Next Article in Journal
Post-Processing Air Temperature Weather Forecast Using Artificial Neural Networks with Measurements from Meteorological Stations
Previous Article in Journal
Applications of Stretching Technique and Time Window Effects on Ultrasonic Velocity Monitoring in Concrete
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Applied Sciences
Volume 12
Issue 14
10.3390/app12147125
applsci-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Communication

Countrywide Mapping of Plant Ecological Communities with 101 Legends including Land Cover Types for the First Time at 10 m Resolution through Convolutional Learning of Satellite Images

by
Ram C. Sharma
Department of Informatics, Tokyo University of Information Sciences, 4-1 Onaridai, Wakaba-ku, Chiba 265-8501, Japan
Appl. Sci. 2022, 12(14), 7125; https://doi.org/10.3390/app12147125
Submission received: 4 June 2022 / Revised: 2 July 2022 / Accepted: 12 July 2022 / Published: 14 July 2022
(This article belongs to the Special Issue Ecological Applications of Remote Sensing and Machine/Deep Learning Techniques)
Browse Figures
Versions Notes

Abstract

:
This paper presents next-generation mapping of plant ecological communities including land cover and agricultural types at 10 m spatial resolution countrywide. This research introduces modelling and mapping of land cover and ecological communities separately in small regions-of-interest (prefecture level), and later integrating the outputs into a large scale (country level) for dealing with regional distribution characteristics of plant ecological communities effectively. The Sentinel-2 satellite images were processed for cloud masking and half-monthly median composite images consisting of ten multi-spectral bands and seven spectral indexes were generated. The reliable ground truth data were prepared from extant multi-source survey databases through the procedure of stratified sampling, cross-checking, and noisy-labels pruning. Deep convolutional learning of the time-series of the satellite data was employed for prefecture-wise classification and mapping of 29–62 classes. The classification accuracy computed with the 10-fold cross-validation method varied from 71.1–87.5% in terms of F1-score and 70.9–87.4% in terms of Kappa coefficient across 48 prefectural regions. This research produced seamless maps of 101 ecological communities including land cover and agricultural types for the first time at a country scale with an average accuracy of 80.5% F1-score.

1. Introduction

Amplification of anthropogenic activities such as land use changes and overexploitation of natural resources has disturbed terrestrial ecosystems worldwide [1,2,3]. Climate change has further intensified such disturbances in many vulnerable ecosystems [4,5,6]. Land cover and vegetation mapping is an essential step in the conservation and management of biodiversity and ecosystem services to understand interrelationships between ecosystems and environmental changes caused by anthropogenic or natural drivers [7,8,9,10].
Earth-observing satellites are suitable technology for monitoring the extent, magnitude, and patterns of ecosystems as well as fragmentation and connectivity of habitats in a consistent and seamless manner [11,12,13,14]. High spatial and spectral resolution images are freely available from an increasing number of earth observation missions [15]. This offers new opportunities for assessing land cover and vegetation conditions across large areas compared with the traditional field-based methods [16,17]. Satellite remote sensing, with a range of spatial, spectral, temporal, and radiometric characteristics, is an effective technology for classification and mapping of vegetation types, particularly at a broad scale [18,19,20].
Researchers have employed different approaches such as supervised classification or unsupervised segmentation and labelling on multi-spectral or hyper-spectral imagery captured from aerial or space platforms for land cover and vegetation mapping [21,22,23,24,25,26]. Nevertheless, machine learning of satellite images with a small set of labelled data and prediction for unlabelled data (supervised approach) has been a popular method for land cover and vegetation mapping in recent years [27,28,29,30,31]. A number of machine learning classifiers, such as Random Forests [32], Support Vector Machines [33], and Gradient Boosted Decision Trees [34] has been utilized effectively for the mapping of a variety of vegetation types such as wetland vegetation [35,36,37], urban vegetation [38,39], tree species and forest types [40,41], and semi-arid vegetation [42,43,44], to give some examples. In recent years, applications of deep learning neural networks have been gaining momentum in the classification and mapping of land cover and vegetation types. For this purpose, model architectures made up of neural-network layers such as multilayer perceptron [45,46], convolutional [47,48], recurrent [49,50], convolutional-recurrent [51,52], and attention-recurrent [53,54] have been employed and shown outstanding performance. A review of the existing research and classification methods shows that most of the previous studies were implemented for local-scale classification and mapping of land cover and vegetation types. Moreover, the recent trend seems to be supervised classification with deep learning neural networks for dealing with the large volume of high-resolution satellite images required for large-scale applications without compromising performance.
Land cover and vegetation mapping include hierarchical organization of land cover and vegetation types with some criteria and delineation of them in a geographic environment [55]. Though some broad categories of vegetation types such as forests, shrublands, grasslands, and cultivated lands, etc. have been incorporated by land use and land cover maps [56,57] and further subdivision into physiognomic categories such as evergreen/deciduous conifer/broadleaf forests [58], vegetation mapping can be described as a specialized field that is mostly focused on the mapping of plant communities. However, defining a plant community, as a relatively uniform collection of plant species distinguishable from its neighboring patches [59], is not straightforward and researchers have attempted different systems for the organization of vegetation types with the support of different criteria such as bioclimate [60,61,62], ecosystem [63], physiognomy [64,65], phytosociological association [66,67], and dominant species [68]. Out of these criteria, bioclimatic, ecosystem, and physiognomic criteria are too broad and are only applicable for classifying vegetation types at coarse scales; whereas criteria of dominant species may be impractical for large-scale applications, especially in areas with mixed vegetation, and the characteristic species-based phytosociological association system may be inaccurately detected by remote sensing-based applications, particularly with high-resolution images.
More recently, the Genus-Physiognomy-Ecosystem (GPE) classification system was developed by Sharma [69] for community-level vegetation classification at a large scale using high-spatial-resolution remote sensing images. In this system, community-level vegetation types are defined with the inference of dominant genus, physiognomy, and shared ecological conditions (Ecosystem). For example, Quercus Evergreen Broadleaf Forests, Quercus Deciduous Broadleaf Forests, Quercus Shrub, Alpine Herb, Coastal Shrub, Wetland Herb, etc. In the current research, the GPE classification system has been redefined as the Dominant Genus-Physiognomy-Ecological (DGPE) classification system and the resulting plant (ecological) community units are denoted as DGPE communities. Though previous study revealed the potential of the DGPE system for the classification of plant ecological communities at a regional scale using Landsat 8 multi-temporal images, it is still necessary to evaluate effectiveness of the DGPE system for seamless and operational mapping at a country level. In addition, from the previous study [69], the regional distribution characteristics of the community-level vegetation types and regional variations of the satellite data were understood. Unlike the broad categories of land cover types, the community-level vegetation types found in one prefecture may not be found in other prefectures. The satellite signals are affected by topographic and climatic variations across regions. The insights gained from the previous study drive the prefecture-wise mapping of plant ecological communities in the current research. Whereas the previous study was merely limited to the classification of plant ecological communities at a regional scale, the current research implements seamless and operational mapping countrywide by adopting the classification system developed in the previous study. The major objective of this research is to produce DGPE community maps including land cover and agricultural types (water, barren, built-up, solar panels, upland field, paddy field, cultivated pasture, and orchard) at 10 m spatial resolution at a country scale by harnessing Sentinel-2 multi-temporal images.

2. Materials and Methods

2.1. Study Area

The study area covers the whole country (Japan) comprising 47 prefectures. This research introduced prefectural region-wise modeling and mapping of land cover and plant ecological communities. Since the largest prefecture (Hokkaido, Japan) was split into two parts, this research dealt with 48 prefectural regions. The location map of the study area is shown in Figure 1.
The geography of the country is bordered by the Pacific Ocean in the east and Sea of Japan in the west, with a mountain range running through the middle of it. The country is divided into four climatic zones: subtropical and warm-temperate in the south, cool-temperate in the central area, and arctic in the north. The altitudinal variations vary from seashore to alpine regions, constituting typical vegetation zones corresponding to temperature gradients, such as Evergreen Broadleaf Forests (Camellia japonica, Cinnamomum japonica, Quercus glauca, etc.) distributed in the lower belt of the subtropical and warm temperate zone, Deciduous Broadleaf Forests (Fagus crenata, Quercus serrata, etc.) and Evergreen Conifer Forests (Pinus densiflora, Abies firma, etc.) distributed in the cool temperate zone, Evergreen Conifer Forests (Tsuga diversifolia, Thujopsis dolabrata, etc.) distributed in the subalpine zone, and Deciduous and Evergreen Shrubs (Pinus pumila, Quercus crispula, etc.) distributed in the alpine zone. This study area was chosen for the implementation of the DGPE classification system because of the diversity of ecological communities incorporated by the country associated with wider variations of both altitudinal and latitudinal ranges.
The community-level vegetation types were enumerated in the research by adopting the Dominant Genus-Physiognomy-Ecological (DGPE) system developed in the previous study [69] for satellite-based classification and mapping of plant ecological communities. Table 1 shows the list of DGPE communities enumerated countrywide. It also includes broad categories of land cover and agricultural types (water, barren, built-up, solar panels, upland field, paddy field, cultivated pastures, and orchard).
The DGPE communities were confirmed through field observations in several prefectures. Photographs taken during field observations are provided in Supplementary Material S1 for identifying some of the typical plant ecological communities of the study area.

