Identifying Spatiotemporal Patterns in Land Use and Cover Samples from Satellite Image Time Series

Abstract

The use of satellite image time series analysis and machine learning methods brings new opportunities and challenges for land use and cover changes (LUCC) mapping over large areas. One of these challenges is the need for samples that properly represent the high variability of land used and cover classes over large areas to train supervised machine learning methods and to produce accurate LUCC maps. This paper addresses this challenge and presents a method to identify spatiotemporal patterns in land use and cover samples to infer subclasses through the phenological and spectral information provided by satellite image time series. The proposed method uses self-organizing maps (SOMs) to reduce the data dimensionality creating primary clusters. From these primary clusters, it uses hierarchical clustering to create subclusters that recognize intra-class variability intrinsic to different regions and periods, mainly in large areas and multiple years. To show how the method works, we use MODIS image time series associated to samples of cropland and pasture classes over the Cerrado biome in Brazil. The results prove that the proposed method is suitable for identifying spatiotemporal patterns in land use and cover samples that can be used to infer subclasses, mainly for crop-types.

Satellite Image Time Series Analysis for Big Earth Observation Data

“…The quantitative evaluation confirms this intuitive insight. As shown by Santos et al [36], removing these and other outliers improves classification results.…”

“…The sits package provides an innovative sample quality control technique based on self-organising maps (SOM) [35,36]. SOM is a dimensionality reduction technique.…”

Estimation of the Conifer-Broadleaf Ratio in Mixed Forests Based on Time-Series Data

“…Traditional k-means clustering aims to partition n observations into k clusters and is widely used for solving time-series data clustering problems [40,41,58]. The main principle of the semi-supervised k-means or k-means clustering algorithm is the minimization of the total distance (typically the ED) between all objects in a cluster from their center [42,[58][59][60], where the cluster center is defined as the mean vector in the cluster [41,43]. Before applying the clustering algorithm, the centroid values for the initial clusters and the optimal number of clusters must be predefined, which also determines the final clustering accuracy [40][41][42].…”

“…Time-series clustering analysis is generally unsupervised or semi-supervised; it groups time-series objects according to their similarity such that the similarities among objects in the same group are maximized while the similarities among objects across groups are minimized [40,41]. Time-series clustering analysis has attracted increasing attention for forest remote sensing owing to its unique advantages, such as reduced sample data requirements, easy implementation, and high accuracy [40][41][42][43]. In this paper, prior information was introduced into machine learning algorithms, and a similarity measurement method based on the Euclidean distance was used.…”

Towards Lifelong Crop Recognition Using Fully Convolutional Recurrent Networks and Sar Image Sequences

“…The authors in (Ajadi et al, 2021) proposed a large scale crop mapping scheme leveraging auxiliary data from the United States, using a XG-Boost model. Similarly, the authors in (Santos et al, 2021) used self-organizing maps (SOMs) to classify crop types using MODIS time series. However, their training features only considered the temporal dimension, leaving the spatial context unexplored.…”