[go: up one dir, main page]
More Web Proxy on the site http://driver.im/
You seem to have javascript disabled. Please note that many of the page functionalities won't work as expected without javascript enabled.
 
 
remotesensing-logo

Journal Browser

Journal Browser

Remote Sensing for Management of Invasive Species

A special issue of Remote Sensing (ISSN 2072-4292). This special issue belongs to the section "Biogeosciences Remote Sensing".

Deadline for manuscript submissions: 30 June 2025 | Viewed by 10135

Special Issue Editor


E-Mail Website
Guest Editor
Manaaki Whenua—Landcare Research, Palmerston North 4472, New Zealand
Interests: environmental monitoring; remote sensing; environmental modelling

Special Issue Information

Dear Colleagues,

Invasive species have devastating effects on ecosystems and economies. They are alien plants, animals, or micro-organisms that are harmful to ecosystems through excessive success in distribution. Often, they are spread by human activities, intentionally or unintentionally, but climate change can also be responsible by giving certain species advantages, which makes them become aggressive and invasive. Spatial information on the status of invasion is essential for understanding drivers and guiding management response, such as prevention, eradication, or control. Remote sensing can be used to map and monitor the spread of invasive species and impact, but it has been challenging as invasive species and impacts are often difficult to detect from space. The advent of new data and methods is improving utility, especially integration with ground surveillance and policy response.

The aim of this Special Issue on “Remote Sensing for Management of Invasive Species” is to bring together recent advances in the field of remote sensing for application to invasive species management. Not only are new remotely sensed data and new analysis methods being used to map more accurate and cost-effective maps of invasive species and their environmental impacts, but new approaches are being developed for integrating remotely sensed data with ground data to provide response managers with more useful information. As such, the themes of this Special Issue will not only cover new remote sensing methods, but also how those methods can provide useful information for practical management of biological invasions.

Contributions focusing on the following themes are welcome for this Special Issue:

  • Mapping and monitoring invasive species;
  • Mapping and monitoring impact of invasive species;
  • Monitoring and predicting distribution of invasive species;
  • The use of remotely sensed information for helping manage response to biological invasions;
  • Integration of remotely sensed data with ground data to help manage biological invasions;
  • New data and methods for detection of invasive species.

Dr. John Dymond
Guest Editor

Manuscript Submission Information

Manuscripts should be submitted online at www.mdpi.com by registering and logging in to this website. Once you are registered, click here to go to the submission form. Manuscripts can be submitted until the deadline. All submissions that pass pre-check are peer-reviewed. Accepted papers will be published continuously in the journal (as soon as accepted) and will be listed together on the special issue website. Research articles, review articles as well as short communications are invited. For planned papers, a title and short abstract (about 100 words) can be sent to the Editorial Office for announcement on this website.

Submitted manuscripts should not have been published previously, nor be under consideration for publication elsewhere (except conference proceedings papers). All manuscripts are thoroughly refereed through a single-blind peer-review process. A guide for authors and other relevant information for submission of manuscripts is available on the Instructions for Authors page. Remote Sensing is an international peer-reviewed open access semimonthly journal published by MDPI.

Please visit the Instructions for Authors page before submitting a manuscript. The Article Processing Charge (APC) for publication in this open access journal is 2700 CHF (Swiss Francs). Submitted papers should be well formatted and use good English. Authors may use MDPI's English editing service prior to publication or during author revisions.

Keywords

  • invasive species and pathogens
  • remote sensing
  • machine learning
  • mapping invasive species
  • monitoring environmental impact
  • species distribution modelling
  • invasive species management
  • biocontrol
  • pest and weed control

Benefits of Publishing in a Special Issue

  • Ease of navigation: Grouping papers by topic helps scholars navigate broad scope journals more efficiently.
  • Greater discoverability: Special Issues support the reach and impact of scientific research. Articles in Special Issues are more discoverable and cited more frequently.
  • Expansion of research network: Special Issues facilitate connections among authors, fostering scientific collaborations.
  • External promotion: Articles in Special Issues are often promoted through the journal's social media, increasing their visibility.
  • e-Book format: Special Issues with more than 10 articles can be published as dedicated e-books, ensuring wide and rapid dissemination.

Further information on MDPI's Special Issue polices can be found here.

Published Papers (8 papers)

Order results
Result details
Select all
Export citation of selected articles as:

Research

Jump to: Other

29 pages, 44436 KiB  
Article
Pragmatically Mapping Phragmites with Unoccupied Aerial Systems: A Comparison of Invasive Species Land Cover Classification Using RGB and Multispectral Imagery
by Alexandra Danielle Evans, Jennifer Cramer, Victoria Scholl and Erika Lentz
Remote Sens. 2024, 16(24), 4691; https://doi.org/10.3390/rs16244691 - 16 Dec 2024
Viewed by 502
Abstract
Unoccupied aerial systems (UASs) are increasingly being deployed in coastal environments to rapidly map and monitor changes to geomorphology, vegetation, and infrastructure, particularly in difficult to access areas. UAS data, relative to airplane or satellite data, typically have higher spatial resolution, sensor customization, [...] Read more.
Unoccupied aerial systems (UASs) are increasingly being deployed in coastal environments to rapidly map and monitor changes to geomorphology, vegetation, and infrastructure, particularly in difficult to access areas. UAS data, relative to airplane or satellite data, typically have higher spatial resolution, sensor customization, and increased flexibility in temporal resolution, which benefits monitoring applications. UAS data have been used to map and monitor invasive species occurrence and expansion, such as Phragmites australis, a reed species in wetlands throughout the eastern United States. To date, the work on this species has been largely opportunistic or ad hoc. Here, we statistically and qualitatively compare results from several sensors and classification workflows to develop baseline understanding of the accuracy of different approaches used to map Phragmites. Two types of UAS imagery were collected in a Phragmites-invaded salt marsh setting—natural color red-green-blue (RGB) imagery and multispectral imagery spanning visible and near infrared wavelengths. We evaluated whether one imagery type provided significantly better classification results for mapping land cover than the other, also considering trade-offs like overall accuracy, financial costs, and effort. We tested the transferability of classification workflows that provided the highest thematic accuracy to another barrier island environment with known Phragmites stands. We showed that both UAS sensor types were effective in classifying Phragmites cover, with neither resulting in significantly better classification results than the other for Phragmites detection (overall accuracy up to 0.95, Phragmites recall up to 0.86 at the pilot study site). We also found the highest accuracy workflows were transferrable to sites in a barrier island setting, although the quality of results varied across these sites (overall accuracy up to 0.97, Phragmites recall up to 0.90 at the additional study sites). Full article
(This article belongs to the Special Issue Remote Sensing for Management of Invasive Species)
Show Figures

