Satellite Image Time Series Analysis for Big Earth Observation Data

Abstract

The development of analytical software for big Earth observation data faces several challenges. Designers need to balance between conflicting factors. Solutions that are efficient for specific hardware architectures can not be used in other environments. Packages that work on generic hardware and open standards will not have the same performance as dedicated solutions. Software that assumes that its users are computer programmers are flexible but may be difficult to learn for a wide audience. This paper describes sits, an open-source R package for satellite image time series analysis using machine learning. To allow experts to use satellite imagery to the fullest extent, sits adopts a time-first, space-later approach. It supports the complete cycle of data analysis for land classification. Its API provides a simple but powerful set of functions. The software works in different cloud computing environments. Satellite image time series are input to machine learning classifiers, and the results are post-processed using spatial smoothing. Since machine learning methods need accurate training data, sits includes methods for quality assessment of training samples. The software also provides methods for validation and accuracy measurement. The package thus comprises a production environment for big EO data analysis. We show that this approach produces high accuracy for land use and land cover maps through a case study in the Cerrado biome, one of the world’s fast moving agricultural frontiers for the year 2018.

Continuous land cover change detection in a critically endangered shrubland ecosystem using neural networks

“…These 1D CNN layers are followed by non-linear activations and potentially also batch normalization and stacked, before class probabilities are produced by a dense and softmax layer. TempCNN performs well on a range of land cover classifications tasks, and has been adopted by multiple packages for classification of satellite image time series (Rußwurm et al, 2020a; Simoes et al, 2021).…”

A Framework for Multivariate Analysis of Land Surface Dynamics and Driving Variables—A Case Study for Indo-Gangetic River Basins

“…In this connection, Sudmanns et al [21] provides an overview of selected architectures and data portals allowing the access to and the analysis of large amounts of EO data, particularly ARD, with the Google Earth Engine [22] being probably the most prominent example. A further solution to store, organize, and analyze EO data is the data cube environment [21,23,24]. In this regard, there is a variety of operational data cubes at regional scale using the Open Data Cube framework [25][26][27][28][29][30].…”

Ha Long—Cam Pha Cities Evolution Analysis Utilizing Remote Sensing Data

“…Indeed, in this recent time-series approach, a set of satellite image scenes taken at many different times-Satellite Image Time Series (SITS) affords a large amount of information compared to a single image couple in the context of temporal tendencies of regional evolution [46]. Despite these benefits, it still raises specific challenges regarding: the irregular temporal phenological signature of different land cover types; the insufficient sampling used to train the supervised classification; the missing temporal data [42]; the network architectures or specific datasets shaping that need to be developed for exploiting the temporal information jointly with the spatial and spectral information of the data [47]. Thus, in a more classical way, other sets of approaches and methods can be used varying from manual change interpretation [48] to bi-temporal linear data transformation [49] or multi-temporal spectral mixture analysis [50] and deep learning [51].…”