To produce high quality paper, lignin should be removed from the pulp.
Quantification of lignin concentrations using standard wet chemistry is
accurate but time consuming and costly, thus not appropriate for a large
number of samples. The ability of hyperspectral remote sensing to predict
foliar lignin concentrations could be utilized to estimate wood lignin
concentrations if meaningful relationships between wood and foliar chemistry
are established. LAI (leaf area index) is a useful parameter that is
incorporated into physiological models in forest assessment. Measuring LAI
over vast areas is labour intensive and expensive; therefore, LAI has been
correlated to vegetation indices using remote sensing. Broadband indices use
average spectral information over broad bandwidths; therefore details on the
characteristics of the forest canopy are compromised and averaged.
Moreover, the broadband indices are known to be highly affected by soil
background at low vegetation cover. The aim of this study is to determine
foliar and wood lignin concentrations of Eucalyptus clones using hyperspectral
lignin indices, and to estimate LAI of Eucalyptus clones from narrowband
vegetation indices in Zululand, South Africa
Twelve Eucalyptus compartments of ages between 6 and 9 years were
selected and 5 trees were randomly sampled from each compartment. A
Hyperion image was acquired within ten days of field sampling, SI and LAI
measurements. Leaf samples were analyzed in the laboratory using the
Klason method as per Tappi standards (Tappi, 1996-1997). Wood samples
were analyzed for lignin concentrations using a NIRS (Near Infrared
Spectroscopy) instrument. The results showed that there is no general model
for predicting wood lignin concentrations from foliar lignin concentrations in
Eucalyptus clones of ages between 6 and 9 years. Regression analysis
performed for individual compartments and on compartments grouped
according to age and SI showed that the relationship between wood and foliar
lignin concentration is site and age specific. A Hyperion image was georeferenced
and atmospherically corrected using ENVI FLAASH 4.2.
The equation to calculate lignin indices for this study was: L1R= ~n5il: A'''''y .
1750 AI680
The relationship between the lignin index and laboratory-measured foliar lignin
was significant with R2 = 0.79. This relationship was used to calculate imagepredicted
foliar lignin concentrations. Firstly, the compartment specific
equations were used to calculate predicted wood lignin concentrations from
predicted foliar lignin concentrations. The relationship between the laboratorymeasured
wood lignin concentrations and predicted wood lignin concentrations
was significant with R2 = 0.91. Secondly, the age and site-specific equations
were used to convert foliar lignin concentration to wood lignin concentrations.
The wood lignin concentrations predicted from these equations were then
compared to the laboratory-measured wood lignin concentrations using linear
regression and the R2 was 0.79 with a p-value lower than 0.001.
Two bands were used to calculate nine vegetation indices; one band from the
near infrared (NIR) region and the other from the short wave infrared (SWIR).
Correlations between the Vis and the LAI measurements were generated and
. then evaluated to determine the most effective VI for estimating LAI of
Eucalyptus plantations. All the results obtained were significant but the NU
and MNU showed possible problems of saturation. The MNDVI*SR and
SAVI*SR produced the most significant relationships with LAI with R2 values
of 0.899 and 0.897 respectively. The standard error for both correlations was
very low, at 0.080, and the p-value of 0.001.
It was concluded that the Eucalyptus wood lignin concentrations can be
predicted using hyperspectral remote sensing, hence wood and foliar lignin
concentrations can be fairly accurately mapped across compartments. LAI
significantly correlated to eight of the nine selected vegetation indices. Seven
Vis are more suitable for LAI estimations in the Eucalyptus plantations in
Zululand. The NU and MNU can only be used for LAI estimations in arid or
semi-arid areas. / Thesis (M.Sc.)-University of KwaZulu-Natal, Pietermaritzburg, 2006.
| Identifer | oai:union.ndltd.org:netd.ac.za/oai:union.ndltd.org:ukzn/oai:http://researchspace.ukzn.ac.za:10413/3470 |
| Date | January 2006 |
| Creators | Mthembu, Ingrid Bongiwe. |
| Contributors | Ahmed, Fethi B. |
| Source Sets | South African National ETD Portal |
| Detected Language | English |
Page generated in 0.002 seconds