Graphical abstract

Graphical abstract
Full article ">Figure 1
<p>Flowchart illustrating the overall study design.</p>
Full article ">Figure 2
<p>(<b>A</b>) Regional map of Cape Cod showing the location of Dog Head Marsh in Mashpee, MA, USA. (<b>B</b>) Study site orthomosaic from the uncrewed aerial system RGB sensor. The yellow box in (<b>B</b>) indicates the study area.</p>
Full article ">Figure 3
<p>Locations of the five study sites in the northern section of Assateague National Seashore (Maryland, USA). Zoomed-in views of clipped orthomosaics made from the RGB imagery are labeled and shown to the right and bottom of the “All Sites” panel.</p>
Full article ">Figure 4
<p>(<b>A</b>) A section of the RGB orthomosaic for ASIS Site 3. The yellow extent indicator corresponds to panel (<b>B</b>). (<b>B</b>) A zoomed-in view of the orthomosaic, illustrating how individual <span class="html-italic">Phragmites</span> plants, both with and without plumes, can be visually identified in the imagery. <span class="html-italic">Phragmites</span> can also be visually distinguished from similar looking species, such as <span class="html-italic">Baccharis halimifolia</span> (a.k.a. “Groundsel”). (<b>C</b>) The same geographic extent shown in (<b>B</b>) but showing the digital surface model rather than the RGB orthomosaic.</p>
Full article ">Figure 5
<p>Feature importance metrics (mean decrease in impurity, “MDI”, and permutation on full model) from the (<b>A</b>) RGB and (<b>B</b>) multispectral random forest classifications when using the “All Features” featuresets from Dog Head Marsh. Feature names are as initially given herein except “Corr.” is correlation, “C.S.” is cluster shadow, “C.P.” is cluster prominence, “H.C.” is Haralick correlation, and “DSM Rug.” is DSM rugosity. The blue bars are the feature importances of the forest, along with their inter-tree variability represented by the error bars.</p>
Full article ">Figure 6
<p>(<b>A</b>) The RGB orthomosaic for visual context. The “best” classified results (bold rows in <a href="#remotesensing-16-04691-t004" class="html-table">Table 4</a>) for the (<b>B</b>) RGB sensor and (<b>C</b>) multispectral sensor. (<b>D</b>) Zoomed-in view of an area containing <span class="html-italic">Phragmites</span>, other vegetation, and not vegetation, the extent of which is shown in (<b>A</b>) in pink. (<b>E</b>) The RGB classification of the area shown in (<b>D</b>). (<b>F</b>) The multispectral classification of the area shown in (<b>D</b>).</p>
Full article ">Figure 7
<p>A comparison of select accuracy metrics calculated per sensor per Assateague (ASIS) site number. Recall is important when the cost of a false negative is high, as in invasive species management (e.g., not classifying <span class="html-italic">Phragmites</span> cover where there is some). Metric values closer to 1 indicate more reliable results.</p>
Full article ">Figure 8
<p>Bar plots showing example important features identified by the random forest algorithms trained and applied per Assateague (ASIS) site according to mean decrease in impurity and mean accuracy decrease metrics. These examples illustrate how the identified important features varied across sites and sensor types. The blue bars are the feature importances of the forest, along with their inter-tree variability represented by the error bars. Larger blue bars indicate higher importance. (<b>A</b>) RGB “All Uncorrelated Features” featureset. (<b>B</b>) Multispectral “All Features” featureset. Feature names are as initially given herein except “Corr.” is correlation, “C.S.” is cluster shadow, “C.P.” is cluster prominence, “H.C.” is Haralick correlation, and “DSM Rug.” is DSM rugosity.</p>
Full article ">Figure 9
<p>Classified cover maps for Assateague (ASIS) sites 1 and 2 from both the RGB and multispectral (“multi”) data. The leftmost column is the RGB orthomosaic for each site, the middle column contains the RGB results, and the rightmost column is the multispectral results. The bottom two rows of panels show zoomed-in views of the sites along with their classified results. The extents of these subsets are shown in the “ASIS 1” and “ASIS 2” panels with pink and blue indicators, respectively.</p>
Full article ">Figure 10
<p>An example of how smoothing via majority filter impacts cover classification appearance. (<b>A</b>) The RGB orthomosaic for Assateague (ASIS) site 1 with the main <span class="html-italic">Phragmites</span> stand, example Groundsel shrubs, and example bare branches annotated. (<b>B</b>) The original classified results from the RGB data using the “All Uncorrelated Features” featureset. (<b>C</b>) The classified results from (<b>B</b>) smoothed with a size 3 majority filter. (<b>D</b>) The classified results from (<b>B</b>) smoothed with a size 5 majority filter.</p>
Full article ">Figure 11
<p><span class="html-italic">Phragmites</span> precision and <span class="html-italic">Phragmites</span> recall across degrees of filtering for the “best” Assateague (ASIS) site results (<b>ASIS 2</b>) and the “worst” ASIS site results (<b>ASIS 1</b>). A metric value closer to 1 indicates more reliable results. “Filtering” refers to a majority filter and “size” refers to the size of the filter in number of pixels.</p>
Full article ">
16 pages, 45241 KiB  
Article
Classifying Serrated Tussock Cover from Aerial Imagery Using RGB Bands, RGB Indices, and Texture Features
by Daniel Pham, Deepak Gautam and Kathryn Sheffield
Remote Sens. 2024, 16(23), 4538; https://doi.org/10.3390/rs16234538 - 4 Dec 2024
Viewed by 482
Abstract
Monitoring the location and severity of invasive plant infestations is critical to the management of their spread. Remote sensing can be an effective tool for mapping invasive plants due to its capture speed, continuous coverage, and low cost, compared to ground-based surveys. Serrated [...] Read more.
Monitoring the location and severity of invasive plant infestations is critical to the management of their spread. Remote sensing can be an effective tool for mapping invasive plants due to its capture speed, continuous coverage, and low cost, compared to ground-based surveys. Serrated tussock (Nassella trichotoma) is a highly problematic invasive plant in Victoria, Australia, as it competes with the species in the communities that it invades. In this study, a workflow was developed and assessed for classifying the cover of serrated tussock in a mix of grazing pastures and grasslands. Using high-resolution RGB aerial imagery and vegetation field survey plots, random forest models were trained to classify the plots based on their fractional coverage of serrated tussock. Three random forest classifiers were trained by utilising spectral features (RGB bands and indices), texture features derived from the Grey-Level Co-occurrence Matrix, and a combination of all the features. The model trained on all the features achieved an overallaccuracy of 67% and a kappa score of 0.52 against a validation dataset. Plots with high and low infestation levels were classified more accurately than plots with moderate or no infestation. Notably, texture features proved more effective than spectral features for classification. The developed random forest model can be used for producing classified maps to depict the spatial distribution of serrated tussock infestation, thus supporting land managers in managing the infestation. Full article
(This article belongs to the Special Issue Remote Sensing for Management of Invasive Species)
Show Figures

