Potentials and Limitations of WorldView-3 Data for the Detection of Invasive Lupinus polyphyllus Lindl. in Semi-Natural Grasslands

Damian Schulze-Brüninghoff,Thomas Astor,Michael Wachendorf

doi:10.3390/rs13214333

Abstract

Semi-natural grasslands contribute highly to biodiversity and other ecosystem services, but they are at risk by the spread of invasive plant species, which alter their habitat structure. Large area grassland monitoring can be a powerful tool to manage invaded ecosystems. Therefore, WorldView-3 multispectral sensor data was utilized to train multiple machine learning algorithms in an automatic machine learning workflow called ‘H2O AutoML’ to detect L. polyphyllus in a nature protection grassland ecosystem. Different degree of L. polyphyllus cover was collected on 3 × 3 m2 reference plots, and multispectral bands, indices, and texture features were used in a feature selection process to identify the most promising classification model and machine learning algorithm based on mean per class error, log loss, and AUC metrics. The best performance was achieved with a binary classification of lupin-free vs. fully invaded 3 × 3 m2 plot classification with a set of 7 features out of 763. The findings reveal that L. polyphyllus detection from WorldView-3 sensor data is limited to large dominant spots and not recommendable for lower plant coverage, especially single plant detection. Further research is needed to clarify if different phenological stages of L. polyphyllus as well as time series increase classification performance.

Highlights

Extensive grasslands, especially at nature conservation sites, are important habitats for multiple endangered species [1]
We identified the threshold value that gives a maximum performance of F0.5-score and F2-score to compare their outcome of L. polyphyllus prediction maps
Our best binary classification showed slightly lower performance compared to other studies using WorldView-3 images to classify invasive plant species. [16] achieved accuracies between 76.6 and 91.2% with an XGBoost algorithm, depending on feature input, which is close to the accuracy of the GBM algorithm in the present study (77%)

Summary

Introduction

Especially at nature conservation sites, are important habitats for multiple endangered species [1]. Thereby, they have a key role in supporting biodiversity, [2]. Extensive grasslands are valuable culturally grown landscapes with increasing significance for species to adapt to the effects of climate change and human activities [4]. It should be one of our main goals to preserve such refugium, because changing climate will increase the challenges for species, which are adapted to specific habitat structures and climate conditions, while at the same time, habitats matching species requirements will become rare. If an invasive species has superior competition advantages, it can rapidly become a dominant species in a habitat, which can be vulnerable to such a degree, that species composition changes drastically and the profile and performance will change in recipient ecosystems, shifting the balance between services and disservices [6]

Objectives

Methods

Results

Discussion

Conclusion