View in EDS HTML Full Text PDF Full Text

Leaf-Level Hyperspectral Discrimination of Wild Carrot from Co-Occurring Weeds and Hybrid Carrots Using Optimized Preprocessing and Machine Learning.

Saved in:
Bibliographic Details
Title: Leaf-Level Hyperspectral Discrimination of Wild Carrot from Co-Occurring Weeds and Hybrid Carrots Using Optimized Preprocessing and Machine Learning.
Authors: Nanayakkara, Dhanesha1,2 (AUTHOR), Bhatia, Nitin2,3 (AUTHOR), Irwin, Matthew1,3 (AUTHOR), McGill, Craig1 (AUTHOR)
Source: Remote Sensing. Jun2026, Vol. 18 Issue 12, p2013. 28p.
Subjects: Spectral reflectance, Machine learning, Plant classification, Carrots, Multispectral imaging, Remote sensing, Weed science, Spectrum analysis
Geographic Terms: New Zealand
Abstract: Highlights: What are the main findings? Leaf-level ASD hyperspectral reflectance (400–2450 nm) combined with min–max normalization and machine learning effectively discriminated wild carrot (Daucus carota subsp. carota) from 19 co-occurring weed species, four hybrid carrot varieties, and their parental lines, with over 90% accuracy in both binary and multiclass tasks using full-spectrum data. PLS-DA-derived VIP band selection identified compact sets of 10–20 diagnostic wavelengths, with full-spectrum models emphasizing the red-edge (704–730 nm) and SWIR (1390–1403 nm and 1875–1880 nm) regions, and airborne-compatible models selecting visible and red-edge bands (400–402, 527, 705–720 nm) that preserved high classification performance in detecting wild carrot while substantially reducing dimensionality. Spatial and temporal analyses indicated negligible site effects (R2 = 0.034–0.055, pseudo-F = 1.46–2.39, p > 0.05) and moderate temporal variation (R2 = 0.136–0.151, p < 0.001), confirming relatively stable wild carrot spectral signatures across locations and during the 5–10 week juvenile stage, supporting the use of pooled datasets and model generalization across space and time. What are the implications of the main findings? The identified visible and red-edge diagnostic bands for airborne-compatible models, together with supporting SWIR features from the full-spectrum analyses, provide a practical spectral basis for designing economical UAV-mounted multispectral sensors targeting wild carrot in hybrid carrot seed production landscapes, with the potential to enable site-specific weed management and reduce broadcast herbicide application. The demonstrated spectral stability of wild carrot across the three tested locations and early growth stages, together with the high classification accuracy for late-season plants of unknown age, suggests strong potential for transferring leaf-level models to field conditions and for deploying operational airborne or UAV/UGV-based wild carrot detection systems with limited recalibration. Wild carrot (Daucus carota subsp. carota), the wild relative of cultivated carrot, is globally identified as an invasive weed that threatens hybrid carrot seed production through natural cross-pollination, resulting in compromised genetic purity. Manual identification across the large areas required to ensure genetic purity in carrot seed crops is impractical. Remote sensing offers an alternative; however, morphological similarities among wild carrot, cultivated carrot, and common weeds hinder reliable detection. Early identification, however, remains essential for preventing genetic contamination. This study evaluated leaf-level hyperspectral reflectance spectroscopy (400–2450 nm) with machine learning to discriminate wild carrot from hybrid carrots, parental lines, and 19 co-occurring weed species. Spectral data from 266 wild carrot plants across three New Zealand sites and six weeks (5–10 weeks after emergence) showed negligible spatial effects (R2 = 0.034–0.055, pseudo-F = 1.46–2.39, p > 0.05) and moderate temporal variation (R2 = 0.136–0.151, pseudo-F = 5.48–6.17, p < 0.001), indicating broadly stable spectral signatures suitable for model generalization. Savitzky–Golay filtering, with min–max normalization outperformed SNV, yielding high full-spectrum accuracies for wild carrot vs. other species (90.35%, κ = 0.80), wild carrot vs. weeds (96.03%, κ = 0.92), and a multi-class model (90.79%, κ = 0.88). After removing atmospheric water-absorption bands to follow airborne sensing, reduced-band models based on airborne-compatible wavelengths maintained strong performance, including 89.40% accuracy (κ = 0.79) for wild carrot vs. weeds using a 20-band Subspace Discriminant model (400–402, 527, 705–720 nm). These findings demonstrate that stable wild carrot spectra and carefully selected visible and red-edge bands can underpin cost-effective UAV/UGV-mounted hyperspectral or multispectral sensors for site-specific wild carrot management. [ABSTRACT FROM AUTHOR]
Copyright of Remote Sensing is the property of MDPI and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)
Database: Engineering Source
Full text is not displayed to guests.
