Assessment of Classifiers and Remote Sensing Features of Hyperspectral Imagery and Stereo-Photogrammetric Point Clouds for Recognition of Tree Species in a Forest Area of High Species Diversity

The monitoring of tree species diversity is important for forest or wetland ecosystem service maintenance or resource management. Remote sensing is an efficient alternative to traditional field work to map tree species diversity over large areas. Previous studies have used light detection and ranging (LiDAR) and imaging spectroscopy (hyperspectral or multispectral remote sensing) for species richness prediction. The recent development of very high spatial resolution (VHR) RGB images has enabled detailed characterization of canopies and forest structures. In this study, we developed a three-step workflow for mapping tree species diversity, the aim of which was to increase knowledge of tree species diversity assessment using deep learning in a tropical wetland (Haizhu Wetland) in South China based on VHR-RGB images and LiDAR points. Firstly, individual trees were detected based on a canopy height model (CHM, derived from LiDAR points) by the local-maxima-based method in the FUSION software (Version 3.70, Seattle, USA). Then, tree species at the individual tree level were identified via a patch-based image input method, which cropped the RGB images into small patches (the individually detected trees) based on the tree apexes detected. Three different deep learning methods (i.e., AlexNet, VGG16, and ResNet50) were modified to classify the tree species, as they can make good use of the spatial context information. Finally, four diversity indices, namely, the Margalef richness index, the Shannon–Wiener diversity index, the Simpson diversity index, and the Pielou evenness index, were calculated from the fixed subset with a size of 30 × 30 m for assessment. In the classification phase, VGG16 had the best performance, with an overall accuracy of 73.25% for 18 tree species. Based on the classification results, mapping of tree species diversity showed reasonable agreement with field survey data (R2Margalef = 0.4562, root-mean-square error RMSEMargalef = 0.5629; R2Shannon–Wiener = 0.7948, RMSEShannon–Wiener = 0.7202; R2Simpson = 0.7907, RMSESimpson = 0.1038; and R2Pielou = 0.5875, RMSEPielou = 0.3053). While challenges remain for individual tree detection and species classification, the deep-learning-based solution shows potential for mapping tree species diversity.

Download Full-text

Discerning Effects of Climate Variability, Local Land Degradation and Tree Species Diversity on Remote Sensing Derived Resilience and Resistance of Dry Afromontane Forest

10.21203/rs.3.rs-91187/v1 ◽

2020 ◽

Author(s):

Hadgu Hishe ◽

Louis Oosterlynck ◽

Kidane Giday ◽

Wanda De Keersmaecker ◽

Ben Somers ◽

...

Keyword(s):

Remote Sensing ◽

Time Series ◽

Species Diversity ◽

Tree Species ◽

Tree Diversity ◽

Tree Species Diversity ◽

Factors Associated ◽

Afromontane Forest ◽

Growth Stability ◽

Dry Afromontane Forest

Abstract Introduction: Anthropogenic disturbances are increasingly affecting the vitality of tropical dry forests. The future condition of this important biome will depend on its capability to resist, and recover from these disturbances. So far, the temporal stability of dryland forests is rarely studied, but could serve as a basis for forest management and restoration. Methodology: In a degraded dry Afromontane forest in northern Ethiopia, we explored remote sensing derived indicators of forest stability, using MODIS satellite derived NDVI time series from 2001 to 2018. Resilience, resistance and variability were measured using the anomalies (remainders) after time series decomposition into seasonality, trend and remainder components. Growth stability was calculated using the integral of the undecomposed NDVI data. These NDVI derived stability indicators were then related to environmental factors of climate, topography, soil, tree species diversity, and disturbance, obtained from a systematic grid of field inventory plots, using boosted regression trees in R. Resilience and resistance were adequately predicted by these factors with an R2 of 0.67 and 0.48, respectively, but the models for variability and growth stability were weaker. Precipitation of the wettest month, distance from settlements and slope were the most important factors associated with resilience, explaining 51% of the effect. Altitude, temperature seasonality and humus accumulation were the significant factors associated with the resistance of the forest, explaining 61% of the overall effect. A positive effect of tree diversity on resilience was also significant, except that the impact of species evenness declined above a threshold value of 0.70, indicating that perfect evenness reduced the resilience of the forest. Conclusion: A combination of climate, topographic variables and disturbance indicators controlled the stability of the dry forest. Tree diversity is an important component that should be considered in the management and restoration programs of such degraded forests. If local disturbances are alleviated the recovery time of dryland forests could be shortened, which is vital to maintain the ecosystem services these forests provide to local communities and global climate change.