Figure 1

Figure 1
<p>The three study sites (Cowies, Glengowie, and Forsyth) within the Western Grassland Reserve (denoted by line), Victoria, were used to investigate the Serrated Tussock classification workflow.</p>
Full article ">Figure 2
<p>The rapid plot assessment data (overlayed on aerial imagery in Forsyth for visualisation) consisted of the percentage ground cover of Serrated Tussock.</p>
Full article ">Figure 3
<p>Optimisation of GLCM pixel offset from 1 to 10 through classification <math display="inline"><semantics> <mrow> <mi>k</mi> <mi>a</mi> <mi>p</mi> <mi>p</mi> <mi>a</mi> </mrow> </semantics></math> score. The optimal pixel offset is used in all subsequent classifications.</p>
Full article ">Figure 4
<p>Cumulative explained variance by the number of principal components. Only the first twelve principal components are displayed for clarity.</p>
Full article ">Figure 5
<p>Comparison of the performance metrics (<math display="inline"><semantics> <mrow> <mi>o</mi> <mi>v</mi> <mi>e</mi> <mi>r</mi> <mi>a</mi> <mi>l</mi> <mi>l</mi> <mi>a</mi> <mi>c</mi> <mi>c</mi> <mi>u</mi> <mi>r</mi> <mi>a</mi> <mi>c</mi> <mi>y</mi> </mrow> </semantics></math>, <math display="inline"><semantics> <mrow> <mi>k</mi> <mi>a</mi> <mi>p</mi> <mi>p</mi> <mi>a</mi> </mrow> </semantics></math> score, and <math display="inline"><semantics> <mrow> <mi>F</mi> <mn>1</mn> </mrow> </semantics></math> score) between the three models (spectral, texture, and combined) under investigation.</p>
Full article ">Figure 6
<p>Confusion Matrices for the three random forest models : (<b>a</b>) Spectral model (trained on spectral features only); (<b>b</b>) Texture model (trained on texture features only); (<b>c</b>) Combined model (trained on all features). The true class is listed on the y-axis, and the predicted class is on the x-axis.</p>
Full article ">Figure 7
<p>Full aerial image of Cowies site classified using the Combined random forest model. The background colour represents the predicted coverage class, while the rapid plots (true classification) are overlaid on top of the image.</p>
Full article ">
17 pages, 4410 KiB  
Article
Assessing Golden Tides from Space: Meteorological Drivers in the Accumulation of the Invasive Algae Rugulopteryx okamurae on Coasts
by Sara Haro, Liam Morrison, Isabel Caballero, Félix L. Figueroa, Nathalie Korbee, Gabriel Navarro and Ricardo Bermejo
Remote Sens. 2024, 16(15), 2689; https://doi.org/10.3390/rs16152689 - 23 Jul 2024
Cited by 1 | Viewed by 1153
Abstract
Massive accumulations of invasive brown algae Rugulopteryx okamurae are exacerbating environmental and socio-economic issues on the Mediterranean and potentially Atlantic coasts. These golden tides, likely intensified by global change processes such as changes in wind direction and intensity and rising temperatures, pose increasing [...] Read more.
Massive accumulations of invasive brown algae Rugulopteryx okamurae are exacerbating environmental and socio-economic issues on the Mediterranean and potentially Atlantic coasts. These golden tides, likely intensified by global change processes such as changes in wind direction and intensity and rising temperatures, pose increasing challenges to coastal management. This study employs the Normalized Difference Vegetation Index (NDVI), with values above 0.08 from Level-2 Sentinel-2 imagery, to effectively monitor these strandings along the coastline of Los Lances beach (Tarifa, Spain) in the Strait of Gibraltar Natural Park from 2018 to 2022. Los Lances beach is one of the most affected by the R. okamurae bioinvasion in Spain. The analysis reveals that wind direction determines the spatial distribution of biomass accumulated on the shore. The highest average NDVI values in the western patch were observed with south-easterly winds, while in the eastern patch, higher average NDVI values were recorded with south-westerly, westerly and north-westerly winds. The maximum coverage correlates with elevated temperatures and minimal rainfall, peaking between July and October. Leveraging these insights, we propose a replicable methodology for the early detection and strategic pre-shore collection of biomass, which could facilitate efficient coastal cleanup strategies and enhance biomass utility for biotechnological applications. This approach promises cost-effective adaptability across different geographic areas impacted by golden tides. Full article
(This article belongs to the Special Issue Remote Sensing for Management of Invasive Species)
Show Figures