2.2. Preparation of Ground Truth Data

This research utilized ground truth data prepared from extant multi-source survey databases through the procedure of stratified sampling, cross-checking, and pruning of noisy labels. The reference data belonging to the DGPE communities were extracted from extant vegetation survey maps (http://gis.biodic.go.jp/webgis, accessed on 10 April 2022) through the procedure of stratified random sampling and these reference data were cross-checked against the vegetation disturbances. For this purpose, vegetation disturbance maps were generated prefecture-wise through utilization of multi-year Sentinel-2 satellite images based on the ground truth data prepared from visual interpretation of annual composite images. The working definition of vegetation disturbance in this research was the loss of vegetation coverage occurring at any time within a study period. The pattern of vegetation disturbance includes logging of forests and conversion into herbaceous or barren lands, transfer of forested or non-forested vegetative areas into barren or built-up areas, etc. Though vegetation may have recovered after the disturbance, the original species composition might have changed, and thus this study masked out the disturbed areas occurring at any time in a study period. The ground truth data belonging to the solar panels were collected recently through visual interpretation of Google Earth images. The agricultural reference data (upland field, paddy field, cultivated pastures, and orchard) were extracted from extant vegetation survey maps through the procedure of stratified random sampling and cross-checked against the land-lot polygons provided by the Ministry of Agriculture, Forestry and Fisheries (https://www.maff.go.jp/j/tokei/porigon, accessed on 10 April 2022). For the built-up areas, the vegetation-survey-based reference data were cross-checked against the built-up areas extracted from open street maps (https://www.openstreetmap.org, accessed on 10 April 2022). The reference data prepared through the procedure of stratified sampling and cross-checking with extant multi-source databases were further strengthened by implementing the procedure of noisy label pruning with reference to Google Earth images. The size of the ground truth data utilized in the current research varied by 1–2% of the total prefectural areas for each prefecture.

2.3. Processing of Satellite Data

All Level-2A product images collected by Sentinel-2 mission satellites between 2018 and 2020 in the study area were utilized in the research. The Sentinel-2 mission acquires frequent optical imagery at high spatial resolution (10–60 m) in visible, near infrared, and short-wave wavelengths [70]. For each scene, cloud masking was undertaken and ten spectral bands (blue, green, red, red edge 1–3, near infrared, mid infrared, and shortwave infrared 1–2) were extracted; and seven spectral indexes, as shown in Table 2, were calculated.
The spectral indexes were selected based on the consideration of diversity of the spectral bands used, for instance, inclusion of the red edge bands; and ease of computation associated with the normalization due to the large scope of the work. The selected vegetation indexes using the red edge bands were effective in the differentiation of land cover and community-level vegetation types, whereas other spectral indexes were useful in the classification of land cover and agricultural types. The spectral-bands and spectral-indexes images were composited by computing half-month median values pixel by pixel. This procedure resulted in 408 features in total. The pixel values, corresponding to the ground truth data (geolocation points), were extracted from the stack of Sentinel-2 feature images consisting of both spectral bands and spectral indexes and utilized for deep convolutional learning. Supplementary Material S2 is provided to describe the capacities of the Sentinel-2 images to discriminate between plant ecological communities by virtue of monthly composite images using spectral band combinations and spectral indexes. The potential of specific spectral vegetation indexes for distinguishing particular vegetation types is also illustrated in Supplementary Material S2.

2.4. Deep Convolutional Learning and Mapping

A custom deep learning model made up of convolutional and densely connected neural network layers [78,79] capable of effectively handling time-series of the satellite data was implemented for the supervised classification of satellite images. A default model architecture was composed of four one-dimensional convolutional layers with ReLU activations followed by three densely connected neural network layers, two with ReLU activations and a final layer with Softmax activation to collect classifications. In this research, since each prefectural region was modeled and mapped separately, the parameters and hyper-parameters of the model including number of layers, number of neurons, learning rate, number of epochs, and batch size also varied with the prefectural regions. A 10-fold cross-validation method was used for fine-tuning the input features and model parameters and hyperparameters with reference to the classification accuracy metrics such as Kappa coefficient and F1-score [80]. After confirming optimum model parameters and hyperparameters, the mapping was undertaken with well-shuffled 85% training data, whereas 15% data were still kept for testing the model performance. The fine-tuned model established in this manner for each prefecture was utilized for prediction with unseen satellite datasets to produce DGPE maps. See Supplementary Material S3 for details on optimizing the deep learning models.

3. Results

3.1. Cross-Validation Accuracies

The class-wise accuracies in terms of Kappa coefficient and F1-score obtained from the 10-fold cross-validation method for each prefectural region are provided in Supplementary Material S4. For instance, the class-wise cross-validation accuracy obtained in one of the prefectural regions (Yamagata prefecture) is shown in Table 3.
A summary of average accuracies across all classes for each prefectural region is shown in Table 4.

3.2. Prefecture-Wise Ecological Communities Maps

The plant ecological community maps by prefectural region, based on the Dominant Genus-Physiognomy-Ecological (DGPE) system, which were produced in this research, are shown in Supplementary Material S5. For example, the plant ecological communities map of the Yamagata prefecture with 50 classes is shown in Figure 2. See Supplementary Material S5 for the plant ecological community maps for other prefectural regions. The resulting maps show clear distribution of land cover and community-level vegetation types in all prefectures.

3.3. Countrywide Distribution of Ecological Communities

The prefecture-wise plant ecological maps produced in the current research were merged together to produce a countrywide map, and the resulting map is shown in Figure 3. The prefecture-wise maps provided in Supplementary Material S5 are recommended in order to see the distribution of the plant ecological communities dealt with in this research.
From the countrywide integration of the plant ecological communities map, it was found that the Deciduous Broadleaf Forests of almost all prefectures are dominated by Quercus DBF in terms of area coverage; whereas Mallotus DBF in Kagoshima and Saga prefectures and Carpinus DBF in Kochi and Miyazaki prefectures are the most common ecological communities of the Deciduous Broadleaf Forests. The major ecological communities of the Evergreen Broadleaf Forests across all prefectures except in Hokkaido prefecture are Quercus EBF, Bamboo EBF, and Castanopsis EBF. The ecological communities of Larix DCF were found in almost all prefectures except in some southern prefectures. The Cryptomeria-Chamaecyparis ECF is the major community of the Evergreen Conifer Forests in most of the prefectures, except in some prefectures such as Iwate, Fukushima, and Hiroshima, where Pinus ECF is the major community of the Evergreen Conifer Forests; and in Hokkaido prefecture, where Abies ECF is the major community of the Evergreen Conifer Forests. The Herbaceous/Grassland areas are dominated by Miscanthus Herb and Open space Herb in almost all prefectures except in Hokkaido and Iwate prefectures where cultivated Pasture is very common. The major ecological communities of the Shrublands are Sasa Shrub and Rhododendron Shrub in almost all prefectures, except in some prefectures where Salix Shrub and Alnus Shrub are most common. The 10 m resolution plant ecological community maps produced in the research offer quantitative analysis on the spatial distribution of countrywide plant ecological communities. Further exploration on the spatial distribution of the plant ecological communities based on the maps produced in the current research is a subject for future investigation.