Download Full-text

Remote-sensing support for the Estonian National Forest Inventory, facilitating the construction of maps for forest height, standing-wood volume, and tree species composition

Forestry Studies / Metsanduslikud Uurimused ◽

10.2478/fsmu-2020-0016 ◽

2020 ◽

Vol 73 (1) ◽

pp. 77-97

Author(s):

Mait Lang ◽

Allan Sims ◽

Kalev Pärna ◽

Raul Kangro ◽

Märt Möls ◽

...

Keyword(s):

Remote Sensing ◽

Species Composition ◽

Tree Species ◽

Forest Inventory ◽

Point Clouds ◽

National Forest Inventory ◽

National Forest ◽

Tree Species Composition ◽

Wood Volume ◽

Forest Height

Abstract Since 1999, Estonia has conducted the National Forest Inventory (NFI) on the basis of sample plots. This paper presents a new module, incorporating remote-sensing feature variables from airborne laser scanning (ALS) and from multispectral satellite images, for the construction of maps of forest height, standing-wood volume, and tree species composition for the entire country. The models for sparse ALS point clouds yield coefficients of determination of 89.5–94.8% for stand height and 84.2–91.7% for wood volume. For the tree species prediction, the models yield Cohen's kappa values (taking 95% confidence intervals) of 0.69–0.72 upon comparing model results against a previous map, and values of 0.51–0.54 upon comparing model results against NFI sample plots. This paper additionally examines the influence of foliage phenology on the predictions and discusses options for further enhancement of the system.

Download Full-text

Season-dependence of remote sensing indicators of tree species diversity

Remote Sensing Letters ◽

10.1080/2150704x.2014.912767 ◽

2014 ◽

Vol 5 (5) ◽

pp. 404-412 ◽

Cited By ~ 10

Author(s):

Eduardo Eiji Maeda ◽

Janne Heiskanen ◽

Koen W. Thijs ◽

Petri K.E. Pellikka

Keyword(s):

Remote Sensing ◽

Species Diversity ◽

Tree Species ◽

Tree Species Diversity

Download Full-text

Tree species diversity mapping using UAS-based digital aerial photogrammetry point clouds and multispectral imageries in a subtropical forest invaded by moso bamboo (Phyllostachys edulis)

International Journal of Applied Earth Observation and Geoinformation ◽

10.1016/j.jag.2021.102587 ◽

2021 ◽

Vol 104 ◽

pp. 102587

Author(s):

Kun Liu ◽

Anke Wang ◽

Sujiong Zhang ◽

Zhenxian Zhu ◽

Yufang Bi ◽

...

Keyword(s):

Species Diversity ◽

Tree Species ◽

Point Clouds ◽

Subtropical Forest ◽

Moso Bamboo ◽

Tree Species Diversity ◽

Phyllostachys Edulis ◽

Aerial Photogrammetry

Download Full-text

TREE SPECIES DIVERSITY, STRUCTURAL CHARACTERISTICS AND CARBON STOCK IN A ONE-HECTARE PLOT OF THE PROTECTION FOREST AREA IN WEST LAMPUNG REGENCY, INDONESIA

REINWARDTIA ◽

10.14203/reinwardtia.v18i1.3574 ◽

2019 ◽

Vol 18 (1) ◽

Author(s):

Nur Muhammad Heriyanto ◽

Ismayadi Samsoedin ◽

Kuswata Kartawinata

Keyword(s):