Graphical abstract

Graphical abstract
Full article ">Figure 1
<p>Location of Los Lances Beach (Tarifa, Cadiz, SW Spain) in the Natural Park of the Strait of Gibraltar, one of the areas most affected by <span class="html-italic">R. okamurae</span> strandings. The study area, i.e., the coastline, is marked with a discontinuous black line. The Level 2 Sentinel-2 True Color image, acquired on 2 August 2019, shows the strandings at the eastern ends of the beach. A photo of <span class="html-italic">R. okamurae</span> strandings in that area is also included. The meteorological station is indicated with a red rhombus.</p>
Full article ">Figure 2
<p>Flowchart showing the proposed methodology to monitor <span class="html-italic">R. okamurae</span> washed ashore. This comprises five steps: (step 1) visualization and selection of Sentinel-2 imagery to be processed; (step 2) creation of the area of interest, i.e., along the entire coastline; (step 3) processing of Sentinel-2 images using Google Earth Engine; (step 4) selection of the largest patches with seaweed accumulations; and (step 5) calculating coverage and average NDVI for the entire coastline and for the largest extension patches using geographic information system (GIS); where start (is plotted in blue), actions (in green), settings (in pink), inputs (in purple) and results (in orange).</p>
Full article ">Figure 3
<p>Average coverage and NDVI (NDVI &gt; 0.08) for <span class="html-italic">R. okamurae</span> along Los Lances beach (Tarifa, Andalusia, Spain) from 130 Level-2 Sentinel-2 images (2018–2022). The coastline is marked with a discontinuous black line, with the maximum extension areas at the beach ends (east and west patches) indicated by a continuous black line.</p>
Full article ">Figure 4
<p>(<b>a</b>,<b>b</b>) Box plots depicting the average monthly and annual coverage of <span class="html-italic">R. okamurae</span> along the coastline from 2018 to 2022. (<b>c</b>,<b>d</b>) Box plots depicting the average monthly and annual NDVI (used as a proxy of <span class="html-italic">R. okamurae</span> biomass washed ashore) along the coastline from 2018 to 2022. The line within each box plot represents the median and the cross sign indicates the mean.</p>
Full article ">Figure 5
<p>(<b>a</b>–<b>d</b>) Seasonal averages of coverage and NDVI associated with golden tides of invasive <span class="html-italic">R. okamurae</span> at the western end of Los Lances Beach (Tarifa, Andalusia, Spain). (<b>e</b>–<b>h</b>) Seasonal averages of coverage and NDVI associated with golden tides of invasive <span class="html-italic">R. okamurae</span> at the eastern end of Los Lances Beach (Tarifa, Andalusia, Spain). Averages were calculated using 22 autumn, 16 winter, 33 spring, and 59 summer Sentinel-2 images taken between 2018 and 2022. The seasons are defined as follows: autumn (21 September–21 December), winter (21 December–21 March), spring (21 March–21 June), and summer (21 June–21 September).</p>
Full article ">Figure 6
<p>(<b>a</b>,<b>b</b>) Box plots showing monthly and annual averages of <span class="html-italic">R. okamurae</span> coverage for the eastern and western patches between 2018 and 2022. (<b>c</b>,<b>d</b>) Box plots showing monthly and annual averages of NDVI (used as a proxy of <span class="html-italic">R. okamurae</span> biomass washed ashore) for the eastern and western patches between 2018 and 2022. The east patch is plotted in orange, and the west patch in blue. The line within each box plot represents the median and the cross sign indicates the mean.</p>
Full article ">Figure 7
<p>(<b>a</b>,<b>b</b>) Monthly average coverage of <span class="html-italic">R. okamurae</span> (ha) washed ashore for the entire coastline at Los Lances beach, compared with the monthly average air temperature (°C) and monthly accumulated precipitation (mm). Error bars indicate the standard error. (<b>c</b>,<b>d</b>) Box plots display the average coverage and NDVI of <span class="html-italic">R. okamurae</span> for both patches according to wind direction (north, N; northeast, NE; east, E; southeast, SE; south, S; southwest, SW; west, W; northwest, NW). The line within each box plot represents the median and the cross sign indicates the mean.</p>
Full article ">
18 pages, 11216 KiB  
Article
Remote Sensing Guides Management Strategy for Invasive Legumes on the Central Plateau, New Zealand
by Paul G. Peterson, James D. Shepherd, Richard L. Hill and Craig I. Davey
Remote Sens. 2024, 16(13), 2503; https://doi.org/10.3390/rs16132503 - 8 Jul 2024
Cited by 1 | Viewed by 757
Abstract
Remote sensing was used to map the invasion of yellow-flowered legumes on the Central Plateau of New Zealand to inform weed management strategy. The distributions of Cytisus scoparius (broom), Ulex europaeus (gorse) and Lupinus arboreus (tree lupin) were captured with high-resolution RGB photographs [...] Read more.
Remote sensing was used to map the invasion of yellow-flowered legumes on the Central Plateau of New Zealand to inform weed management strategy. The distributions of Cytisus scoparius (broom), Ulex europaeus (gorse) and Lupinus arboreus (tree lupin) were captured with high-resolution RGB photographs of the plants while flowering. The outcomes of herbicide operations to control C. scoparius and U. europaeus over time were also assessed through repeat photography and change mapping. A grid-square sampling tool previously developed by Manaaki Whenua—Landcare Research was used to help transfer data rapidly from photography to maps using manual classification. Artificial intelligence was trialled and ruled out because the number of false positives could not be tolerated. Future actions to protect the natural values and vistas of the Central Plateau from legume invasion were identified. While previous control operations have mostly targeted large, highly visible legume patches, the importance of removing outlying plants to prevent the establishment of new seed banks and slow spread has been underestimated. Outliers not only establish new, large, long-lived seed banks in previously seed-free areas, but they also contribute more to range expansion than larger patches. Our C. scoparius and U. europaeus change mapping confirms and helps to visualise the establishment and expansion of uncontrolled outliers. The power of visualizing weed control strategies through remote sensing has supported recommendations to improve outlier control to achieve long-term, sustainable landscape-scale suppression of invasive legumes. Full article
(This article belongs to the Special Issue Remote Sensing for Management of Invasive Species)
Show Figures