4. Discussion

In recent years, land cover and land use mapping by the utilization of high-spatial resolution satellite images such as Sentinel-2 or Landsat 8 has increased worldwide. In spite of high-resolution land cover mapping research focused at a local scale [81,82], mapping attempts at country, continental, or global scales are still limited. At the global scale, for example, Chen et al. [83] produced a worldwide land cover map at 30 m resolution using Landsat images. More recently, the European Space Agency produced a worldwide land cover map including vegetation physiognomic classes at 10 m resolution for 2020 based on Sentinel-1/2 images (https://esa-worldcover.org/en, accessed on 10 April 2022). At the continental scale, Li et al. [84] produced a land cover map of Africa at 10 m resolution from multi-source remote sensing data. Venter and Sydenham [85] produced land cover maps without vegetation physiognomic classes at 10 m resolution over Europe. However, both of these continental-scale maps do not include vegetation physiognomic classes and include land cover classes only. At the country level, for example, Inglada et al. [86] produced a high-resolution land cover map of France using satellite time-series images. The countrywide Dominant Genus-Physiognomy-Ecological (DGPE) mapping with 101 legends in this research can be considered as a paradigm shift in the field of land cover and vegetation mapping because of its unique characteristics of community-level vegetation mapping at a country scale. On the other hand, there is much research conducted in the classification of tree or shrub species [87,88,89,90], forest types [91,92,93,94], or crop types [95,96,97] using remote sensing images. However, most of this research was targeted at local scales in contrast to the country-level application of this research.
In recent years, there has been a surge in utility-scale installations of solar panels as a cleaner option for energy, with possible unintended environmental consequences such as land cover change, habitat fragmentation, and barriers to species movement [98,99]. In line with this, the mapping of an up-to-date distribution of solar panels at a country scale should be useful to better understand the consequences of increasing solar panel installations to nature and biodiversity conservation. In addition, classification and mapping of agricultural land types such as upland fields, paddy fields, orchards, and cultivated pasture, as conducted in the research, should provide datasets required to meet food security and sustainable development. When splitting the satellite-based raster images into prefectural boundaries using the vector polygons, the quality of boundary delineation depends on the resolution of the satellite images used. During integration of the prefecture-wise maps into the country level map, shortcomings in the delineation of a prefecture boundary pertaining to the resolution of the images was overcome by using the boundary of neighboring prefectures, except in outer coastal boundaries, and thus the seamless countrywide map was obtained. However, utilization of very-high-resolution satellite images or super-resolution of the high-resolution images through deep learning methods could produce land cover and community-level vegetation maps with precise delineation of the boundaries, improving the limitation pertaining to the high-resolution images.
As far as the methodology is concerned, most of the recent research on land cover and vegetation mapping is based on a supervised classification approach by employing machine learning classifiers in the presence of ground truth data [100,101,102]. Moreover, application of deep learning neural networks has been advancing rapidly in recent years as a versatile technology for the mapping of land cover and vegetation types [103,104,105,106]. This research based on the combination of convolutional and dense neural network layers applied efficiently with the ground truth data for the mapping of 101 classes at a country scale should be a timely and important contribution. The recent studies in the classification of ecological communities using satellite images have emphasized the usage of multi-temporal satellite images for improving performance [107,108,109]. Consequently, Kluczek et al. [110] achieved an F1-score in the range of 76–90% for the classification of 13 mountain forest and non-forest plant communities. Another study by Bhatt et al. [111] obtained a Kappa coefficient of 75% using the Random Forests classifier for the classification of habitat communities. Similar results (80% Kappa coefficient) were obtained from the Random Forests classifier in the classification of coastal wetlands by Martínez Prentice et al. [112]. Since those accuracies were obtained in local-scale classification tasks, an achievement of 80.5% average accuracy in terms of Kappa coefficient across 48 prefectural regions in the current research for the discrimination of complex ecological communities using multi-temporal Sentinel-2 images is a significant contribution.

5. Conclusions

Maintenance of high-resolution land cover and community-level vegetation information at a country scale is a challenging task. This research demonstrated production of Dominant Genus-Physiognomy-Ecological (DGPE) maps with 101 legends for the first time at a country scale from the satellite images. This research was made possible by utilizing entire archives of the multi-spectral images collected by a constellation of two polar-orbiting Sentinel-2 mission satellites. This research applied an innovative and robust methodology by automated scripted processing of satellite data and implementation of a convolutional neural network model for the production of land cover and community-level vegetation maps effectively at a country scale. It involved paying special attention to the ground truth data, utilization of both temporally and spectrally represented satellite features, and computation with graphical processing units to make deep-learning-based production at 10 m resolution efficient at a country scale. For precise delineation of the geographical boundaries, mainly the coastlines, utilization of very-high-resolution satellite images or super-resolution of the high-resolution images through deep learning methods is recommended in the future.
Achieving 80.2% average accuracy in terms of F1-score and 80.5% average accuracy in terms of Kappa coefficient across 48 prefectural regions shows the effectiveness of the methodology implemented in the research. This level of performance was achieved by implementing a unique prefecture-wise mapping approach by better matching the regional characteristics of the satellite data and vegetation distribution. This research highlights the importance of land cover and community-level vegetation mapping separately in small regions-of-interest and later integrating the outputs into a large scale for better performance. The promising results obtained in all prefectures in the current research confirmed the effectiveness of the DGPE classification system for large-scale classification and mapping of community-level vegetation types. The consistent accuracy in terms of F1-score across all prefectural regions obtained in the research indicates its suitability for application in other countries.

Supplementary Materials

The following supporting information can be downloaded at: https://www.mdpi.com/article/10.3390/app12147125/s1, Supplementary Material S1. Photographs of some typical plant ecological communities of the study area taken during field surveys. Supplementary Material S2. Discriminative potential of the Sentinel-2 images and Spectral-indexes. Supplementary Material S3. Optimization of Deep Learning Model Parameters and Hyper-parameters. Supplementary Material S4. Cross-validation accuracies for each prefecture. Supplementary Material S5. Land cover and community-level vegetation maps produced in the research.

Funding

This research received no external funding.

Acknowledgments

Sentinel-2 data were available from the European Space Agency (ESA) Copernicus program. The prefectural names and boundaries used in this research were based on the Land Numerical Information Data V2.3 of MLIT. Keitarou Hara is appreciated for logistic support to the research. The author is thankful to the anonymous reviewers and editors of the Journal.

Conflicts of Interest

The author declares no conflict of interest.