Species Diversity ◽

Tree Species ◽

Carbon Stock ◽

Dominant Species ◽

Structural Characteristics ◽

Floristic Composition ◽

Montane Forest ◽

Forest Area ◽

Tree Species Diversity ◽

Protection Forest

HERIYANTO, N. M., SAMSOEDIN, I. & KARTAWINATA, K. 2018. Tree species diversity, structural characteristics and carbon stock in a one-hectare plot of the protection forest area in West Lampung Regency, Indonesia. Reinwardtia 18(1): 1‒18. — A study of species composition, structure and carbon stock in the lower montane forest in the Register 45B of the protection forest area in the Tri Budi Syukur District, Kebun Tebu Village, West Lampung Regency, Lampung Province was conducted in September 2016. The objective of the study was to undertake quantified measurements of floristic composition and structure of and carbon storage in the lower montane forest at 965 m asl in the protection forest area. A one hectare plot (100 m × 100 m) was established randomly. The plot was further divided into 25 subplots of 20 m × 20 m each to record trees. Quadrats of 5 m × 5 m for saplings and subquadrats of 2 m × 2 m for seedlings were nested in the tree subplots. We recorded 247 trees with diameter at breast height ≥ 10 cm representing 25 species and 19 families, with a total basal area of 59.14 m2. Overall including seedlings and saplings we recorded 31 species. The species richness was very low due to disturbances, and was the lowest compared to that of other forests in Sumatra, Kalimantan and Java. The dominant species in terms of importance values (IV) were Litsea cf. fulva (IV=77.02), Lithocarpus reinwardtii (IV=45.21) and Altingia excelsa (IV=26.95). Dominant species in seedling and sapling stages were Polyalthia lateriflora (IV=27.54) and Memecylon multiflorum (IV=41.58). Biomass and carbon stock of trees with DBH ≥ 10 cm was 50.87 ton/ha and 25.43 ton C/ha, respectively. Regeneration was poor. Structurally and floristically the forest was a developing disturbed forest and the composition will remain unchanged in many years to come. The successions leading to terminal communities similar to the original conditions would be very slow and should be assisted and enhanced by applying ecological restoration through planting tree species native to the site.

Download Full-text

Short Communication: Trees species diversity and indicator species in Bedugul forest ecosystem, Bali, Indonesia

Biodiversitas Journal of Biological Diversity ◽

10.13057/biodiv/d190629 ◽

2018 ◽

Vol 19 (6) ◽

pp. 2213-2218

Author(s):

SUTOMO SUTOMO ◽

I DEWA PUTU DARMA ◽

ARIEF PRIYADI ◽

RAJIF IRYADI

Keyword(s):

Species Diversity ◽

Indicator Species ◽

Forest Ecosystem ◽

Tree Species ◽

Short Communication ◽

Forest Area ◽

Ratio Analysis ◽

Vegetation Diversity ◽

Endorheic Basin ◽

Steep Slopes

Sutomo, Darma IDP, Priyadi A, Iryadi R. 2018. Trees species diversity and indicator species in Bedugul forest ecosystem, Bali, Indonesia. Biodiversitas 19: 2213-2218. Bedugul area is an endorheic basin landscape with 3 lakes namely, Beratan, Buyan, Tamblingan, which is surrounded by Bukit Mangu, Tapak and Lesung. Topography of the area shows sloping to steep slopes with altitude on the lake surface ± 1,100 m asl and the highest peak of Bukit Mangu 2002 m asl. Ecological studies have not been optimal so identification of comprehensive ecological potential is carried out. Measurement of tree vegetation diversity was carried out by Centered Quarter Method and important value ratio analysis and location elevation class. The results of the inventory of tree species diversity in the Bedugul Bali forest area recorded 35 species and 13 indicator tree species. From the number of indicator tree species in the Mangu hill forest area there are 5 types of Ficus sp, Platea latifolia, Polyosma integrifolia, Lindera sp. and Syzygium sp., Bukit Tapak forest area consists of 4 species, Casuarina junghuhniana, Acronychia trifoliate, Astronia spectabilis and Homalanthus giganteus, the forest area of Bukit Lesung consists of 4 types of Lophopetalum javanicum, Syzygium racemosum, Dysoxylum nutans and Dendrocnide peltata.