Figure 1

Figure 1
<p>The 23,000 ha area of interest (AOI) for control of <span class="html-italic">C. scoparius</span>, <span class="html-italic">U. europaeus</span> and <span class="html-italic">L. arboreus</span> (dotted red lines) with adjacent map of NZ showing location (red shading).</p>
Full article ">Figure 2
<p>Example of a successful <span class="html-italic">U. europaeus</span> control operation between 2012 (<b>a</b>) and 2016 (<b>b</b>). Red circle shows presence of <span class="html-italic">U. europaeus</span>.</p>
Full article ">Figure 3
<p>Orthorectified aerial photography mosaics of flowering invasive legumes in the AOI between 2012 and 2022.</p>
Full article ">Figure 4
<p>Maps of the distribution and density of <span class="html-italic">C. scoparius</span> (<b>a</b>,<b>b</b>), <span class="html-italic">U. europaeus</span> (<b>c</b>,<b>d</b>) and <span class="html-italic">L</span>. <span class="html-italic">arboreus</span> (<b>e</b>) in the AOI between 2012 and 2020.</p>
Full article ">Figure 4 Cont.
<p>Maps of the distribution and density of <span class="html-italic">C. scoparius</span> (<b>a</b>,<b>b</b>), <span class="html-italic">U. europaeus</span> (<b>c</b>,<b>d</b>) and <span class="html-italic">L</span>. <span class="html-italic">arboreus</span> (<b>e</b>) in the AOI between 2012 and 2020.</p>
Full article ">Figure 5
<p>Examples of low- (<b>a</b>), moderate- (<b>b</b>) and high-density (<b>c</b>) <span class="html-italic">U. europaeus</span> grid squares. The image on the left shows a 1 km<sup>2</sup> area alongside the Desert Road, and the image on the right is a 200 × 200 m grid square within this area. The red circles highlight <span class="html-italic">U. europaeus</span> plants.</p>
Full article ">Figure 6
<p>Flowering <span class="html-italic">C. scoparius</span> (<b>a</b>) and <span class="html-italic">L</span>. <span class="html-italic">arboreus</span> (<b>b</b>) visible in aerial photography.</p>
Full article ">Figure 7
<p>LRIS portal screenshot showing orthomosaic of photographs over the entire AOI (<b>a</b>) and zoomed-in image of a high-density <span class="html-italic">U. europaeus</span> area alongside the Desert Road (<b>b</b>).</p>
Full article ">Figure 8
<p>LRIS portal screenshot showing semi-transparent grid-density vector layer on the <span class="html-italic">U. europaeus</span> 2013 photography for easy weed target detection (red = high density, orange = moderate density, yellow = low density).</p>
Full article ">Figure 9
<p><span class="html-italic">C. scoparius</span> 2012–2016 (<b>a</b>) and <span class="html-italic">U. europaeus</span> 2013–2020 (<b>b</b>) change maps showing where weeds increased, decreased, stayed the same or were removed from grid squares.</p>
Full article ">Figure 9 Cont.
<p><span class="html-italic">C. scoparius</span> 2012–2016 (<b>a</b>) and <span class="html-italic">U. europaeus</span> 2013–2020 (<b>b</b>) change maps showing where weeds increased, decreased, stayed the same or were removed from grid squares.</p>
Full article ">Figure 10
<p>Example of a new <span class="html-italic">U. europaeus</span> outlier (red circle) present within a grid square in 2020 (<b>b</b>) but not in 2013 (<b>a</b>).</p>
Full article ">Figure 11
<p>Example of a <span class="html-italic">U. europaeus</span> control operation conducted between 2013 (<b>a</b>) and 2020 (<b>b</b>) in a grid square.</p>
Full article ">Figure 12
<p>An example of yellow reflectance (circled in red) from <span class="html-italic">Podocarpus nivalis</span> remaining in imagery taken during (<b>a</b>) December 2016 and (<b>b</b>) October 2020.</p>
Full article ">
24 pages, 37691 KiB  
Article
African Lovegrass Segmentation with Artificial Intelligence Using UAS-Based Multispectral and Hyperspectral Imagery
by Pirunthan Keerthinathan, Narmilan Amarasingam, Jane E. Kelly, Nicolas Mandel, Remy L. Dehaan, Lihong Zheng, Grant Hamilton and Felipe Gonzalez
Remote Sens. 2024, 16(13), 2363; https://doi.org/10.3390/rs16132363 - 27 Jun 2024
Viewed by 1160
Abstract
The prevalence of the invasive species African Lovegrass (Eragrostis curvula, ALG thereafter) in Australian landscapes presents significant challenges for land managers, including agricultural losses, reduced native species diversity, and heightened bushfire risks. Uncrewed aerial system (UAS) remote sensing combined with AI [...] Read more.
The prevalence of the invasive species African Lovegrass (Eragrostis curvula, ALG thereafter) in Australian landscapes presents significant challenges for land managers, including agricultural losses, reduced native species diversity, and heightened bushfire risks. Uncrewed aerial system (UAS) remote sensing combined with AI algorithms offer a powerful tool for accurately mapping the spatial distribution of invasive species and facilitating effective management strategies. However, segmentation of vegetations within mixed grassland ecosystems presents challenges due to spatial heterogeneity, spectral similarity, and seasonal variability. The performance of state-of-the-art artificial intelligence (AI) algorithms in detecting ALG in the Australian landscape remains unknown. This study compared the performance of four supervised AI models for segmenting ALG using multispectral (MS) imagery at four sites and developed segmentation models for two different seasonal conditions. UAS surveys were conducted at four sites in New South Wales, Australia. Two of the four sites were surveyed in two distinct seasons (flowering and vegetative), each comprised of different data collection settings. A comparative analysis was also conducted between hyperspectral (HS) and MS imagery at a single site within the flowering season. Of the five AI models developed (XGBoost, RF, SVM, CNN, and U-Net), XGBoost and the customized CNN model achieved the highest validation accuracy at 99%. The AI model testing used two approaches: quadrat-based ALG proportion prediction for mixed environments and pixel-wise classification in masked regions where ALG and other classes could be confidently differentiated. Quadrat-based ALG proportion ground truth values were compared against the prediction for the custom CNN model, resulting in 5.77% and 12.9% RMSE for the seasons, respectively, emphasizing the superiority of the custom CNN model over other AI algorithms. The comparison of the U-Net demonstrated that the developed CNN effectively captures ALG without requiring the more intricate architecture of U-Net. Masked-based testing results also showed higher F1 scores, with 91.68% for the flowering season and 90.61% for the vegetative season. Models trained on single-season data exhibited decreased performance when evaluated on data from a different season with varying collection settings. Integrating data from both seasons during training resulted in a reduction in error for out-of-season predictions, suggesting improved generalizability through multi-season data integration. Moreover, HS and MS predictions using the custom CNN model achieved similar test results with around 20% RMSE compared to the ground truth proportion, highlighting the practicality of MS imagery over HS due to operational limitations. Integrating AI with UAS for ALG segmentation shows great promise for biodiversity conservation in Australian landscapes by facilitating more effective and sustainable management strategies for controlling ALG spread. Full article
(This article belongs to the Special Issue Remote Sensing for Management of Invasive Species)
Show Figures