References

  1. Foley, J.A.; DeFries, R.; Asner, G.P.; Barford, C.; Bonan, G.; Carpenter, S.R.; Chapin, F.S.; Coe, M.T.; Daily, G.C.; Gibbs, H.K.; et al. Global Consequences of Land Use. Science 2005, 309, 570–574. [Google Scholar] [CrossRef] [Green Version]
  2. Chen, B.; Zhang, X.; Tao, J.; Wu, J.; Wang, J.; Shi, P.; Zhang, Y.; Yu, C. The Impact of Climate Change and Anthropogenic Activities on Alpine Grassland over the Qinghai-Tibet Plateau. Agric. For. Meteorol. 2014, 189–190, 11–18. [Google Scholar] [CrossRef]
  3. Lehosmaa, K.; Jyväsjärvi, J.; Virtanen, R.; Ilmonen, J.; Saastamoinen, J.; Muotka, T. Anthropogenic Habitat Disturbance Induces a Major Biodiversity Change in Habitat Specialist Bryophytes of Boreal Springs. Biol. Conserv. 2017, 215, 169–178. [Google Scholar] [CrossRef]
  4. Grimm, N.B.; Chapin, F.S.; Bierwagen, B.; Gonzalez, P.; Groffman, P.M.; Luo, Y.; Melton, F.; Nadelhoffer, K.; Pairis, A.; Raymond, P.A.; et al. The Impacts of Climate Change on Ecosystem Structure and Function. Front. Ecol. Environ. 2013, 11, 474–482. [Google Scholar] [CrossRef] [Green Version]
  5. Schirpke, U.; Kohler, M.; Leitinger, G.; Fontana, V.; Tasser, E.; Tappeiner, U. Future Impacts of Changing Land-Use and Climate on Ecosystem Services of Mountain Grassland and Their Resilience. Ecosyst. Serv. 2017, 26, 79–94. [Google Scholar] [CrossRef] [PubMed]
  6. Li, D.; Wu, S.; Liu, L.; Zhang, Y.; Li, S. Vulnerability of the Global Terrestrial Ecosystems to Climate Change. Glob. Chang. Biol. 2018, 24, 4095–4106. [Google Scholar] [CrossRef]
  7. Kuchler, A.W. Problems in Classifying and Mapping Vegetation for Ecological Regionalization. Ecology 1973, 54, 512–523. [Google Scholar] [CrossRef]
  8. Henderson, E.B.; Bell, D.M.; Gregory, M.J. Vegetation Mapping to Support Greater Sage-grouse Habitat Monitoring and Management: Multi- or Univariate Approach? Ecosphere 2019, 10, e02838. [Google Scholar] [CrossRef]
  9. Schrodt, F.; de la Barreda Bautista, B.; Williams, C.; Boyd, D.S.; Schaepman-Strub, G.; Santos, M.J. Integrating Biodiversity, Remote Sensing, and Auxiliary Information for the Study of Ecosystem Functioning and Conservation at Large Spatial Scales. In Remote Sensing of Plant Biodiversity; Cavender-Bares, J., Gamon, J.A., Townsend, P.A., Eds.; Springer International Publishing: Cham, Switzerland, 2020; pp. 449–484. ISBN 978-3-030-33156-6. [Google Scholar]
  10. Bowler, D.E.; Bjorkman, A.D.; Dornelas, M.; Myers-Smith, I.H.; Navarro, L.M.; Niamir, A.; Supp, S.R.; Waldock, C.; Winter, M.; Vellend, M.; et al. Mapping Human Pressures on Biodiversity across the Planet Uncovers Anthropogenic Threat Complexes. People Nat. 2020, 2, 380–394. [Google Scholar] [CrossRef] [Green Version]
  11. Iverson, L.R.; Graham, R.L.; Cook, E.A. Applications of Satellite Remote Sensing to Forested Ecosystems. Landscape Ecol. 1989, 3, 131–143. [Google Scholar] [CrossRef]
  12. Pettorelli, N.; Schulte to Bühne, H.; Tulloch, A.; Dubois, G.; Macinnis-Ng, C.; Queirós, A.M.; Keith, D.A.; Wegmann, M.; Schrodt, F.; Stellmes, M.; et al. Satellite Remote Sensing of Ecosystem Functions: Opportunities, Challenges and Way Forward. Remote Sens. Ecol. Conserv. 2018, 4, 71–93. [Google Scholar] [CrossRef]
  13. Guo, X.; Coops, N.C.; Gergel, S.E.; Bater, C.W.; Nielsen, S.E.; Stadt, J.J.; Drever, M. Integrating Airborne Lidar and Satellite Imagery to Model Habitat Connectivity Dynamics for Spatial Conservation Prioritization. Landsc. Ecol. 2018, 33, 491–511. [Google Scholar] [CrossRef]
  14. Babí Almenar, J.; Bolowich, A.; Elliot, T.; Geneletti, D.; Sonnemann, G.; Rugani, B. Assessing Habitat Loss, Fragmentation and Ecological Connectivity in Luxembourg to Support Spatial Planning. Landsc. Urban Plan. 2019, 189, 335–351. [Google Scholar] [CrossRef]
  15. Fu, W.; Ma, J.; Chen, P.; Chen, F. Remote Sensing Satellites for Digital Earth. In Manual of Digital Earth; Guo, H., Goodchild, M.F., Annoni, A., Eds.; Springer: Singapore, 2020; pp. 55–123. ISBN 978-981-329-914-6. [Google Scholar]
  16. Rhodes, C.J.; Henrys, P.; Siriwardena, G.M.; Whittingham, M.J.; Norton, L.R. The Relative Value of Field Survey and Remote Sensing for Biodiversity Assessment. Methods Ecol. Evol. 2015, 6, 772–781. [Google Scholar] [CrossRef]
  17. Macintyre, P.; van Niekerk, A.; Mucina, L. Efficacy of Multi-Season Sentinel-2 Imagery for Compositional Vegetation Classification. Int. J. Appl. Earth Obs. Geoinf. 2020, 85, 101980. [Google Scholar] [CrossRef]
  18. Brook, R.K.; Kenkel, N.C. A Multivariate Approach to Vegetation Mapping of Manitoba’s Hudson Bay Lowlands. International J. Remote Sens. 2002, 23, 4761–4776. [Google Scholar] [CrossRef]
  19. Xie, Y.; Sha, Z.; Yu, M. Remote Sensing Imagery in Vegetation Mapping: A Review. J. Plant Ecol. 2008, 1, 9–23. [Google Scholar] [CrossRef]
  20. Ustin, S.L.; Middleton, E.M. Current and Near-Term Advances in Earth Observation for Ecological Applications. Ecol. Process 2021, 10, 1. [Google Scholar] [CrossRef]
  21. Adam, E.; Mutanga, O.; Rugege, D. Multispectral and Hyperspectral Remote Sensing for Identification and Mapping of Wetland Vegetation: A Review. Wetl. Ecol. Manag. 2010, 18, 281–296. [Google Scholar] [CrossRef]
  22. Natividade, J.; Prado, J.; Marques, L. Low-Cost Multi-Spectral Vegetation Classification Using an Unmanned Aerial Vehicle. In Proceedings of the 2017 IEEE International Conference on Autonomous Robot Systems and Competitions (ICARSC), Coimbra, Portugal, 26–28 April 2017; IEEE: Piscataway, NJ, USA, 2017; pp. 336–342. [Google Scholar]
  23. Langford, Z.L.; Kumar, J.; Hoffman, F.M.; Breen, A.L.; Iversen, C.M. Arctic Vegetation Mapping Using Unsupervised Training Datasets and Convolutional Neural Networks. Remote Sens. 2019, 11, 69. [Google Scholar] [CrossRef] [Green Version]
  24. Westinga, E.; Beltran, A.P.R.; de Bie, C.A.J.M.; van Gils, H.A.M.J. A Novel Approach to Optimize Hierarchical Vegetation Mapping from Hyper-Temporal NDVI Imagery, Demonstrated at National Level for Namibia. Int. J. Appl. Earth Obs. Geoinf. 2020, 91, 102152. [Google Scholar] [CrossRef]
  25. Yeo, S.; Lafon, V.; Alard, D.; Curti, C.; Dehouck, A.; Benot, M.-L. Classification and Mapping of Saltmarsh Vegetation Combining Multispectral Images with Field Data. Estuar. Coast. Shelf Sci. 2020, 236, 106643. [Google Scholar] [CrossRef]
  26. Lassiter, A.; Darbari, M. Assessing Alternative Methods for Unsupervised Segmentation of Urban Vegetation in Very High-Resolution Multispectral Aerial Imagery. PLoS ONE 2020, 15, e0230856. [Google Scholar] [CrossRef] [PubMed]
  27. Shih, H.; Stow, D.A.; Tsai, Y.H. Guidance on and Comparison of Machine Learning Classifiers for Landsat-Based Land Cover and Land Use Mapping. Int. J. Remote Sens. 2019, 40, 1248–1274. [Google Scholar] [CrossRef]
