Using hyperspectral data and PLSR modelling to assess acid sulphate soil in subsurface

Abstract

Purpose - Acid sulphate soil (ASS) has raised increasing environmental concerns because of its capability to produce strong acidity and consequent trace metal release. It is difficult to assess the occurrence and severity of ASS in the subsurface by conventional methods, either by chemical measurements following intensive field survey or by airborne/spaceborne remote sensing. This paper aims to explore a new way to rapidly assess the occurrence and severity of the harmful ASS in the subsurface. Materials and methods - This paper introduced a proximal hyperspectral instrument, namely Hylogger™ system, which can rapidly scan soil cores and provide high-resolution hyperspectral data to assess ASS occurring in the subsurface. Traditional soil coring and chemical measurements were also applied to assist the assessment. Furthermore, partial least squares regression (PLSR) was used to establish the relationship between soil pH values and reflectance spectral features. Results and discussion - The main results include mineral distribution, which was mapped using scanned hypespectral data on soil cores, soil pH map and the distribution of the two types of ASS, including harmful actual acid sulphate soil and harmless potential acid sulphate soil. Furthermore, the relationship between the soil pH values and spectral features was established by PLSR modelling. Conclusions - Conclusively, ASS in the subsurface was characterised spectrally, the mineralogy was mapped using hyperspectral data from soil cores, and the AASS and the PASS were separated as well. The model established could be used to predict soil pH in the future; thus, it could further accelerate the assessment of ASS.