Figure 1

Figure 1
<p>Overview of the study methodology, illustrating the key steps in data acquisition, data preprocessing, pixel-wise labeling, multispectral-based prediction, and multispectral and hyperspectral comparison.</p>
Full article ">Figure 2
<p>Map of the study sites. Site 1 and Site 2 correspond to Bunyan sites, while Site 3 and Site 4 correspond to Cooma sites, located in in New South Wales, Australia.</p>
Full article ">Figure 3
<p>Illustration of quadrat species diversity at Bunyan and Cooma sites.</p>
Full article ">Figure 4
<p>Labeled polygons of three randomly selected quadrats from Sites 1 and 4, along with their corresponding close-up images.</p>
Full article ">Figure 5
<p>Spectral signature differences for spectral indices.</p>
Full article ">Figure 6
<p>Modelling and augmentation of data points during ALG model development.</p>
Full article ">Figure 7
<p>Custom CNN model architecture for MS-based ALG segmentation. For MS and HS comparison, the third dimension of the first layer captures the channel depth, which is 5 for MS and 448 for HS imagery. The remaining dimensions are unchanged.</p>
Full article ">Figure 8
<p>U-Net architecture used for ALG classification.</p>
Full article ">Figure 9
<p>Multispectral-based prediction maps of three quadrats from test sites using the models developed from the combined seasonal dataset. The filled black regions represent the ALG.</p>
Full article ">Figure 10
<p>The ground truth and the predicted ALG proportion from the Bunyan test site (flowering) using the models developed from the combined seasonal dataset.</p>
Full article ">Figure 11
<p>The ground truth and the predicted ALG proportion from the Cooma test site (vegetative) using the models developed from the combined seasonal dataset.</p>
Full article ">Figure 12
<p>Multispectral-based segmented ALG spatial distribution map of test sites. (<b>a</b>) Cooma site; (<b>b</b>) Bunyan site. The hashed black polygon represents the ALG detected region.</p>
Full article ">Figure 13
<p>The ground truth and the predicted ALG proportion of the quadrats from the test region of Site 2 by the custom CNN model.</p>
Full article ">Figure 14
<p>Comparison of multispectral and hyperspectral imagery-based models prediction maps of three quadrats from test sites. The filled black regions represent the ALG.</p>
Full article ">
30 pages, 14644 KiB  
Article
Integrating Artificial Intelligence and UAV-Acquired Multispectral Imagery for the Mapping of Invasive Plant Species in Complex Natural Environments
by Narmilan Amarasingam, Fernando Vanegas, Melissa Hele, Angus Warfield and Felipe Gonzalez
Remote Sens. 2024, 16(9), 1582; https://doi.org/10.3390/rs16091582 - 29 Apr 2024
Cited by 5 | Viewed by 2254
Abstract
The proliferation of invasive plant species poses a significant ecological threat, necessitating effective mapping strategies for control and conservation efforts. Existing studies employing unmanned aerial vehicles (UAVs) and multispectral (MS) sensors in complex natural environments have predominantly relied on classical machine learning (ML) [...] Read more.
The proliferation of invasive plant species poses a significant ecological threat, necessitating effective mapping strategies for control and conservation efforts. Existing studies employing unmanned aerial vehicles (UAVs) and multispectral (MS) sensors in complex natural environments have predominantly relied on classical machine learning (ML) models for mapping plant species in natural environments. However, a critical gap exists in the literature regarding the use of deep learning (DL) techniques that integrate MS data and vegetation indices (VIs) with different feature extraction techniques to map invasive species in complex natural environments. This research addresses this gap by focusing on mapping the distribution of the Broad-leaved pepper (BLP) along the coastal strip in the Sunshine Coast region of Southern Queensland in Australia. The methodology employs a dual approach, utilising classical ML models including Random Forest (RF), eXtreme Gradient Boosting (XGBoost), and Support Vector Machine (SVM) in conjunction with the U-Net DL model. This comparative analysis allows for an in-depth evaluation of the performance and effectiveness of both classical ML and advanced DL techniques in mapping the distribution of BLP along the coastal strip. Results indicate that the DL U-Net model outperforms classical ML models, achieving a precision of 83%, recall of 81%, and F1–score of 82% for BLP classification during training and validation. The DL U-Net model attains a precision of 86%, recall of 76%, and F1–score of 81% for BLP classification, along with an Intersection over Union (IoU) of 68% on the separate test dataset not used for training. These findings contribute valuable insights to environmental conservation efforts, emphasising the significance of integrating MS data with DL techniques for the accurate mapping of invasive plant species. Full article
(This article belongs to the Special Issue Remote Sensing for Management of Invasive Species)
Show Figures