Download Full-text

Relationships between tree species diversity and above-ground biomass in Central African rainforests: implications for REDD

Environmental Conservation ◽

10.1017/s0376892913000295 ◽

2013 ◽

Vol 41 (1) ◽

pp. 64-72 ◽

Cited By ~ 35

Author(s):

MICHAEL DAY ◽

CRISTINA BALDAUF ◽

ERVAN RUTISHAUSER ◽

TERRY C. H. SUNDERLAND

Keyword(s):

Species Diversity ◽

Biodiversity Conservation ◽

Tropical Forests ◽

Tree Species ◽

Tree Species Diversity ◽

Above Ground Biomass ◽

Ground Biomass ◽

High Species Diversity ◽

The Relationship ◽

Impacts Of Climate Change

SUMMARYTropical forests are both important stores of carbon and among the most biodiverse ecosystems on the planet. Reducing emissions from deforestation and degradation (REDD) schemes are designed to mitigate the impacts of climate change, by conserving tropical forests threatened by deforestation or degradation. REDD schemes also have the potential to contribute significantly to biodiversity conservation efforts within tropical forests, however biodiversity conservation and carbon sequestration need to be aligned more closely for this potential to be realized. This paper analyses the relationship between tree species diversity and above-ground biomass (AGB) derived from 1-ha tree plots in Central African rainforests. There was a weakly significant correlation between tree biomass and tree species diversity (r = 0.21, p = 0.03), and a significantly higher mean species diversity in plots with larger AGB estimates (M = 44.38 species in the top eight plots, compared to M = 35.22 in the lower eight plots). In these Central African plots, the relationship between tree species diversity and AGB appeared to be highly variable; nonetheless, high species diversity may often be related to higher biomass and, in such cases, REDD schemes may enhance biodiversity by targeting species diverse forests.

Download Full-text

Method of Estimating Degraded Forest Area: Cases from Dominant Tree Species from Guangdong and Tibet in China

Forests ◽

10.3390/f11090930 ◽

2020 ◽

Vol 11 (9) ◽

pp. 930 ◽

Cited By ~ 1

Author(s):

Biyun Wu ◽

Xiang Meng ◽

Qiaolin Ye ◽

Ram P. Sharma ◽

Guangshuang Duan ◽

...

Keyword(s):

Remote Sensing ◽

Tree Species ◽

Guangdong Province ◽

Forest Degradation ◽

Forest Area ◽

Forest Site ◽

Degraded Forest ◽

Autonomous Region ◽

Tibet Autonomous Region ◽

Total Forest Area

Forest degradation has been considered as one of the main causes of climate change in recent years. The knowledge of estimating degraded forest areas without the application of remote sensing tools can be useful in finding solutions to resolve degradation problems through appropriate restoration methods. Using the existing knowledge through literature review and field-based primary information, we generated new knowledge by combining the information obtained from multi-criteria decision analyses with an analytic hierarchy process, and this was then used to estimate degraded forest area. Estimation involves determining forest degradation index (FDI) and degradation threshold. Continuous inventory data of permanent sample plots collected from degraded forests, consisting of various forest types divided by dominant tree species in the Guangdong province and Tibet autonomous region of China, were used for the purposes. We identified four different forest degradation levels through the determination and comprehensive evaluation of FDI. The degraded forest area with broad-leaved species as dominant tree species in the Guangdong province was estimated to be 83.3% of a total forest area of 24,037 km2. In the same province, the degraded forest area with eucalyptus as a dominant tree species was 59.5% of a total forest area of 18,665 km2. In the Tibet autonomous region, the degraded forest area with spruce as a dominant tree species was 99.1% of a total forest area of 17,614 km2, and with fir as a dominant tree species, the degraded area was 98.4% of a forest area of 12,103 km2. A sampling accuracy of forest areas with national forest inventory was about 95% in both provinces. Our study concludes that the FDI method used has a certain scientific rationality in estimating degraded forest area. The forest provides a variety of tangible and intangible goods and services for humans. Therefore, forest management should focus on the improvement of its overall productivity, which is only possible with improving forest site quality. One of the important steps to improve the quality of a forest site is to resolve its degradation issues. The presented method in this article will be useful in finding the solutions to forest degradation problems. This method, which does not need any remote sensing tool, is simple and can be easily applied for estimating any degraded forest area and developing effective forest restoration plans.

