from 0.6% during spring to 3.2% during winter. The aquatic C loss was most significant during spring contributing 8% to the annual NLCB. We further found that abiotic controls (e.g., air temperature and incoming radiation) regulated the temporal variability of the NLCB whereas land cover types (e.g., mire vs. forest) and management practices (e.g., clear-cutting) determined their spatial variability. Our study advocates the need for integrating terrestrial and aquatic fluxes at the landscape scale based on tall-tower eddy covariance measurements combined with biomass stock and stream monitoring to develop a holistic understanding of the NLCB of managed boreal forest landscapes and to better evaluate their potential for mitigating climate change.
|
|
