Leaf Water Potential in a Mixed Mediterranean Forest from Machine Learning and Unmanned Aerial Vehicle (UAV)-Based Hyperspectral Imaging
Leaf water potential (<inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mi>ψ</mi></mrow></semantics></math></inline-formula><sub>leaf</sub>) is a key indicato...
Saved in:
| Main Authors: | , , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2024-12-01
|
| Series: | Remote Sensing |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2072-4292/17/1/106 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1841549026422423552 |
|---|---|
| author | Netanel Fishman Yehuda Yungstein Assaf Yaakobi Sophie Obersteiner Laura Rez Gabriel Mulero Yaron Michael Tamir Klein David Helman |
| author_facet | Netanel Fishman Yehuda Yungstein Assaf Yaakobi Sophie Obersteiner Laura Rez Gabriel Mulero Yaron Michael Tamir Klein David Helman |
| author_sort | Netanel Fishman |
| collection | DOAJ |
| description | Leaf water potential (<inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mi>ψ</mi></mrow></semantics></math></inline-formula><sub>leaf</sub>) is a key indicator of plant water status, but its measurement is labor-intensive and limited in spatial coverage. While remote sensing has emerged as a useful tool for estimating vegetation water status, <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mi>ψ</mi></mrow></semantics></math></inline-formula><sub>leaf</sub> remains unexplored, particularly in mixed forests. Here, we use spectral indices derived from unmanned aerial vehicle-based hyperspectral imaging and machine learning algorithms to assess <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mi>ψ</mi></mrow></semantics></math></inline-formula><sub>leaf</sub> in a mixed, multi-species Mediterranean forest comprised of five key woody species: <i>Pinus halepensis</i>, <i>Quercus calliprinos</i>, <i>Cupressus sempervirens</i>, <i>Ceratonia siliqua</i>, and <i>Pistacia lentiscus</i>. Hyperspectral images (400–1000 nm) were acquired monthly over one year, concurrent with <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mi>ψ</mi></mrow></semantics></math></inline-formula><sub>leaf</sub> measurements in each species. Twelve spectral indices and thousands of normalized difference spectral index (NDSI) combinations were evaluated. Three machine learning algorithms—random forest (RF), extreme gradient boosting (XGBoost), and support vector machine (SVM)—were used to model <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mi>ψ</mi></mrow></semantics></math></inline-formula><sub>leaf</sub>. We compared the machine learning model results with linear models based on spectral indices and the NDSI. SVM, using species information as a feature, performed the best with a relatively good <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mi>ψ</mi></mrow></semantics></math></inline-formula><sub>leaf</sub> assessment (R<sup>2</sup> = 0.53; RMSE = 0.67 MPa; rRMSE = 28%), especially considering the small seasonal variance in <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mi>ψ</mi></mrow></semantics></math></inline-formula><sub>leaf</sub> (<inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mo>±</mo><mi>σ</mi></mrow></semantics></math></inline-formula> = 0.8 MPa). Predictions were best for <i>Cupressus sempervirens</i> (R<sup>2</sup> = 0.80) and <i>Pistacia lentiscus</i> (R<sup>2</sup> = 0.49), which had the largest <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mi>ψ</mi></mrow></semantics></math></inline-formula><sub>leaf</sub> variances (<inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mo>±</mo><mi>σ</mi></mrow></semantics></math></inline-formula> > 1 MPa). Aggregating data at the plot scale in a ‘general’ model markedly improved the <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mi>ψ</mi></mrow></semantics></math></inline-formula><sub>leaf</sub> model (R<sup>2</sup> = 0.79, RMSE = 0.31 MPa; rRMSE = 13%), providing a promising tool for monitoring mixed forest <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mi>ψ</mi></mrow></semantics></math></inline-formula><sub>leaf</sub>. The fact that a non-species-specific, ‘general’ model could predict <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mi>ψ</mi></mrow></semantics></math></inline-formula><sub>leaf</sub> implies that such a model can also be used with coarser resolution satellite data. Our study demonstrates the potential of combining hyperspectral imagery with machine learning for non-invasive <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mi>ψ</mi></mrow></semantics></math></inline-formula><sub>leaf</sub> estimation in mixed forests while highlighting challenges in capturing interspecies variability. |