Download Full-text

TREE SPECIES CLASSIFICATION BASED ON 3D SPECTRAL POINT CLOUDS AND ORTHOMOSAICS ACQUIRED BY SNAPSHOT HYPERSPECTRAL UAS SENSOR

ISPRS - International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences ◽

10.5194/isprs-archives-xlii-2-w13-379-2019 ◽

2019 ◽

Vol XLII-2/W13 ◽

pp. 379-384 ◽

Cited By ~ 2

Author(s):

C. Iseli ◽

A. Lucieer

Keyword(s):

Tree Species ◽

Classification Accuracy ◽

Point Cloud ◽

Hyperspectral Imagery ◽

Spectral Feature ◽

Point Clouds ◽

Sensor Calibration ◽

Unmanned Aerial Systems ◽

3D Point Clouds

Abstract. In recent years, there has been a growing number of small hyperspectral sensors suitable for deployment on unmanned aerial systems (UAS. The introduction of the hyperspectral snapshot sensor provides interesting opportunities for acquisition of three-dimensional (3D) hyperspectral point clouds based on the structure-from-motion (SfM) workflow. In this study, we describe the integration of a 25-band hyperspectral snapshot sensor (PhotonFocus camera with IMEC 600&thinsp;&ndash;&thinsp;875&thinsp;nm 5x5 mosaic chip) on a multi-rotor UAS. The sensor was integrated with a dual frequency GNSS receiver for accurate time synchronisation and geolocation. We describe the sensor calibration workflow, including dark current and flat field characterisation. An SfM workflow was implemented to derive hyperspectral 3D point clouds and orthomosaics from overlapping frames. On-board GNSS coordinates for each hyperspectral frame assisted in the SfM process and allowed for accurate direct georeferencing (&lt;&thinsp;10&thinsp;cm absolute accuracy). We present the processing workflow to generate seamless hyperspectral orthomosaics from hundreds of raw images. Spectral reference panels and in-field spectral measurements were used to calibrate and validate the spectral signatures. This process provides a novel data type which contains both 3D, geometric structure and detailed spectral information in a single format. First, to determine the potential improvements that such a format could provide, the core aim of this study was to compare the use of 3D hyperspectral point clouds to conventional hyperspectral imagery in the classification of two Eucalyptus tree species found in Tasmania, Australia. The IMEC SM5x5 hyperspectral snapshot sensor was flown over a small native plantation plot, consisting of a mix of the Eucalyptus pauciflora and E. tenuiramis species. High overlap hyperspectral imagery was captured and then processed using SfM algorithms to generate both a hyperspectral orthomosaic and a dense hyperspectral point cloud. Additionally, to ensure the optimum spectral quality of the data, the characteristics of the hyperspectral snapshot imaging sensor were analysed utilising measurements captured in a laboratory environment. To coincide with the generated hyperspectral point cloud data, both a file format and additional processing and visualisation software were developed to provide the necessary tools for a complete classification workflow. Results based on the classification of the E. pauciflora and E. tenuiramis species revealed that the hyperspectral point cloud produced an increased classification accuracy over conventional hyperspectral imagery based on random forest classification. This was represented by an increase in classification accuracy from 67.2% to 73.8%. It was found that even when applied separately, the geometric and spectral feature sets from the point cloud both provided increased classification accuracy over the hyperspectral imagery.

Download Full-text