  28. Talukdar, S.; Singha, P.; Mahato, S.; Shahfahad; Pal, S.; Liou, Y.-A.; Rahman, A. Land-Use Land-Cover Classification by Machine Learning Classifiers for Satellite Observations—A Review. Remote Sens. 2020, 12, 1135. [Google Scholar] [CrossRef] [Green Version]
  29. Furuya, D.E.G.; Aguiar, J.A.F.; Estrabis, N.V.; Pinheiro, M.M.F.; Furuya, M.T.G.; Pereira, D.R.; Gonçalves, W.N.; Liesenberg, V.; Li, J.; Marcato Junior, J.; et al. A Machine Learning Approach for Mapping Forest Vegetation in Riparian Zones in an Atlantic Biome Environment Using Sentinel-2 Imagery. Remote Sens. 2020, 12, 4086. [Google Scholar] [CrossRef]
  30. Hamylton, S.M.; Morris, R.H.; Carvalho, R.C.; Roder, N.; Barlow, P.; Mills, K.; Wang, L. Evaluating Techniques for Mapping Island Vegetation from Unmanned Aerial Vehicle (UAV) Images: Pixel Classification, Visual Interpretation and Machine Learning Approaches. Int. J. Appl. Earth Obs. Geoinf. 2020, 89, 102085. [Google Scholar] [CrossRef]
  31. Pan, X.; Wang, Z.; Gao, Y.; Dang, X.; Han, Y. Detailed and Automated Classification of Land Use/Land Cover Using Machine Learning Algorithms in Google Earth Engine. Geocarto Int. 2021. [Google Scholar] [CrossRef]
  32. Breiman, L. Random Forests. Mach. Learn. 2001, 45, 5–32. [Google Scholar] [CrossRef] [Green Version]
  33. Chang, C.-C.; Lin, C.-J. LIBSVM: A Library for Support Vector Machines. ACM Trans. Intell. Syst. Technol. 2011, 2, 1–27. [Google Scholar] [CrossRef]
  34. Chen, T.; Guestrin, C. XGBoost: A Scalable Tree Boosting System. In Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, San Francisco, CA, USA, 13–17 August 2016; ACM: New York, NY, USA, 2016; pp. 785–794. [Google Scholar]
  35. Szantoi, Z.; Escobedo, F.J.; Abd-Elrahman, A.; Pearlstine, L.; Dewitt, B.; Smith, S. Classifying Spatially Heterogeneous Wetland Communities Using Machine Learning Algorithms and Spectral and Textural Features. Env. Monit. Assess. 2015, 187, 262. [Google Scholar] [CrossRef] [PubMed]
  36. Hird, J.; DeLancey, E.; McDermid, G.; Kariyeva, J. Google Earth Engine, Open-Access Satellite Data, and Machine Learning in Support of Large-Area Probabilistic Wetland Mapping. Remote Sens. 2017, 9, 1315. [Google Scholar] [CrossRef] [Green Version]
  37. Whyte, A.; Ferentinos, K.P.; Petropoulos, G.P. A New Synergistic Approach for Monitoring Wetlands Using Sentinels -1 and 2 Data with Object-Based Machine Learning Algorithms. Environ. Model. Softw. 2018, 104, 40–54. [Google Scholar] [CrossRef] [Green Version]
  38. Zhang, X.; Feng, X.; Jiang, H. Object-Oriented Method for Urban Vegetation Mapping Using IKONOS Imagery. Int. J. Remote Sens. 2010, 31, 177–196. [Google Scholar] [CrossRef]
  39. Ouerghemmi, W.; Gadal, S.; Mozgeris, G.; Jonikavicius, D. Urban Vegetation Mapping by Airborne Hyperspetral Imagery; Feasibility and Limitations. In Proceedings of the 2018 9th Workshop on Hyperspectral Image and Signal Processing: Evolution in Remote Sensing (WHISPERS), Amsterdam, The Netherlands, 23–26 September 2018; IEEE: Piscataway, NJ, USA, 2018; pp. 1–5. [Google Scholar]
  40. Vaglio Laurin, G.; Puletti, N.; Hawthorne, W.; Liesenberg, V.; Corona, P.; Papale, D.; Chen, Q.; Valentini, R. Discrimination of Tropical Forest Types, Dominant Species, and Mapping of Functional Guilds by Hyperspectral and Simulated Multispectral Sentinel-2 Data. Remote Sens. Environ. 2016, 176, 163–176. [Google Scholar] [CrossRef] [Green Version]
  41. Grabska, E.; Frantz, D.; Ostapowicz, K. Evaluation of Machine Learning Algorithms for Forest Stand Species Mapping Using Sentinel-2 Imagery and Environmental Data in the Polish Carpathians. Remote Sens. Environ. 2020, 251, 112103. [Google Scholar] [CrossRef]
  42. Su, L. Optimizing Support Vector Machine Learning for Semi-Arid Vegetation Mapping by Using Clustering Analysis. ISPRS J. Photogramm. Remote Sens. 2009, 64, 407–413. [Google Scholar] [CrossRef]
  43. Adam, E.; Mureriwa, N.; Newete, S. Mapping Prosopis Glandulosa (Mesquite) in the Semi-Arid Environment of South Africa Using High-Resolution WorldView-2 Imagery and Machine Learning Classifiers. J. Arid Environ. 2017, 145, 43–51. [Google Scholar] [CrossRef]
  44. Nguyen, U.; Glenn, E.P.; Dang, T.D.; Pham, L.T.H. Mapping Vegetation Types in Semi-Arid Riparian Regions Using Random Forest and Object-Based Image Approach: A Case Study of the Colorado River Ecosystem, Grand Canyon, Arizona. Ecol. Inform. 2019, 50, 43–50. [Google Scholar] [CrossRef]
  45. Yuan, H.; Van Der Wiele, C.; Khorram, S. An Automated Artificial Neural Network System for Land Use/Land Cover Classification from Landsat TM Imagery. Remote Sens. 2009, 1, 243–265. [Google Scholar] [CrossRef] [Green Version]
  46. Alqadhi, S.; Mallick, J.; Balha, A.; Bindajam, A.; Singh, C.K.; Hoa, P.V. Spatial and Decadal Prediction of Land Use/Land Cover Using Multi-Layer Perceptron-Neural Network (MLP-NN) Algorithm for a Semi-Arid Region of Asir, Saudi Arabia. Earth Sci. Inf. 2021, 14, 1547–1562. [Google Scholar] [CrossRef]
  47. Kussul, N.; Lavreniuk, M.; Skakun, S.; Shelestov, A. Deep Learning Classification of Land Cover and Crop Types Using Remote Sensing Data. IEEE Geosci. Remote Sens. Lett. 2017, 14, 778–782. [Google Scholar] [CrossRef]
  48. Zhang, C.; Sargent, I.; Pan, X.; Li, H.; Gardiner, A.; Hare, J.; Atkinson, P.M. Joint Deep Learning for Land Cover and Land Use Classification. Remote Sens. Environ. 2019, 221, 173–187. [Google Scholar] [CrossRef] [Green Version]
  49. Ienco, D.; Gaetano, R.; Dupaquier, C.; Maurel, P. Land Cover Classification via Multitemporal Spatial Data by Deep Recurrent Neural Networks. IEEE Geosci. Remote Sens. Lett. 2017, 14, 1685–1689. [Google Scholar] [CrossRef] [Green Version]
  50. Sun, Z.; Di, L.; Fang, H. Using Long Short-Term Memory Recurrent Neural Network in Land Cover Classification on Landsat and Cropland Data Layer Time Series. Int. J. Remote Sens. 2019, 40, 593–614. [Google Scholar] [CrossRef]
  51. Mazzia, V.; Khaliq, A.; Chiaberge, M. Improvement in Land Cover and Crop Classification Based on Temporal Features Learning from Sentinel-2 Data Using Recurrent-Convolutional Neural Network (R-CNN). Appl. Sci. 2019, 10, 238. [Google Scholar] [CrossRef] [Green Version]
  52. Zhu, Y.; Geis, C.; So, E.; Jin, Y. Multitemporal Relearning With Convolutional LSTM Models for Land Use Classification. IEEE J. Sel. Top. Appl. Earth Obs. Remote Sens. 2021, 14, 3251–3265. [Google Scholar] [CrossRef]
  53. Rußwurm, M.; Körner, M. Self-Attention for Raw Optical Satellite Time Series Classification. ISPRS J. Photogramm. Remote Sens. 2020, 169, 421–435. [Google Scholar] [CrossRef]
  54. Praveen, B.; Menon, V. A Bidirectional Deep-Learning-Based Spectral Attention Mechanism for Hyperspectral Data Classification. Remote Sens. 2022, 14, 217. [Google Scholar] [CrossRef]