| format | Article |
| id | doaj-art-d0fe93827c1f4f01ac0121ba6745a2af |
| institution | Kabale University |
| issn | 2072-4292 |
| language | English |
| publishDate | 2024-12-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Remote Sensing |
| spelling | doaj-art-d0fe93827c1f4f01ac0121ba6745a2af2025-01-10T13:20:15ZengMDPI AGRemote Sensing2072-42922024-12-0117110610.3390/rs17010106Leaf Water Potential in a Mixed Mediterranean Forest from Machine Learning and Unmanned Aerial Vehicle (UAV)-Based Hyperspectral ImagingNetanel Fishman0Yehuda Yungstein1Assaf Yaakobi2Sophie Obersteiner3Laura Rez4Gabriel Mulero5Yaron Michael6Tamir Klein7David Helman8Department of Soil and Water Sciences, Institute of Environmental Sciences, The Robert H. Smith Faculty of Agriculture, Food and Environment, The Hebrew University of Jerusalem, Rehovot 7610001, IsraelDepartment of Soil and Water Sciences, Institute of Environmental Sciences, The Robert H. Smith Faculty of Agriculture, Food and Environment, The Hebrew University of Jerusalem, Rehovot 7610001, IsraelDepartment of Plant and Environmental Sciences, Weizmann Institute of Science, Rehovot 7610001, IsraelDepartment of Plant and Environmental Sciences, Weizmann Institute of Science, Rehovot 7610001, IsraelDepartment of Plant and Environmental Sciences, Weizmann Institute of Science, Rehovot 7610001, IsraelDepartment of Soil and Water Sciences, Institute of Environmental Sciences, The Robert H. Smith Faculty of Agriculture, Food and Environment, The Hebrew University of Jerusalem, Rehovot 7610001, IsraelDepartment of Soil and Water Sciences, Institute of Environmental Sciences, The Robert H. Smith Faculty of Agriculture, Food and Environment, The Hebrew University of Jerusalem, Rehovot 7610001, IsraelDepartment of Plant and Environmental Sciences, Weizmann Institute of Science, Rehovot 7610001, IsraelDepartment of Soil and Water Sciences, Institute of Environmental Sciences, The Robert H. Smith Faculty of Agriculture, Food and Environment, The Hebrew University of Jerusalem, Rehovot 7610001, IsraelLeaf water potential (<inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mi>ψ</mi></mrow></semantics></math></inline-formula><sub>leaf</sub>) is a key indicator of plant water status, but its measurement is labor-intensive and limited in spatial coverage. While remote sensing has emerged as a useful tool for estimating vegetation water status, <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mi>ψ</mi></mrow></semantics></math></inline-formula><sub>leaf</sub> remains unexplored, particularly in mixed forests. Here, we use spectral indices derived from unmanned aerial vehicle-based hyperspectral imaging and machine learning algorithms to assess <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mi>ψ</mi></mrow></semantics></math></inline-formula><sub>leaf</sub> in a mixed, multi-species Mediterranean forest comprised of five key woody species: <i>Pinus halepensis</i>, <i>Quercus calliprinos</i>, <i>Cupressus sempervirens</i>, <i>Ceratonia siliqua</i>, and <i>Pistacia lentiscus</i>. Hyperspectral images (400–1000 nm) were acquired monthly over one year, concurrent with <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mi>ψ</mi></mrow></semantics></math></inline-formula><sub>leaf</sub> measurements in each species. Twelve spectral indices and thousands of normalized difference spectral index (NDSI) combinations were evaluated. Three machine learning algorithms—random forest (RF), extreme gradient boosting (XGBoost), and support vector machine (SVM)—were used to model <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mi>ψ</mi></mrow></semantics></math></inline-formula><sub>leaf</sub>. We compared the machine learning model results with linear models based on spectral indices and the NDSI. SVM, using species information as a feature, performed the best with a relatively good <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mi>ψ</mi></mrow></semantics></math></inline-formula><sub>leaf</sub> assessment (R<sup>2</sup> = 0.53; RMSE = 0.67 MPa; rRMSE = 28%), especially considering the small seasonal variance