![Hier klicken, um den Treffer aus der Auswahl zu entfernen](images/unchecked.gif) |
Titel |
Leaf Surface Effects on Retrieving Chlorophyll Content from Hyperspectral Remote Sensing |
VerfasserIn |
Feng Qiu, Jing Ming Chen, Weimin Ju, Jun Wang, Qian Zhang |
Konferenz |
EGU General Assembly 2017
|
Medientyp |
Artikel
|
Sprache |
en
|
Digitales Dokument |
PDF |
Erschienen |
In: GRA - Volume 19 (2017) |
Datensatznummer |
250137928
|
Publikation (Nr.) |
EGU/EGU2017-802.pdf |
|
|
|
Zusammenfassung |
Light reflected directly from the leaf surface without entering the surface layer is not
influenced by leaf internal biochemical content. Leaf surface reflectance varies from leaf to
leaf due to differences in the surface roughness features and is relatively more important in
strong absorption spectral regions. Therefore it introduces dispersion of data points in
the relationship between biochemical concentration and reflectance (especially in
the visible region). Separation of surface from total leaf reflection is important
to improve the link between leaf pigments content and remote sensing data. This
study aims to estimate leaf surface reflectance from hyperspectral remote sensing
data and retrieve chlorophyll content by inverting a modified PROSPECT model.
Considering leaf surface reflectance is almost the same in the visible and near infrared
spectral regions, a surface layer with a reflectance independent of wavelength but
varying from leaf to leaf was added to the PROSPECT model. The specific absorption
coefficients of pigments were recalibrated. Then the modified model was inverted on
independent datasets to check the performance of the model in predicting the chlorophyll
content.
Results show that differences in estimated surface layer reflectance of various species are
noticeable. Surface reflectance of leaves with epicuticular waxes and trichomes is usually
higher than other samples. Reconstruction of leaf reflectance and transmittance in the
400-1000 nm wavelength region using the modified PROSPECT model is excellent with low
root mean square error (RMSE) and bias. Improvements for samples with high surface
reflectance (e.g. maize) are significant, especially for high pigment leaves. Moreover,
chlorophyll retrieved from inversion of the modified model is consequently improved (RMSE
from 5.9-13.3 ug/cm2 with mean value 8.1 ug/cm2, while mean correlation coefficient is
0.90) compared to results of PROSPECT-5 (RMSE from 9.6-20.2 ug/cm2 with mean value
13.1 ug/cm2, while mean correlation coefficient is 0.81). Underestimation of high
chlorophyll content, which is due to underestimation of reflectance in the visible region
of PROSPECT, is partially corrected or alleviated. Improvements are particularly
noticeable for leaves with high surface reflectance or high chlorophyll content, which
both lead to large proportions of surface reflectance to the total leaf reflectance. |
|
|
|
|
|