  55. Grekousis, G.; Mountrakis, G.; Kavouras, M. An Overview of 21 Global and 43 Regional Land-Cover Mapping Products. Int. J. Remote Sens. 2015, 36, 5309–5335. [Google Scholar] [CrossRef]
  56. Running, S.W.; Loveland, T.R.; Pierce, L.L.; Nemani, R.R.; Hunt, E.R. A Remote Sensing Based Vegetation Classification Logic for Global Land Cover Analysis. Remote Sens. Environ. 1995, 51, 39–48. [Google Scholar] [CrossRef]
  57. Cihlar, J. Land Cover Mapping of Large Areas from Satellites: Status and Research Priorities. Int. J. Remote Sens. 2000, 21, 1093–1114. [Google Scholar] [CrossRef]
  58. Kaplan, G. Broad-Leaved and Coniferous Forest Classification in Google Earth Engine Using Sentinel Imagery. In Proceedings of the 1st International Electronic Conference on Forests—Forests for a Better Future: Sustainability, Innovation, Interdisciplinarity, Online, 15–30 November 2020; p. 64. [Google Scholar]
  59. Poore, M.E.D. The Use of Phytosociological Methods in Ecological Investigations: I. The Braun-Blanquet System. J. Ecol. 1955, 43, 226. [Google Scholar] [CrossRef]
  60. Köppen, W. Das Geographische System Der Klimate. 1936. Available online: https://cir.nii.ac.jp/crid/1571417124443846784 (accessed on 10 April 2022).
  61. Metzger, M.J.; Bunce, R.G.H.; Jongman, R.H.G.; Sayre, R.; Trabucco, A.; Zomer, R. A High-Resolution Bioclimate Map of the World: A Unifying Framework for Global Biodiversity Research and Monitoring: High-Resolution Bioclimate Map of the World. Glob. Ecol. Biogeogr. 2013, 22, 630–638. [Google Scholar] [CrossRef] [Green Version]
  62. Bailey, R.G. Ecosystem Geography: From Ecoregions to Sites; Springer Science & Business Media: Berlin/Heidelberg, Germany, 2009; ISBN 0-387-89515-9. [Google Scholar]
  63. Bailey, R.G. Suggested Hierarchy of Criteria for Multi-Scale Ecosystem Mapping. Landsc. Urban Plan. 1987, 14, 313–319. [Google Scholar] [CrossRef] [Green Version]
  64. Küchler, A.W. A Physiognomic Classification of Vegetation. Ann. Assoc. Am. Geogr. 1949, 39, 201–210. [Google Scholar] [CrossRef]
  65. Beard, J.S. The Physiognomic Approach. In Classification of Plant Communities; Whittaker, R.H., Ed.; Springer: Dordrecht, The Netherlands, 1978; pp. 33–64. ISBN 978-94-009-9183-5. [Google Scholar]
  66. Braun-Blanquet, J.P.; Schoenichen, W. Grundzuge Der Vegetationskunde. Auf Wien 1964, 865. [Google Scholar] [CrossRef]
  67. Theurillat, J.; Willner, W.; Fernández-González, F.; Bültmann, H.; Čarni, A.; Gigante, D.; Mucina, L.; Weber, H. International Code of Phytosociological Nomenclature. 4th Edition. Appl. Veg. Sci. 2021, 24, e12491. [Google Scholar] [CrossRef] [Green Version]
  68. Bredenkamp, G.; Chytrý, M.; Fischer, H.S.; Neuhäuslová, Z.; van der Maarel, E. Vegetation Mapping: Theory, Methods and Case Studies: Introduction. Appl. Veg. Sci. 1998, 1, 162–164. [Google Scholar] [CrossRef]
  69. Sharma, R.C. Genus-Physiognomy-Ecosystem (GPE) System for Satellite-Based Classification of Plant Communities. Ecologies 2021, 2, 203–213. [Google Scholar] [CrossRef]
  70. Drusch, M.; Del Bello, U.; Carlier, S.; Colin, O.; Fernandez, V.; Gascon, F.; Hoersch, B.; Isola, C.; Laberinti, P.; Martimort, P.; et al. Sentinel-2: ESA’s Optical High-Resolution Mission for GMES Operational Services. Remote Sens. Environ. 2012, 120, 25–36. [Google Scholar] [CrossRef]
  71. Rouse, J.W.; Haas, R.H.; Schell, J.A.; Deering, D.W. Monitoring Vegetation Systems in the Great Plains with ERTS. NASA Spec. Publ. 1974, 351, 309. [Google Scholar]
  72. McFeeters, S.K. The Use of the Normalized Difference Water Index (NDWI) in the Delineation of Open Water Features. Int. J. Remote Sens. 1996, 17, 1425–1432. [Google Scholar] [CrossRef]
  73. Riggs, G.A.; Hall, D.K.; Salomonson, V.V. A Snow Index for the Landsat Thematic Mapper and Moderate Resolution Imaging Spectroradiometer. In Proceedings of the IGARSS ’94—1994 IEEE International Geoscience and Remote Sensing Symposium, Pasadena, CA, USA, 8–12 August 1994; IEEE: Piscataway, NJ, USA, 1994; Volume 4, pp. 1942–1944. [Google Scholar]
  74. Chandrasekar, K.; Sesha Sai, M.V.R.; Roy, P.S.; Dwevedi, R.S. Land Surface Water Index (LSWI) Response to Rainfall and NDVI Using the MODIS Vegetation Index Product. Int. J. Remote Sens. 2010, 31, 3987–4005. [Google Scholar] [CrossRef]
  75. Falkowski, M.J.; Gessler, P.E.; Morgan, P.; Hudak, A.T.; Smith, A.M.S. Characterizing and Mapping Forest Fire Fuels Using ASTER Imagery and Gradient Modeling. For. Ecol. Manag. 2005, 217, 129–146. [Google Scholar] [CrossRef] [Green Version]
  76. Gitelson, A.; Merzlyak, M.N. Spectral Reflectance Changes Associated with Autumn Senescence of Aesculus hippocastanum L. and Acer platanoides L. Leaves. Spectral Features and Relation to Chlorophyll Estimation. J. Plant Physiol. 1994, 143, 286–292. [Google Scholar] [CrossRef]
  77. Maccioni, A.; Agati, G.; Mazzinghi, P. New Vegetation Indices for Remote Measurement of Chlorophylls Based on Leaf Directional Reflectance Spectra. J. Photochem. Photobiol. B Biol. 2001, 61, 52–61. [Google Scholar] [CrossRef]
  78. Schmidhuber, J. Deep Learning in Neural Networks: An Overview. Neural Netw. 2015, 61, 85–117. [Google Scholar] [CrossRef] [Green Version]
  79. LeCun, Y.; Bengio, Y.; Hinton, G. Deep Learning. Nature 2015, 521, 436–444. [Google Scholar] [CrossRef]
  80. Goutte, C.; Gaussier, E. A Probabilistic Interpretation of Precision, Recall and F-Score, with Implication for Evaluation. In Advances in Information Retrieval; Losada, D.E., Fernández-Luna, J.M., Eds.; Springer: Berlin/Heidelberg, Germany, 2005; Volume 3408, pp. 345–359. ISBN 978-3-540-25295-5. [Google Scholar]
  81. Carrasco, L.; O’Neil, A.; Morton, R.; Rowland, C. Evaluating Combinations of Temporally Aggregated Sentinel-1, Sentinel-2 and Landsat 8 for Land Cover Mapping with Google Earth Engine. Remote Sens. 2019, 11, 288. [Google Scholar] [CrossRef] [Green Version]
  82. González-González, A.; Clerici, N.; Quesada, B. A 30 M-Resolution Land Use-Land Cover Product for the Colombian Andes and Amazon Using Cloud-Computing. Int. J. Appl. Earth Obs. Geoinf. 2022, 107, 102688. [Google Scholar] [CrossRef]
  83. Chen, J.; Chen, J.; Liao, A.; Cao, X.; Chen, L.; Chen, X.; He, C.; Han, G.; Peng, S.; Lu, M.; et al. Global Land Cover Mapping at 30 m Resolution: A POK-Based Operational Approach. ISPRS J. Photogramm. Remote Sens. 2015, 103, 7–27. [Google Scholar] [CrossRef] [Green Version]
  84. Li, Q.; Qiu, C.; Ma, L.; Schmitt, M.; Zhu, X. Mapping the Land Cover of Africa at 10 m Resolution from Multi-Source Remote Sensing Data with Google Earth Engine. Remote Sens. 2020, 12, 602. [Google Scholar] [CrossRef] [Green Version]
  85. Venter, Z.S.; Sydenham, M.A.K. Continental-Scale Land Cover Mapping at 10 m Resolution Over Europe (ELC10). Remote Sens. 2021, 13, 2301. [Google Scholar] [CrossRef]
