Predicting water stress induced by Thaumastocoris peregrinus infestations in plantation forests using field spectroscopy and neural networks

Abstract

Field spectroscopy was tested in predicting water stress induced by Thaumastocoris peregrinus infestations, a pest which is causing significant damage to eucalypt plantations internationally. Water indices and known water absorption bands calculated from hyperspectral field reflectance were input into a neural network algorithm to predict plant water content (PWC) and equivalent water thickness (EWT). The integrated approach involving field spectral data and neural networks predicted PWC and EWT with correlation coefficients of 0.88 and 0.71 on independent test datasets. The results indicate the potential of high-resolution field spectral data in detecting the early stages of insect infestation due to physiological changes that alter water content.

Keywords::

• Thaumastocoris peregrinus
• plant water content
• equivalent water thickness
• field spectroscopy
• neural networks

Acknowledgements

We thank Muhammad Sheik Oumar and Romano Lottering for helping with the field work. Funding for this research was provided by the National Research Foundation of South Africa and the University of KwaZulu-Natal.