Multi-Temporal Sentinel-2 Data in Classification of Mountain Vegetation

Martyna Wakulińska,Adriana Marcinkowska-Ochtyra

doi:10.3390/rs12172696

Abstract

The electromagnetic spectrum registered via satellite remote sensing methods became a popular data source that can enrich traditional methods of vegetation monitoring. The European Space Agency Sentinel-2 mission, thanks to its spatial (10–20 m) and spectral resolution (12 spectral bands registered in visible-, near-, and mid-infrared spectrum) and primarily its short revisit time (5 days), helps to provide reliable and accurate material for the identification of mountain vegetation. Using the support vector machines (SVM) algorithm and reference data (botanical map of non-forest vegetation, field survey data, and high spatial resolution images) it was possible to classify eight vegetation types of Giant Mountains: bogs and fens, deciduous shrub vegetation, forests, grasslands, heathlands, subalpine tall forbs, subalpine dwarf pine scrubs, and rock and scree vegetation. Additional variables such as principal component analysis (PCA) bands and selected vegetation indices were included in the best classified dataset. The results of the iterative classification, repeated 100 times, were assessed as approximately 80% median overall accuracy (OA) based on multi-temporal datasets composed of images acquired through the vegetation growing season (from late spring to early autumn 2018), better than using a single-date scene (70%–72% OA). Additional variables did not significantly improve the results, showing the importance of spectral and temporal information themselves. Our study confirms the possibility of fully available data for the identification of mountain vegetation for management purposes and protection within national parks.

Highlights

Mountain vegetation is vulnerable to climate change, where the changes of the tree line and plant floor borders become visible [1]
We divided the discussion section into three parts: the first is devoted to mountain vegetation classification with special attention to the Giant Mountains study area (Section 4.1); in the second, we describe the use of multi-temporal data in classification (Section 4.2); in the third, we discuss the sense of including additional variables in the classification (Section 4.3)
Analogous studies with multispectral Sentinel-2 data were provided by Kupková et al [7] where support vector machines (SVM), MLC, and neural nets (NNs) algorithms were used to classify eight vegetation classes in Eastern tundra of the Giant Mountains, which yielded lower overall accuracy (OA) than our results—71.0% and 79.5% based on SVM, respectively

Summary

Introduction

Mountain vegetation is vulnerable to climate change, where the changes of the tree line and plant floor borders become visible [1]. The occurrence of species from various geographical regions in a relatively small area, often being glacial relics, endemics, or endangered species, makes their identification and monitoring extremely important for preserving natural wealth [2]. It is important to provide up-to-date vegetation maps of mountain protected areas. Field mapping requires a lot of time and work. In the case of high-mountain vegetation, the limited availability and shorter vegetation period compared to lowlands significantly affect the possibilities of field research. Due to rapid technological progress, remote sensing data, characterized by both greater objectivity and spatial coverage, are increasingly used [3]. The electromagnetic spectrum registered by remote sensing instruments, which create unique spectral characteristics of the analyzed objects, can support traditional methods of vegetation mapping by the use of image classification [3]

Methods

Discussion

Conclusion