Spectral Unmixing-based Arctic Plant Species Analysis using a Spectral Library and Terrestrial Hyperspectral Imagery: A Case Study in Adventdalen, Svalbard

Abstract

Remote sensing is an invaluable tool for understanding and monitoring the rapid changes in the distribution and composition of Arctic vegetation caused by global warming. Although hyperspectral data consisting of contiguous spectral bands enables the quantitative analysis of remote sensing data, monitoring studies on the Arctic vegetation based on hyperspectral remote sensing have rarely been discussed because of the difficulty in acquiring data from the Arctic region. To address this limitation, we collected hyperspectral information on the dominant vegetation species in Adventdalen Valley, Svalbard, and investigated various approaches for mapping Arctic vegetation using these data. First, labeled datasets were constructed for Arctic vegetation species by extracting pixel data from ground-based hyperspectral images. We then quantitatively and qualitatively compared the classification performances of three machine learning-based classifiers, random forest (RF), support vector machine (SVM), and one-dimensional convolutional neural network (1D-CNN), using the datasets and hyperspectral images. The 1D-CNN classifier combined with the first-order derivative of smoothed reflectance (RS,FD) achieved the highest statistical accuracies of 0.9892 (Kappa = 0.9880) and 0.9352 (Kappa = 0.9280) for the two independent test sets and produced the most accurate vegetation maps. Additionally, a spectral library was developed using the mean specta per class of the labeled datasets. The spectral library using the RS,FD spectrum showed high spectral discriminability and potential for estimating the abundance of classes from mixed pixels, which is commonly formed in low-spatial resolution images. Accordingly, the spectral library developed in this study can serve as a fundamental reference for large-scale mapping studies using remote sensing. The proposed approaches in this study can be widely applied to various remote sensing datasets, leading to effective monitoring of Arctic vegetation changes in response to climate change.