Field spectroscopy of plant water content in Eucalyptus grandis forest stands in KwaZulu-Natal, South Africa

Abstract

The measurement of plant water content is essential to assess stress and

disturbance in forest plantations. Traditional techniques to assess plant water

content are costly, time consuming and spatially restrictive. Remote sensing

techniques offer the alternative of a non destructive and instantaneous method of

assessing plant water content over large spatial scales where ground

measurements would be impossible on a regular basis. The aim of this research

was to assess the relationship between plant water content and reflectance data in

Eucalyptus grandis forest stands in KwaZulu-Natal, South Africa. Field reflectance

and first derivative reflectance data were correlated with plant water content. The

first derivative reflectance performed better than the field reflectance data in

estimating plant water content with high correlations in the visible and mid-infrared

portions of the electromagnetic spectrum. Several reflectance indices were also

tested to evaluate their effectiveness in estimating plant water content and were

compared to the red edge position. The red edge position calculated from the first

derivative reflectance and from the linear four-point interpolation method performed

better than all the water indices tested. It was therefore concluded that the red

edge position can be used in association with other water indices as a stable

spectral parameter to estimate plant water content on hyperspectral data. The

South African satellite SumbandilaSat is due for launch in the near future and it is

essential to test the utility of this satellite in estimating plant water content, a study

which has not been done before. The field reflectance data from this study was

resampled to the SumbandilaSat band settings and was put into a neural network

to test its potential in estimating plant water content. The integrated approach

involving neural networks and the resampled field spectral data successfully

predicted plant water content with a correlation coefficient of 0.74 and a root mean

square error (RMSE) of 1.41 on an independent test dataset outperforming the

traditional multiple regression method of estimation. The potential of the

SumbandilaSat wavebands to estimate plant water content was tested using a

sensitivity analysis. The results from the sensitivity analysis indicated that the xanthophyll, blue and near infrared wavebands are the three most important

wavebands used by the neural network in estimating plant water content. It was

therefore concluded that these three bands of the SumbandilaSat are essential for

plant water estimation. In general this study showed the potential of up-scaling field

spectral data to the SumbandilaSat, the second South African satellite scheduled

for launch in the near future.