  86. Inglada, J.; Vincent, A.; Arias, M.; Tardy, B.; Morin, D.; Rodes, I. Operational High Resolution Land Cover Map Production at the Country Scale Using Satellite Image Time Series. Remote Sens. 2017, 9, 95. [Google Scholar] [CrossRef] [Green Version]
  87. Hościło, A.; Lewandowska, A. Mapping Forest Type and Tree Species on a Regional Scale Using Multi-Temporal Sentinel-2 Data. Remote Sens. 2019, 11, 929. [Google Scholar] [CrossRef] [Green Version]
  88. Hartling, S.; Sagan, V.; Sidike, P.; Maimaitijiang, M.; Carron, J. Urban Tree Species Classification Using a WorldView-2/3 and LiDAR Data Fusion Approach and Deep Learning. Sensors 2019, 19, 1284. [Google Scholar] [CrossRef] [Green Version]
  89. Jombo, S.; Adam, E.; Tesfamichael, S. Classification of Urban Tree Species Using LiDAR Data and WorldView-2 Satellite Imagery in a Heterogeneous Environment. Geocarto Int. 2022, 12, 1–24. [Google Scholar] [CrossRef]
  90. Carbonell-Rivera, J.P.; Torralba, J.; Estornell, J.; Ruiz, L.Á.; Crespo-Peremarch, P. Classification of Mediterranean Shrub Species from UAV Point Clouds. Remote Sens. 2022, 14, 199. [Google Scholar] [CrossRef]
  91. Pasquarella, V.J.; Holden, C.E.; Woodcock, C.E. Improved Mapping of Forest Type Using Spectral-Temporal Landsat Features. Remote Sens. Environ. 2018, 210, 193–207. [Google Scholar] [CrossRef]
  92. Liu, Y.; Gong, W.; Hu, X.; Gong, J. Forest Type Identification with Random Forest Using Sentinel-1A, Sentinel-2A, Multi-Temporal Landsat-8 and DEM Data. Remote Sens. 2018, 10, 946. [Google Scholar] [CrossRef] [Green Version]
  93. Wessel, M.; Brandmeier, M.; Tiede, D. Evaluation of Different Machine Learning Algorithms for Scalable Classification of Tree Types and Tree Species Based on Sentinel-2 Data. Remote Sens. 2018, 10, 1419. [Google Scholar] [CrossRef] [Green Version]
  94. Yu, Y.; Li, M.; Fu, Y. Forest Type Identification by Random Forest Classification Combined with SPOT and Multitemporal SAR Data. J. For. Res. 2018, 29, 1407–1414. [Google Scholar] [CrossRef]
  95. Orynbaikyzy, A.; Gessner, U.; Conrad, C. Crop Type Classification Using a Combination of Optical and Radar Remote Sensing Data: A Review. Int. J. Remote Sens. 2019, 40, 6553–6595. [Google Scholar] [CrossRef]
  96. Sun, C.; Bian, Y.; Zhou, T.; Pan, J. Using of Multi-Source and Multi-Temporal Remote Sensing Data Improves Crop-Type Mapping in the Subtropical Agriculture Region. Sensors 2019, 19, 2401. [Google Scholar] [CrossRef] [Green Version]
  97. Tufail, R.; Ahmad, A.; Javed, M.A.; Ahmad, S.R. A Machine Learning Approach for Accurate Crop Type Mapping Using Combined SAR and Optical Time Series Data. Adv. Space Res. 2022, 69, 331–346. [Google Scholar] [CrossRef]
  98. Hernandez, R.R.; Easter, S.B.; Murphy-Mariscal, M.L.; Maestre, F.T.; Tavassoli, M.; Allen, E.B.; Barrows, C.W.; Belnap, J.; Ochoa-Hueso, R.; Ravi, S.; et al. Environmental Impacts of Utility-Scale Solar Energy. Renew. Sustain. Energy Rev. 2014, 29, 766–779. [Google Scholar] [CrossRef] [Green Version]
  99. Dhar, A.; Naeth, M.A.; Jennings, P.D.; Gamal El-Din, M. Perspectives on Environmental Impacts and a Land Reclamation Strategy for Solar and Wind Energy Systems. Sci. Total Environ. 2020, 718, 134602. [Google Scholar] [CrossRef]
  100. Ghorbanian, A.; Kakooei, M.; Amani, M.; Mahdavi, S.; Mohammadzadeh, A.; Hasanlou, M. Improved Land Cover Map of Iran Using Sentinel Imagery within Google Earth Engine and a Novel Automatic Workflow for Land Cover Classification Using Migrated Training Samples. ISPRS J. Photogramm. Remote Sens. 2020, 167, 276–288. [Google Scholar] [CrossRef]
  101. Shafizadeh-Moghadam, H.; Khazaei, M.; Alavipanah, S.K.; Weng, Q. Google Earth Engine for Large-Scale Land Use and Land Cover Mapping: An Object-Based Classification Approach Using Spectral, Textural and Topographical Factors. GISci. Remote Sens. 2021, 58, 914–928. [Google Scholar] [CrossRef]
  102. Pizarro, S.E.; Pricope, N.G.; Vargas-Machuca, D.; Huanca, O.; Ñaupari, J. Mapping Land Cover Types for Highland Andean Ecosystems in Peru Using Google Earth Engine. Remote Sens. 2022, 14, 1562. [Google Scholar] [CrossRef]
  103. Pelletier, C.; Webb, G.; Petitjean, F. Temporal Convolutional Neural Network for the Classification of Satellite Image Time Series. Remote Sens. 2019, 11, 523. [Google Scholar] [CrossRef] [Green Version]
  104. Vali, A.; Comai, S.; Matteucci, M. Deep Learning for Land Use and Land Cover Classification Based on Hyperspectral and Multispectral Earth Observation Data: A Review. Remote Sens. 2020, 12, 2495. [Google Scholar] [CrossRef]
  105. Safari, K.; Prasad, S.; Labate, D. A Multiscale Deep Learning Approach for High-Resolution Hyperspectral Image Classification. IEEE Geosci. Remote Sens. Lett. 2021, 18, 167–171. [Google Scholar] [CrossRef]
  106. Ghayour, L.; Neshat, A.; Paryani, S.; Shahabi, H.; Shirzadi, A.; Chen, W.; Al-Ansari, N.; Geertsema, M.; Pourmehdi Amiri, M.; Gholamnia, M.; et al. Performance Evaluation of Sentinel-2 and Landsat 8 OLI Data for Land Cover/Use Classification Using a Comparison between Machine Learning Algorithms. Remote Sens. 2021, 13, 1349. [Google Scholar] [CrossRef]
  107. Li, D.; Ke, Y.; Gong, H.; Li, X. Object-Based Urban Tree Species Classification Using Bi-Temporal WorldView-2 and WorldView-3 Images. Remote Sens. 2015, 7, 16917–16937. [Google Scholar] [CrossRef] [Green Version]
  108. Karlson, M.; Ostwald, M.; Reese, H.; Bazié, H.R.; Tankoano, B. Assessing the Potential of Multi-Seasonal WorldView-2 Imagery for Mapping West African Agroforestry Tree Species. Int. J. Appl. Earth Obs. Geoinf. 2016, 50, 80–88. [Google Scholar] [CrossRef]
  109. Adamo, M.; Tomaselli, V.; Tarantino, C.; Vicario, S.; Veronico, G.; Lucas, R.; Blonda, P. Knowledge-Based Classification of Grassland Ecosystem Based on Multi-Temporal WorldView-2 Data and FAO-LCCS Taxonomy. Remote Sens. 2020, 12, 1447. [Google Scholar] [CrossRef]
  110. Kluczek, M.; Zagajewski, B.; Kycko, M. Airborne HySpex Hyperspectral Versus Multitemporal Sentinel-2 Images for Mountain Plant Communities Mapping. Remote Sens. 2022, 14, 1209. [Google Scholar] [CrossRef]
  111. Bhatt, P.; Maclean, A.; Dickinson, Y.; Kumar, C. Fine-Scale Mapping of Natural Ecological Communities Using Machine Learning Approaches. Remote Sens. 2022, 14, 563. [Google Scholar] [CrossRef]
  112. Martínez Prentice, R.; Villoslada Peciña, M.; Ward, R.D.; Bergamo, T.F.; Joyce, C.B.; Sepp, K. Machine Learning Classification and Accuracy Assessment from High-Resolution Images of Coastal Wetlands. Remote Sens. 2021, 13, 3669. [Google Scholar] [CrossRef]
Applsci 12 07125 g001 550
Figure 1. Location map of the study area.
Figure 1. Location map of the study area.
Applsci 12 07125 g001
Applsci 12 07125 g002 550
Figure 2. Plant ecological communities map of the Yamagata prefecture with 50 classes including land cover and agricultural types, produced in the research.
