IEEE Geoscience and Remote Sensing Magazine - September 2019 - 111

use in three different case studies related to the identification
of vegetation types based on temporal profiles, the identification of suitable time-series dissimilarity functions in the
recognition of eucalyptus pixels, and the assessment of the
recovery of a vegetation region subjected to severe drought
events. This tool is available on GitHub, and some of its components have already been well received by the target community, notably with their recent integration into a private
cloud-computing platform.

in an Amazon Basin region subjected to extreme
drought events.
◗ discussion of a component that implements a recently
proposed soft-computing framework for the discovery of time-series dissimilarity combination functions
in the context of eucalyptus pixel classification
◗ extension of the BE to support structural-break detection based on time windows and smoothed time series.

User

TUCUMÃ
BACKGROUND
This section presents the proposed toolbox, highlighting
The detection and characterization of temporal changes are
its main modules.
important indicators for understanding how complex phenomena behave and evolve. Particularly in environmental
DATA ACQUISITION MONITOR
studies, those temporal changes carry plenty of useful inThe DAM module supports data acquisition. The current
formation that may be used, for example, to define public
version allows downloading of Landsat time series as either
and research policies or support the analysis of different
single pixels or entire images. Figure 2 provides functional
ecological phenomena related to climate change, carbon
and architectural views of the tool. First, users define locredits, and biodiversity [1], [2]. Despite the large applicabilcations of interest by providing either a comma-separated
ity of temporal-change analysis methods, several tools have
values (CSV) file containing a list of pixel coordinates
been proposed to account for more detailed elements and
(Pixel tab) or a shapefile (Raster tab). A specific engine
further improve the understanding of complex spatiotemis in charge of automatically downloading the requested
poral changes [3]-[7].
data from Google Earth Engine. Finally, obtained prodThe use of remote sensing images and techniques has
ucts are made available to users for further analysis.
been thoroughly pursued over the past decades to detect and
investigate temporal changes on Earth's surface [8], [9]. The
TIME-SERIES RETRIEVER
use of appropriate tools [10]-[14] that can support complex
In Figure 3, we present a functional overview of the TSR modanalyses of image time series is of paramount importance
ule. In this module, the user is expected to provide-through
in various domains, from the purely scientific to the eduthe Input tab-preprocessed, stacked remote sensing imcational. However, although there are many existing tools,
ages. After the user provides those data, it is possible to
it is often necessary to switch among them several times.
interact with the Exploratory and Retrieval tabs. In the
Typical interoperability issues, therefore, range from system
E xplorator y tab, the user may visualize raster layers at
mismatch to the variety of user profiles and data types.
different time stamps and select a specific spatial point to
In summary, our contributions are as follows:
view its entire time series. In the Retrieval tab, the user
must first select a spatial point to compute the distance of
◗ presentation of an in-depth overview of the main comthis point to all others, generating a distance map. Based
ponents of the tool, which may guide possible users and
on the distance map, the TSR module computes the elbow
developers in their customizations for different applicaplot and runs the k-means algorithm with the number of
tions and domains
clusters (k) chosen by the user. All resulting images/plots
◗ explanation of possible usage scenarios in the context
are presented in the user view.
of different relevant scientific investigations, which depend on the detection and characterization of temporal changes
* With regard to the DAM and
DAM
TSR modules, we consider the
Provides
BE
TSR
identification of vegetation
CSV
Pixel Tab
ROI
types in the Emas National
GPC
Requests
Definition
Park region based on tempoData
ral profiles.
Provides
Shapefile
* In relation to the GPC module,
Raster Tab
we address the identification
of suitable time-series dissimiProvides
Time-Series
larity functions in the recogniData
tion of eucalyptus pixels.
Download Module
* With regard to the BE module, we discuss the assessment of vegetation recovery FIGURE 2. A functional overview of the DAM module. ROI: region of interest.
SEPTEMBER 2019

IEEE GEOSCIENCE AND REMOTE SENSING MAGAZINE

111

IEEE Geoscience and Remote Sensing Magazine - September 2019

Table of Contents for the Digital Edition of IEEE Geoscience and Remote Sensing Magazine - September 2019