FullText Links:
  – Type: pdflink
Text:
  Availability: 1
Header DbId: egs
DbLabel: Engineering Source
An: 194915146
AccessLevel: 6
PubType: Academic Journal
PubTypeId: academicJournal
PreciseRelevancyScore: 0
IllustrationInfo
Items – Name: Title
  Label: Title
  Group: Ti
  Data: Leaf-Level Hyperspectral Discrimination of Wild Carrot from Co-Occurring Weeds and Hybrid Carrots Using Optimized Preprocessing and Machine Learning.
– Name: Author
  Label: Authors
  Group: Au
  Data: &lt;searchLink fieldCode=&quot;AR&quot; term=&quot;%22Nanayakkara%2C+Dhanesha%22&quot;&gt;Nanayakkara, Dhanesha&lt;/searchLink&gt;&lt;relatesTo&gt;1,2&lt;/relatesTo&gt; (AUTHOR)&lt;br /&gt;&lt;searchLink fieldCode=&quot;AR&quot; term=&quot;%22Bhatia%2C+Nitin%22&quot;&gt;Bhatia, Nitin&lt;/searchLink&gt;&lt;relatesTo&gt;2,3&lt;/relatesTo&gt; (AUTHOR)&lt;br /&gt;&lt;searchLink fieldCode=&quot;AR&quot; term=&quot;%22Irwin%2C+Matthew%22&quot;&gt;Irwin, Matthew&lt;/searchLink&gt;&lt;relatesTo&gt;1,3&lt;/relatesTo&gt; (AUTHOR)&lt;br /&gt;&lt;searchLink fieldCode=&quot;AR&quot; term=&quot;%22McGill%2C+Craig%22&quot;&gt;McGill, Craig&lt;/searchLink&gt;&lt;relatesTo&gt;1&lt;/relatesTo&gt; (AUTHOR)
– Name: TitleSource
  Label: Source
  Group: Src
  Data: &lt;searchLink fieldCode=&quot;JN&quot; term=&quot;%22Remote+Sensing%22&quot;&gt;Remote Sensing&lt;/searchLink&gt;. Jun2026, Vol. 18 Issue 12, p2013. 28p.
– Name: Subject
  Label: Subjects
  Group: Su
  Data: &lt;searchLink fieldCode=&quot;DE&quot; term=&quot;%22Spectral+reflectance%22&quot;&gt;Spectral reflectance&lt;/searchLink&gt;&lt;br /&gt;&lt;searchLink fieldCode=&quot;DE&quot; term=&quot;%22Machine+learning%22&quot;&gt;Machine learning&lt;/searchLink&gt;&lt;br /&gt;&lt;searchLink fieldCode=&quot;DE&quot; term=&quot;%22Plant+classification%22&quot;&gt;Plant classification&lt;/searchLink&gt;&lt;br /&gt;&lt;searchLink fieldCode=&quot;DE&quot; term=&quot;%22Carrots%22&quot;&gt;Carrots&lt;/searchLink&gt;&lt;br /&gt;&lt;searchLink fieldCode=&quot;DE&quot; term=&quot;%22Multispectral+imaging%22&quot;&gt;Multispectral imaging&lt;/searchLink&gt;&lt;br /&gt;&lt;searchLink fieldCode=&quot;DE&quot; term=&quot;%22Remote+sensing%22&quot;&gt;Remote sensing&lt;/searchLink&gt;&lt;br /&gt;&lt;searchLink fieldCode=&quot;DE&quot; term=&quot;%22Weed+science%22&quot;&gt;Weed science&lt;/searchLink&gt;&lt;br /&gt;&lt;searchLink fieldCode=&quot;DE&quot; term=&quot;%22Spectrum+analysis%22&quot;&gt;Spectrum analysis&lt;/searchLink&gt;
– Name: SubjectGeographic
  Label: Geographic Terms
  Group: Su
  Data: &lt;searchLink fieldCode=&quot;DE&quot; term=&quot;%22New+Zealand%22&quot;&gt;New Zealand&lt;/searchLink&gt;
– Name: Abstract
  Label: Abstract
  Group: Ab
  Data: Highlights: What are the main findings? Leaf-level ASD hyperspectral reflectance (400–2450 nm) combined with min–max normalization and machine learning effectively discriminated wild carrot (Daucus carota subsp. carota) from 19 co-occurring weed species, four hybrid carrot varieties, and their parental lines, with over 90% accuracy in both binary and multiclass tasks using full-spectrum data. PLS-DA-derived VIP band selection identified compact sets of 10–20 diagnostic wavelengths, with full-spectrum models emphasizing the red-edge (704–730 nm) and SWIR (1390–1403 nm and 1875–1880 nm) regions, and airborne-compatible models selecting visible and red-edge bands (400–402, 527, 705–720 nm) that preserved high classification performance in detecting wild carrot while substantially reducing dimensionality. Spatial and temporal analyses indicated negligible site effects (R2 = 0.034–0.055, pseudo-F = 1.46–2.39, p &gt; 0.05) and moderate temporal variation (R2 = 0.136–0.151, p &lt; 0.001), confirming relatively stable wild carrot spectral signatures across locations and during the 5–10 week juvenile stage, supporting the use of pooled datasets and model generalization across space and time. What are the implications of the main findings? The identified visible and red-edge diagnostic bands for airborne-compatible models, together with supporting SWIR features from the full-spectrum analyses, provide a practical spectral basis for designing economical UAV-mounted multispectral sensors targeting wild carrot in hybrid carrot seed production landscapes, with the potential to enable site-specific weed management and reduce broadcast herbicide application. The demonstrated spectral stability of wild carrot across the three tested locations and early growth stages, together with the high