| 1. |
- Axelsson, Arvid, et al.
(författare)
-
Exploring Multispectral ALS Data for Tree Species Classification
- 2018
-
Ingår i: Remote Sensing. - : MDPI AG. - 2072-4292. ; 10
-
Tidskriftsartikel (refereegranskat)abstract
- Multispectral Airborne Laser Scanning (ALS) is a new technology and its output data have not been fully explored for tree species classification purposes. The objective of this study was to investigate what type of features from multispectral ALS data (wavelengths of 1550 nm, 1064 nm and 532 nm) are best suited for tree species classification. Remote sensing data were gathered over hemi-boreal forest in southern Sweden (58 degrees 2718.35N, 13 degrees 398.03E) on 21 July 2016. The field data consisted of 179 solitary trees from nine genera and ten species. Two new methods for feature extraction were tested and compared to features of height and intensity distributions. The features that were most important for tree species classification were intensity distribution features. Features from the upper part of the upper and outer parts of the crown were better for classification purposes than others. The best classification model was created using distribution features of both intensity and height in multispectral data, with a leave-one-out cross-validated accuracy of 76.5%. As a comparison, only structural features resulted in an highest accuracy of 43.0%. Picea abies and Pinus sylvestris had high user's and producer's accuracies and were not confused with any deciduous species. Tilia cordata was the deciduous species with a large sample that was most frequently confused with many other deciduous species. The results, although based on a small and special data set, suggest that multispectral ALS is a technology with great potential for tree species classification.
|
|
| 2. |
- Axelsson, Arvid
(författare)
-
Tree Species Classification : Analyzing Multitemporal Satellite Imagery and Multispectral Airborne Laser Scanning Data
- 2023
-
Doktorsavhandling (övrigt vetenskapligt/konstnärligt)abstract
- Tree species composition of forests affects the whole ecosystem and is part of the information needed for an efficient planning of forest management. This thesis explores how recent developments in remote sensing can provide more accurate tree species mapping. I try to answer the question of how the properties of these data can be used to derive more information on tree species. Out of the four papers in this thesis, two papers examine how multitemporal satellite imagery from the Sentinel-2 mission can be of use, and the other two papers investigate what properties of multispectral airborne laser scanning (MSALS) data that contain the most information on tree species. We applied a Bayesian method to multitemporal satellite imagery for tree species classification of pixels in the hemiboreal forest of Remningstorp in southwestern Sweden. The Bayesian method was applied to 142 Sentinel-2 images, and to a subset of images ranked and selected by the separability of tree species classes. The method was also compared to a Random Forest classifier for 45 Sentinel-2 images of boreal forest in mid-Sweden. The Bayesian method performed better for homogeneous tree species classes, while Random Forest performed better for heterogeneous classes. Data from two MSALS systems were used for classifying the tree species of individual trees. Optech Titan-X data were used to classify free-standing trees of nine species in Remningstorp. By using Riegl VQ-1560i-DW data, we performed a tree species classification in a more operational setting for three tree species in closed-canopy hemiboreal forest in Asa in southern Sweden. Multispectral intensity features provided a great improvement in classification accuracy in both cases, compared to using only structural features or combining them with monospectral intensity features. For Optech Titan-X, the green wavelength performed poorly, but for Riegl VQ-1560i-DW, the green wavelength provided the most information for separability, especially for birch (Betula spp.). There are two main conclusions in this thesis. The first is that Bayesian methods that updates probabilities as new observations are made provides an opportunity to automate the addition of satellite images for an updated classification. The second is that MSALS data provides more information on tree species than monospectral data and tree crown structure do, with the most information coming from the upper parts of the canopy. Nonetheless, what wavelengths of light that contribute most to tree species classification accuracy is highly dependent on what MSALS system that is used.
|
|
| 3. |
- Axelsson, Arvid, et al.
(författare)
-
Tree species classification using Sentinel-2 imagery and Bayesian inference
- 2021
-
Ingår i: International Journal of Applied Earth Observation and Geoinformation. - : Elsevier BV. - 1569-8432 .- 0303-2434. ; 100
-
Tidskriftsartikel (refereegranskat)abstract
- The increased temporal frequency of optical satellite data acquisitions provides a data stream that has the potential to improve land cover mapping, including mapping of tree species. However, for large area operational mapping, partial cloud cover and different image extents can pose challenges. Therefore, methods are needed to assimilate new images in a straightforward way without requiring a total spatial coverage for each new image. This study shows that Bayesian inference applied sequentially has the potential to solve this problem. To test Bayesian inference for tree species classification in the boreo-nemoral zone of southern Sweden, field data from the study area of Remningstorp (58°27′18.35″ N, 13°39′8.03″ E) were used. By updating class likelihood with an increasing number of combined Sentinel-2 images, a higher and more stable cross-validated overall accuracy was achieved. Based on a Mahalanobis distance, 23 images were automatically chosen from the period of 2016 to 2018 (from 142 images total). An overall accuracy of 87% (a Cohen’s kappa of 78.5%) was obtained for four tree species classes: Betula spp., Picea abies, Pinus sylvestris, and Quercus robur. This application of Bayesian inference in a boreo-nemoral forest suggests that it is a practical way to provide a high and stable classification accuracy. The method could be applied where data are not always complete for all areas. Furthermore, the method requires less reference data than if all images were used for classification simultaneously.
|
|
| 4. |
|
|
| 5. |
|
|
| 6. |
|
|
| 7. |
- Nilsson, Mats, et al.
(författare)
-
Data homogeneity impact in tree species classification based on Sentinel-2 multitemporal data case study in central Sweden
- 2024
-
Ingår i: International Journal of Remote Sensing. - 0143-1161 .- 1366-5901. ; 45, s. 5050-5075
-
Tidskriftsartikel (refereegranskat)abstract
- Spatial information on forest composition is invaluable for achieving scientific, ecological, and management objectives and for monitoring multiple changes in forest ecosystems. The increased flow of optical satellite data provides new opportunities to improve tree species mapping. However, the accuracy of such maps is affected by training data, and in particular on the homogeneity of individual classes. Thus, we evaluated the effect of data homogeneity in tree species classification. We performed tree species classification by considering different ways to partition data into tree species classes. The class sets considered were (i) only mixed coniferous and mixed deciduous forest classes, (ii) single-species classes, (iii) single-species, mixed coniferous and mixed deciduous classes, and (iv) single-species, mixed coniferous and mixed deciduous classes and a true mixed class. Using data from the Swedish National Forest Inventory, we varied the threshold that defined dominating species. Tree species were classified for a study area in central Sweden using Sentinel-2 data and two classification approaches: Bayesian inference and random forest (RF). Images were selected by class separability and the most informative images based on variable selection with RF. The most informative images tended to be selected by both methods. However, in forests with tree species of similar spectral behaviour, image selection on the basis of class separability was found to be more reliable. More accurate classification results were achieved as the number of classes decreased and the threshold of plot purity increased. The Bayesian classification approach of only mixed coniferous and mixed deciduous classes gave the highest OA, always greater than 90%. When discriminating between pure plots of Birch (Betula spp.), Spruce (Picea abies), Scots pine (Pinus sylvestris) and Lodgepole pine (Pinus contorta), the best OA values were 84% for Bayesian and 80% for RF. In more complicated scenarios, RF resulted in higher overall accuracies (OA).
|
|