Figure 1

Figure 1
<p>Processing pipeline for BLP identification using UAV-based spectral data and AI.</p>
Full article ">Figure 2
<p>Study site of coastal strip from Point Cartwright to Wurtulla.</p>
Full article ">Figure 3
<p>BLP ground truth locations at Wurtulla site.</p>
Full article ">Figure 4
<p>MicaSense Altum processing workflow for multispectral orthomosaic generation (including Reflectance Calibration) in Agisoft Metashape Professional.</p>
Full article ">Figure 5
<p>Outline of processing pipeline for georeferencing of high-resolution RGB and multispectral imagery.</p>
Full article ">Figure 6
<p>Ground truth labeling of BLP at the Bokarina site, highlighting BLP population in red within a small region highlighting in yellow dotted lines.</p>
Full article ">Figure 7
<p>Processing pipeline for classical machine learning model training for BLP classification.</p>
Full article ">Figure 8
<p>Processing pipeline for DL U-Net model training for BLP classification.</p>
Full article ">Figure 9
<p>U-Net architecture used for BLP classification.</p>
Full article ">Figure 10
<p>Spectral signature differences for various VIs.</p>
Full article ">Figure 11
<p>Spectral signature difference for spectral bands.</p>
Full article ">Figure 12
<p>Variable importance factor (VIF) values for different VIs.</p>
Full article ">Figure 13
<p>Correlation heatmap for different VIs used in this study.</p>
Full article ">Figure 14
<p>BLP prediction outcomes generated by RF model at region of interest in Bokarina study site: yellow polygons represent ground truth labels, regions highlighted in red denote prediction outcomes (<b>a</b>) training phase, (<b>b</b>) testing phase.</p>
Full article ">Figure 15
<p>Visual representation of the RGB regions of interest with respective labelled mask, multispectral, and selected VIs.</p>
Full article ">Figure 16
<p>Comparison of U-Net model performance (precision (P) and recall (R) for BLP) using different feature sets in testing dataset.</p>
Full article ">Figure 17
<p>U-Net BLP model training plot: (<b>a</b>) accuracy, (<b>b</b>) precision (BLP).</p>
Full article ">Figure 18
<p>K-fold cross-validation for BLP U-Net model: (<b>a</b>) accuracy, (<b>b</b>) loss metrics.</p>
Full article ">Figure 19
<p>BLP prediction outcomes generated by U-Net model showcasing selected ROIs from testing phase. Representation includes RGB images, MS views, corresponding labelled masks, and U-Net model predictions.</p>
Full article ">Figure 20
<p>U-Net predictions, encompassing high-resolution RGB image of entire study area of Bokarina2 site overlaid on Google Earth satellite image. (<b>a</b>) U-Net predictions depicted with BLP highlighted in red, (<b>b</b>) offers zoomed-in view of prediction, providing closer examination of model’s output at a more detailed level.</p>
Full article ">Figure 21
<p>Integration of Google Earth satellite image depicting site within Bokarina2 overlaid with BLP predictions highlighted in red, illustrating model’s spatial accuracy in identifying target species.</p>
Full article ">
56 pages, 143133 KiB  
Article
Analysis and Quantification of the Distribution of Marabou (Dichrostachys cinerea (L.) Wight & Arn.) in Valle de los Ingenios, Cuba: A Remote Sensing Approach
by Eduardo Moreno, Encarnación Gonzalez, Reinaldo Alvarez and Julio Menendez
Remote Sens. 2024, 16(5), 752; https://doi.org/10.3390/rs16050752 - 21 Feb 2024
Viewed by 1137
Abstract
Cuba is struggling with a growing environmental problem: the uncontrolled spread of the allochthonous weed species marabou (Dichrostachys cinerea) throughout the country. Over the last 70 years, marabou has become a formidable invasive species that poses a threat to Cuban biodiversity [...] Read more.
Cuba is struggling with a growing environmental problem: the uncontrolled spread of the allochthonous weed species marabou (Dichrostachys cinerea) throughout the country. Over the last 70 years, marabou has become a formidable invasive species that poses a threat to Cuban biodiversity and agricultural productivity. In this paper, we present a free and affordable method for regularly mapping the spatial distribution of the marabou based on the Google Earth Engine platform and ecological surveys. To test its accuracy, we develop an 18-year remote sensing analysis (2000–2018) of marabou dynamics using the Valle de los Ingenios, a Cuban UNESCO World Heritage Site, as an experimental model. Our spatial analysis reveals clear patterns of marabou distribution and highlights areas of concentrated growth. Temporal trends illustrate the aggressive nature of the species, identifying periods of expansion and decline. In addition, our system is able to detect specific, large-scale human interventions against the marabou plague in the area. The results highlight the urgent need for remedial strategies to maintain the fragile ecological balance in the region. Full article
(This article belongs to the Special Issue Remote Sensing for Management of Invasive Species)
Show Figures