in <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mi>ψ</mi></mrow></semantics></math></inline-formula><sub>leaf</sub> (<inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mo>±</mo><mi>σ</mi></mrow></semantics></math></inline-formula> = 0.8 MPa). Predictions were best for <i>Cupressus sempervirens</i> (R<sup>2</sup> = 0.80) and <i>Pistacia lentiscus</i> (R<sup>2</sup> = 0.49), which had the largest <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mi>ψ</mi></mrow></semantics></math></inline-formula><sub>leaf</sub> variances (<inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mo>±</mo><mi>σ</mi></mrow></semantics></math></inline-formula> > 1 MPa). Aggregating data at the plot scale in a ‘general’ model markedly improved the <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mi>ψ</mi></mrow></semantics></math></inline-formula><sub>leaf</sub> model (R<sup>2</sup> = 0.79, RMSE = 0.31 MPa; rRMSE = 13%), providing a promising tool for monitoring mixed forest <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mi>ψ</mi></mrow></semantics></math></inline-formula><sub>leaf</sub>. The fact that a non-species-specific, ‘general’ model could predict <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mi>ψ</mi></mrow></semantics></math></inline-formula><sub>leaf</sub> implies that such a model can also be used with coarser resolution satellite data. Our study demonstrates the potential of combining hyperspectral imagery with machine learning for non-invasive <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mi>ψ</mi></mrow></semantics></math></inline-formula><sub>leaf</sub> estimation in mixed forests while highlighting challenges in capturing interspecies variability.https://www.mdpi.com/2072-4292/17/1/106NDVIrandom forestremote sensingSVMwaterXGBoost |
| spellingShingle | Netanel Fishman Yehuda Yungstein Assaf Yaakobi Sophie Obersteiner Laura Rez Gabriel Mulero Yaron Michael Tamir Klein David Helman Leaf Water Potential in a Mixed Mediterranean Forest from Machine Learning and Unmanned Aerial Vehicle (UAV)-Based Hyperspectral Imaging Remote Sensing NDVI random forest remote sensing SVM water XGBoost |
| title | Leaf Water Potential in a Mixed Mediterranean Forest from Machine Learning and Unmanned Aerial Vehicle (UAV)-Based Hyperspectral Imaging |
| title_full | Leaf Water Potential in a Mixed Mediterranean Forest from Machine Learning and Unmanned Aerial Vehicle (UAV)-Based Hyperspectral Imaging |
| title_fullStr | Leaf Water Potential in a Mixed Mediterranean Forest from Machine Learning and Unmanned Aerial Vehicle (UAV)-Based Hyperspectral Imaging |
| title_full_unstemmed | Leaf Water Potential in a Mixed Mediterranean Forest from Machine Learning and Unmanned Aerial Vehicle (UAV)-Based Hyperspectral Imaging |
| title_short | Leaf Water Potential in a Mixed Mediterranean Forest from Machine Learning and Unmanned Aerial Vehicle (UAV)-Based Hyperspectral Imaging |
| title_sort | leaf water potential in a mixed mediterranean forest from machine learning and unmanned aerial vehicle uav based hyperspectral imaging |
| topic | NDVI random forest remote sensing SVM water XGBoost |
| url | https://www.mdpi.com/2072-4292/17/1/106 |
| work_keys_str_mv | AT netanelfishman leafwaterpotentialinamixedmediterraneanforestfrommachinelearningandunmannedaerialvehicleuavbasedhyperspectralimaging AT yehudayungstein leafwaterpotentialinamixedmediterraneanforestfrommachinelearningandunmannedaerialvehicleuavbasedhyperspectralimaging AT assafyaakobi leafwaterpotentialinamixedmediterraneanforestfrommachinelearningandunmannedaerialvehicleuavbasedhyperspectralimaging AT sophieobersteiner leafwaterpotentialinamixedmediterraneanforestfrommachinelearningandunmannedaerialvehicleuavbasedhyperspectralimaging AT laurarez leafwaterpotentialinamixedmediterraneanforestfrommachinelearningandunmannedaerialvehicleuavbasedhyperspectralimaging AT gabrielmulero leafwaterpotentialinamixedmediterraneanforestfrommachinelearningandunmannedaerialvehicleuavbasedhyperspectralimaging AT yaronmichael leafwaterpotentialinamixedmediterraneanforestfrommachinelearningandunmannedaerialvehicleuavbasedhyperspectralimaging AT tamirklein leafwaterpotentialinamixedmediterraneanforestfrommachinelearningandunmannedaerialvehicleuavbasedhyperspectralimaging AT davidhelman leafwaterpotentialinamixedmediterraneanforestfrommachinelearningandunmannedaerialvehicleuavbasedhyperspectralimaging |