IEEE Geoscience and Remote Sensing Magazine - September 2019 - 120

drought caused by El Niño Southern Oscillation (ENSO)
climatic variability.
In Figure 16(b), we again detected a decrease in the
EVI2 value. With h = 0.21, the first break was detected
in mid-1998 and the last one in mid-2001. The second
break occurred at the beginning of 2000, indicating that
the vegetation took approximately a year and a half to
initiate its recovery (slightly longer than in the 1991/1992
drought) and another year to complete its recovery process. Again, by the dates of the detected breaks, we can
infer that the time pattern relates to the 1997/1998 ENSO-caused drought.
Finally, in Figure 16(c), we present the BFAST output
for the third time window, to which we set h = 0.18. The
first structural break was detected early in 2007 and the
last one in mid-2011. The second break was detected at the
end of 2009, indicating that the vegetation took almost
three years to start recovering (much longer than was observed in the 1997/1998 drought) but also fully recovered
in approximately one year. Although our objective was to
analyze the effects of the 2010 Atlantic drought, the slow
drop in vegetation productivity appears to have been
caused by a similar slow drought signal not directly related to the Atlantic but by some anomalies in the Pacific
(i.e., ENSO) combined with the 2010 drought. From the
U.S. National Oceanic and Atmospheric Administration
website [41], we can see a weak El Niño in the 2006/2007
summer, which may have caused the productivity drop.
That drop is followed by another weak El Niño in the
2009/2010 summer and, finally, the water deficit caused
by the 2010 severe drought.
CONCLUSIONS
This article introduced Tucumã, developed as a convenient
exploratory set of tools that permits easy and fast remote
sensing image analysis through time-series visualization,
structural-break detection, and dissimilarity searches. Several case studies, ranging from ecological to agricultural applications, are discussed to introduce possible usage scenarios of Tucumã components. Those usage scenarios include
the identification of vegetation types based on temporal
profiles, identification of proper dissimilarity functions to
guide the recognition of eucalyptus plantations, and analysis of the recovery process of vegetation subjected to severe
drought events. Also, we presented the use of a GP-based
framework [15], implemented as one component of the
toolbox, in the eucalyptus pixel classification problem. In
addition, we described extensions to the BE tool to support
time-series-break detection based on time windows and
time-series smoothing procedures.
Ongoing work is concerned with the extension of the
toolbox to support other kinds of data (e.g., MODIS and
near-surface images [39]). We also plan to incorporate resilience metrics to support studies on the stability of complex systems. The most up-to-date version of this toolbox
is available on GitHub (https://github.com/nathmenini/
120

tucuma-toolbox), as are the DAM and BE tool source files
in separate repositories [42], [43]. In addition, the BE tool
was recently integrated into the System for Earth Observation Data Access, Processing, and Analysis for Land Monitoring (SEPAL) of the Food and Agriculture Organization
of the United Nations, which is their cloud-computing platform. The open source files of the SEPAL project are also
located on GitHub [44].
ACKNOWLEDGMENTS
We thank the Coordenação de Aperfeiçoamento de Pessoal
de Nível Superior-Brasil (CAPES) (grant 88881.145912/2017
-01), Brazilian National Council for Scientific and Technological Development (grants 307560/2016-3 and 132847/
2015-9), São Paulo Research Foundation (FAPESP) (grants
2018/06918-3, 2017/12646-3, 2016/26170-8, 2016/08085-3,
and 2014/12236-1), and the FAPESP-Microsoft Virtual Institute (grants 2016/08085-3, 2015/02105-0, 2014/50715-9,
2013/50169-1, and 2013/50155-0) for financial support. This
study was financed in part by CAPES (finance code 001) and
the Serrapilheira Institute (grant Serra-1709-18983).
AUTHOR INFORMATION
Nathalia Menini (nathmenini@gmail.com) received her
B.Sc. degree with distinction in statistics in 2017 and her
M.Sc. degree in computer science in 2019 from the University of Campinas, Brazil. She is a data scientist at Quaasar
Machine Learning, São Paulo, Brazil.
Alexandre E. Almeida (almeida.xan@gmail.com) received
his B.Sc. degree with distinction in statistics in 2015 and his
M.Sc. degree in computer science in 2017 from the University of Campinas, Brazil. He is a cofounder of and data scientist
at Quaasar Machine Learning, São Paulo, Brazil.
Rubens Lamparelli (lamparel@g.unicamp.br) received
his Ph.D. degree in transport engineering from the University of São Paulo, Brazil, in 1998. He is currently a researcher, as well as an agricultural engineer, at the Interdisciplinary Center on Energy Planning, specializing in remote
sensing and geoprocessing. His research interests include
interactions of the environment, agriculture, and energy.
Guerric Le Maire (guerric.le_maire@cirad.fr) received his
diploma from AgroParisTech (French Higher Educational
Institution), his M.Sc. degree in ecology, and his Ph.D. degree in ecophysiology. He is a researcher at the French Agricultural Research Centre for International Development,
Montpellier, France, and is currently a visiting professor at
the University of Campinas, Brazil, where he develops projects related to monitoring eucalyptus plantations at the regional scale. He has published more than 60 articles in the
field of remote sensing or process-based forest modeling.
His research interests include remote sensing of forests, in
close relation with carbon and water budget modeling, and
the classification and estimation of forest characteristics
from satellite images, such as leaf area index, chlorophyll
content, or biomass, with many applications for eucalyptus plantations.
IEEE GEOSCIENCE AND REMOTE SENSING MAGAZINE

SEPTEMBER 2019

https://www.github.com/nathmenini/tucuma-toolbox https://www.github.com/nathmenini/tucuma-toolbox

IEEE Geoscience and Remote Sensing Magazine - September 2019

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