Figure 1

Figure 1
<p>Cartographic sources provided by the University of Sancti Spiritus (Cuba).</p>
Full article ">Figure 2
<p>Work flowchart.</p>
Full article ">Figure 3
<p>Final delimitation of the Valle de los Ingenios (red area).</p>
Full article ">Figure 4
<p>(<b>a</b>) Landsat 8 NDVI time series (the red line represents NDVI mean value). (<b>b</b>) Harmonic model.</p>
Full article ">Figure 5
<p>Example of Landsat 8 original and harmonic model fitted values for marabou (year 2014).</p>
Full article ">Figure 6
<p>Marabou distribution maps for 2004 (<b>a</b>), 2006 (<b>b</b>), 2016 (<b>c</b>) and 2021 (<b>d</b>), showing marabou in green.</p>
Full article ">Figure A1
<p>Marabou year 2002.</p>
Full article ">Figure A2
<p>Marabou year 2004.</p>
Full article ">Figure A3
<p>Marabou year 2006.</p>
Full article ">Figure A4
<p>Marabou year 2008.</p>
Full article ">Figure A5
<p>Marabou year 2010.</p>
Full article ">Figure A6
<p>Marabou year 2012.</p>
Full article ">Figure A7
<p>Marabou year 2014.</p>
Full article ">Figure A8
<p>Marabou year 2015.</p>
Full article ">Figure A9
<p>Marabou year 2016.</p>
Full article ">Figure A10
<p>Marabou year 2017.</p>
Full article ">Figure A11
<p>Marabou year 2018.</p>
Full article ">Figure A12
<p>Marabou year 2019.</p>
Full article ">Figure A13
<p>Marabou year 2020.</p>
Full article ">Figure A14
<p>Marabou year 2021.</p>
Full article ">

Other

Jump to: Research

16 pages, 17976 KiB  
Technical Note
Advanced Detection of Invasive Neophytes in Agricultural Landscapes: A Multisensory and Multiscale Remote Sensing Approach
by Florian Thürkow, Christopher Günter Lorenz, Marion Pause and Jens Birger
Remote Sens. 2024, 16(3), 500; https://doi.org/10.3390/rs16030500 - 28 Jan 2024
Cited by 3 | Viewed by 1338
Abstract
The sustainable provision of ecological products and services, both natural and man-made, faces a substantial threat emanating from invasive plant species (IPS), which inflict considerable economic and ecological harm on a global scale. They are widely recognized as one of the primary drivers [...] Read more.
The sustainable provision of ecological products and services, both natural and man-made, faces a substantial threat emanating from invasive plant species (IPS), which inflict considerable economic and ecological harm on a global scale. They are widely recognized as one of the primary drivers of global biodiversity decline and have become the focal point of an increasing number of studies. The integration of remote sensing (RS) and geographic information systems (GIS) plays a pivotal role in their detection and classification across a diverse range of research endeavors, emphasizing the critical significance of accounting for the phenological stages of the targeted species when endeavoring to accurately delineate their distribution and occurrences. This study is centered on this fundamental premise, as it endeavors to amass terrestrial data encompassing the phenological stages and spectral attributes of the specified IPS, with the overarching objective of ascertaining the most opportune time frames for their detection. Moreover, it involves the development and validation of a detection and classification algorithm, harnessing a diverse array of RS datasets, including satellite and unmanned aerial vehicle (UAV) imagery spanning the spectrum from RGB to multispectral and near-infrared (NIR). Taken together, our investigation underscores the advantages of employing an array of RS datasets in conjunction with the phenological stages, offering an economically efficient and adaptable solution for the detection and monitoring of invasive plant species. Such insights hold the potential to inform both present and future policymaking pertaining to the management of invasive species in agricultural and natural ecosystems. Full article
(This article belongs to the Special Issue Remote Sensing for Management of Invasive Species)
Show Figures

Figure 1

Figure 1
<p>Geographical distribution of the selected study sites across Saxony-Anhalt, illustrating the diverse range of agricultural landscapes and neophyte occurrences.</p>
Full article ">Figure 2
<p>General workflow of the classification approach.</p>
Full article ">Figure 3
<p>Classification result of a HySpex image targeting <span class="html-italic">Fallopia spec.</span> highlighted in red. Objects outlined in red represent misassignments identified as different species (trees).</p>
Full article ">Figure 4
<p>Result of the automatic segment-based classification of <span class="html-italic">Heracleum mantegazzianum</span> and <span class="html-italic">Fallopia spec.</span> in the Wülperode study area from a Gyrocopter HySpex dataset.</p>
Full article ">Figure 5
<p>DOP captured in May near Ebeleben in Thuringia depicting occurrences of <span class="html-italic">Bunias orientalis</span> visible as yellow spots on the left side and highlighted in red after the automated segmentation classification on the right side.</p>
Full article ">Figure 6
<p>Spectra from the RGB dataset dated 2 August 2018, for the Wittenberg study area (grayscale value determination: average values from 10,000 pixels/class).</p>
Full article ">
Back to TopTop