Comparative study of three satellite image time-series decomposition methods for vegetation change detection

Ali Ben Abbes,Oumayma Bounouh,Imed Riadh Farah,Rogier De Jong,Beatriz Martínez

doi:10.1080/22797254.2018.1465360

Abstract

ABSTRACTSatellite image time-series (SITS) methods have contributed notably to detection of global change over the last decades, for instance by tracking vegetation changes. Compared with multi-temporal change detection methods, temporally highly resolved SITS methods provide more information in a single analysis, for instance on the type and consistency of change. In particular, SITS decomposition methods show a great potential in extracting various components from non-stationary time series, which allows for an improved interpretation of the temporal variability. Even though many case studies have applied SITS decomposition methods, a systematic comparison of common algorithms is still missing.In this study, the seasonal trend loess (STL), breaks for additive season and trend (BFAST) and multi-resolution analysis-wavelet transform (MRA-WT) were explored in order to evaluate their performance in modelling, monitoring and detecting land-cover changes with pronounced seasonal variations from simulated normal difference vegetation index time series. The selected methods have all proven their ability to characterize the non-stationary vegetation dynamics along with different physical processes driving the vegetation dynamics. Our results indicated that BFAST is the most accurate method for the examined simulated dataset in terms of RMSE, whereas MRA-WT showed a great potential for the extraction of multi-level vegetation dynamics. Considering the computational efficiency, both STL and MRA-WT outperformed BFAST.

Highlights

Vegetation change detection from remotely sensed images with high temporal resolution is important to promote better decisions for sustainable land management
Our results indicated that breaks for additive season and trend (BFAST) is the most accurate method for the examined simulated dataset in terms of root mean square error (RMSE), whereas multi-resolution analysis-wavelet transform (MRA-WT) showed a great potential for the extraction of multi-level vegetation dynamics
The details of the seasonal component were better estimated by BFAST and MRAWT approaches. ● Boundary effect: As shown in Figure 2 case (a), both seasonal trend loess (STL) and multi-resolution analysis (MRA)-WT trends diverge

Summary

Introduction

Vegetation change detection from remotely sensed images with high temporal resolution is important to promote better decisions for sustainable land management. The field of change detection started with approaches for bi-temporal datasets, including image differencing (Bruzzone & Prieto, 2000; Mas, 1999), change vector analysis (Bovolo, Marchesi, & Bruzzone, 2012), principal component analysis (Byrne, Crapper, & Mayo, 1980) or tasselled cap transformation-based methods (Minu & Shetty, 2015). These methods are still very popular and can be applied to more than two images, they may be impractical for long and dense SITS and may suffer from some limitations. High temporal-resolution SITS methods account for several – not all – of these drawbacks and, are an interesting tool for tracing complex trajectories of land-cover change

Objectives

Results

Discussion

Conclusion