classification accuracy for late-season plants of unknown age, suggests strong potential for transferring leaf-level models to field conditions and for deploying operational airborne or UAV/UGV-based wild carrot detection systems with limited recalibration. Wild carrot (Daucus carota subsp. carota), the wild relative of cultivated carrot, is globally identified as an invasive weed that threatens hybrid carrot seed production through natural cross-pollination, resulting in compromised genetic purity. Manual identification across the large areas required to ensure genetic purity in carrot seed crops is impractical. Remote sensing offers an alternative; however, morphological similarities among wild carrot, cultivated carrot, and common weeds hinder reliable detection. Early identification, however, remains essential for preventing genetic contamination. This study evaluated leaf-level hyperspectral reflectance spectroscopy (400–2450 nm) with machine learning to discriminate wild carrot from hybrid carrots, parental lines, and 19 co-occurring weed species. Spectral data from 266 wild carrot plants across three New Zealand sites and six weeks (5–10 weeks after emergence) showed negligible spatial effects (R2 = 0.034–0.055, pseudo-F = 1.46–2.39, p &gt; 0.05) and moderate temporal variation (R2 = 0.136–0.151, pseudo-F = 5.48–6.17, p &lt; 0.001), indicating broadly stable spectral signatures suitable for model generalization. Savitzky–Golay filtering, with min–max normalization outperformed SNV, yielding high full-spectrum accuracies for wild carrot vs. other species (90.35%, κ = 0.80), wild carrot vs. weeds (96.03%, κ = 0.92), and a multi-class model (90.79%, κ = 0.88). After removing atmospheric water-absorption bands to follow airborne sensing, reduced-band models based on airborne-compatible wavelengths maintained strong performance, including 89.40% accuracy (κ = 0.79) for wild carrot vs. weeds using a 20-band Subspace Discriminant model (400–402, 527, 705–720 nm). These findings demonstrate that stable wild carrot spectra and carefully selected visible and red-edge bands can underpin cost-effective UAV/UGV-mounted hyperspectral or multispectral sensors for site-specific wild carrot management. [ABSTRACT FROM AUTHOR]
– Name: AbstractSuppliedCopyright
  Label:
  Group: Ab
  Data: &lt;i&gt;Copyright of Remote Sensing is the property of MDPI and its content may not be copied or emailed to multiple sites without the copyright holder&#39;s express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract.&lt;/i&gt; (Copyright applies to all Abstracts.)
PLink https://search.ebscohost.com/login.aspx?direct=true&site=eds-live&db=egs&AN=194915146
RecordInfo BibRecord:
  BibEntity:
    Identifiers:
      – Type: doi
        Value: 10.3390/rs18122013
    Languages:
      – Code: eng
        Text: English
    PhysicalDescription:
      Pagination:
        PageCount: 28
        StartPage: 2013
    Subjects:
      – SubjectFull: Spectral reflectance
        Type: general
      – SubjectFull: Machine learning
        Type: general
      – SubjectFull: Plant classification
        Type: general
      – SubjectFull: Carrots
        Type: general
      – SubjectFull: Multispectral imaging
        Type: general
      – SubjectFull: Remote sensing
        Type: general
      – SubjectFull: Weed science
        Type: general
      – SubjectFull: Spectrum analysis
        Type: general
      – SubjectFull: New Zealand
        Type: general
    Titles:
      – TitleFull: Leaf-Level Hyperspectral Discrimination of Wild Carrot from Co-Occurring Weeds and Hybrid Carrots Using Optimized Preprocessing and Machine Learning.
        Type: main
  BibRelationships:
    HasContributorRelationships:
      – PersonEntity:
          Name:
            NameFull: Nanayakkara, Dhanesha
      – PersonEntity:
          Name:
            NameFull: Bhatia, Nitin
      – PersonEntity:
          Name:
            NameFull: Irwin, Matthew
      – PersonEntity:
          Name:
            NameFull: McGill, Craig
    IsPartOfRelationships:
      – BibEntity:
          Dates:
            – D: 15
              M: 06
              Text: Jun2026
              Type: published
              Y: 2026
          Identifiers:
            – Type: issn-print
              Value: 20724292
          Numbering:
            – Type: volume
              Value: 18
            – Type: issue
              Value: 12
          Titles:
            – TitleFull: Remote Sensing
              Type: main
ResultId 1