Figure 2. Plant ecological communities map of the Yamagata prefecture with 50 classes including land cover and agricultural types, produced in the research.
Applsci 12 07125 g002
Applsci 12 07125 g003 550
Figure 3. Plant ecological communities map produced at a country scale.
Figure 3. Plant ecological communities map produced at a country scale.
Applsci 12 07125 g003
Table
Table 1. List of Dominant Genus-Physiognomy-Ecological (DGPE) communities enumerated countrywide including land cover and agricultural types.
Table 1. List of Dominant Genus-Physiognomy-Ecological (DGPE) communities enumerated countrywide including land cover and agricultural types.
1. Abandoned land35. Elaeagnus Shrub69. Populus DBF
2. Abies ECF36. Elaeocarpus EBF70. Prunus DBF
3. Acacia EBF37. Euptelea DBF71. Pterocarya DBF
4. Acer DBF38. Eurya EBF72. Pterostyrax DBF
5. Acer Shrub39. Eurya ECF73. Quercus DBF
6. Alnus DBF40. Evodia DBF74. Quercus EBF
7. Alnus Shrub41. Fagus DBF75. Quercus Shrub
8. Alpine Herb42. Fraxinus DBF76. Rhododendron Shrub
9. Alpine Shrub43. Heritiera EBF77. Robinia DBF
10. Ardisia EBF44. Hibiscus EBF78. Salix DBF
11. Bamboo EBF45. Hydrangea Shrub79. Salix Shrub
12. Bamboo Shrub46. Juglans DBF80. Sasa Shrub
13. Barren47. Juniperus Shrub81. Schima EBF
14. Betula DBF48. Lagestroemia DBF82. Sciadopitys ECF
15. Betula Shrub49. Larix DCF83. Solar panels
16. Built-up50. Leucaena DBF84. Sorbus DBF
17. Calamagrostis Herb51. Lithocarpus EBF85. Stewartia DBF
18. Calophyllum EBF52. Litsea EBF86. Symplocos EBF
19. Camellia EBF53. Machilus EBF87. Taxus Shrub
20. Carpinus DBF54. Mallotus DBF88. Thuja ECF
21. Carpinus Shrub55. Mangroves EBF89. Thujopsis ECF
22. Castanopsis EBF56. Melia DBF90. Tilia DBF
23. Casuarina ECF57. Miscanthus Herb91. Treefern ECF
24. Celtis DBF58. Open space Herb92. Trema EBF
25. Cercidiphyllum DBF59. Orchard93. Trochodendron EBF
26. Chionanthus EBF60. Paddy field94. Tsuga ECF
27. Cinnamomum EBF61. Palm ECF95. Ulmus DBF
28. Coastal Herb62. Pandanus ECF96. Upland field
29. Cornus DBF63. Pasture97. Water
30. Costal Shrub64. Picea ECF98. Weigela Shrub
31. Cryptomeria ECF65. Pinus ECF99. Wetland Herb
32. Deciduous Shrub66. Pinus Shrub100. Zelkova DBF
33. Diospyros EBF67. Podocarpus ECF101. Zoysia Herb
34. Distylium EBF68. Pongamia EBF
Table
Table 2. List of spectral indexes used in the research.
Table 2. List of spectral indexes used in the research.
Spectral IndexesReference
Normalized difference vegetation index (NDVI)Rouse et al. [71]
Normalized difference water index (NDWI)McFeeters [72]
Normalized difference snow index (NDSI)Riggs et al. [73]
Land surface water index (LSWI)Chandrasekar et al. [74]
Green red vegetation index (GRVI)Falkowski et al. [75]
Red edge normalized difference vegetation index (RENDVI)Gitelson and Merzlyak [76]
Normalized inner reflectance in the green and red edge (NDVIRE)Maccioni et al. [77]
Table
Table 3. Class-wise cross-validation accuracy obtained for Yamagata prefecture.
Table 3. Class-wise cross-validation accuracy obtained for Yamagata prefecture.
LegendKappaF1-ScoreLegendKappaF1-Score
Abandoned land0.7850.786Open space Herb0.8540.855
Abies ECF0.8450.846Orchard0.7080.709
Acer DBF0.9240.929Paddy field0.8180.819
Acer Shrub0.9070.910Pasture0.9040.906
Alnus DBF0.8390.843Pinus ECF0.8490.850
Alnus Shrub0.9050.908Pinus Shrub0.9810.982
Alpine Herb0.9250.926Populus DBF1.0001.000
Alpine Shrub0.9440.944Pterocarya DBF0.8150.817
Bamboo EBF0.9430.945Quercus DBF0.7920.794
Barren0.8120.814Quercus Shrub0.8860.887
Betula DBF0.9200.924Rhododendron Shrub0.9800.980
Built-up0.9840.990Robinia DBF0.8520.859
Carpinus DBF0.8630.863Salix DBF0.7730.775
Coastal Herb0.9210.921Salix Shrub0.8370.838
Coastal Shrub0.9610.961Sasa Shrub0.8800.881
Cryptomeria ECF0.7890.791Solar panels0.9850.986
Deciduous Shrub0.9530.953Thujopsis ECF0.9660.967
Fagus DBF0.8670.868Tilia DBF0.9660.966
Fraxinus DBF0.9500.951Tsuga ECF0.8060.807
Hydrangea Shrub0.7930.795Ulmus DBF0.7500.750
Juglans DBF0.7290.732Upland field0.7930.794
Larix DCF0.8790.880Water0.8430.844
Machilus EBF0.9320.932Weigela Shrub0.8830.884
Mallotus DBF0.9350.935Wetland Herb0.8590.860
Miscanthus Herb0.7590.761Zelkova DBF0.8520.853
Table
Table 4. Summary of the cross-validation accuracies calculated as average accuracies in terms of Kappa coefficient and F1-score across all classes in each prefectural region.
Table 4. Summary of the cross-validation accuracies calculated as average accuracies in terms of Kappa coefficient and F1-score across all classes in each prefectural region.
PrefecturesClassKappaF1-ScorePrefecturesClassKappaF1-Score
Aichi370.7840.786Miyagi500.8060.808
Akita510.8280.831Miyazaki490.8190.821
Aomori470.8140.817Nagano520.7880.791
Chiba320.7680.771Nagasaki510.8340.836
Ehime440.7950.797Nara420.8070.810
Fukui410.8160.820Niigata530.8640.865
Fukuoka380.7660.768Oita460.8000.802
Fukushima510.8190.822Okayama410.7990.802
Gifu510.8120.814Okinawa430.8020.805
Gunma480.7900.792Osaka290.7940.797
Hiroshima450.8170.820Saga370.8180.821
HokkaidoA440.8190.822Saitama390.7590.762
HokkaidoB400.8360.838Shiga420.8320.835
Hyogo490.7990.801Shimane440.8130.816
Ibaraki400.7420.744Shizuoka500.7530.755
Ishikawa450.8110.814Tochigi490.8190.821
Iwate480.8200.822Tokushima480.7840.786
Kagawa350.7090.711Tokyo460.8250.828
Kagoshima620.8090.812Tottori390.8110.814
Kanagawa430.7320.734Toyama510.8190.822
Kochi480.8050.808Wakayama370.7620.765
Kumamoto440.8170.819Yamagata500.8740.875
Kyoto410.8380.840Yamaguchi420.7900.792
Mie420.8080.81Yamanashi430.7790.781
Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Share and Cite

MDPI and ACS Style

Sharma, R.C. Countrywide Mapping of Plant Ecological Communities with 101 Legends including Land Cover Types for the First Time at 10 m Resolution through Convolutional Learning of Satellite Images. Appl. Sci. 2022, 12, 7125. https://doi.org/10.3390/app12147125

AMA Style

Sharma RC. Countrywide Mapping of Plant Ecological Communities with 101 Legends including Land Cover Types for the First Time at 10 m Resolution through Convolutional Learning of Satellite Images. Applied Sciences. 2022; 12(14):7125. https://doi.org/10.3390/app12147125

Chicago/Turabian Style

Sharma, Ram C. 2022. "Countrywide Mapping of Plant Ecological Communities with 101 Legends including Land Cover Types for the First Time at 10 m Resolution through Convolutional Learning of Satellite Images" Applied Sciences 12, no. 14: 7125. https://doi.org/10.3390/app12147125

APA Style

Sharma, R. C. (2022). Countrywide Mapping of Plant Ecological Communities with 101 Legends including Land Cover Types for the First Time at 10 m Resolution through Convolutional Learning of Satellite Images. Applied Sciences, 12(14), 7125. https://doi.org/10.3390/app12147125

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop