Brazil Data Cube is a project that is being developed by the Brazil’s National Institute for Space Research (INPE), since January 2019, that aims to create multidimensional data cubes of analysis-ready from medium-resolution Earth observation images….
On August 12, 13 and 14, 2020, the Brazil Data Cube project team conducted a virtual hands-on training on the classification of remote sensing image time series using the R package for Satellite Image Time software package Series (SITS).
The training was coordinated by Dr Gilberto Camara with support from the Brazil Data Cube (BDC) team members Alber Sanchez, Gilberto Queiroz, Felipe Carlos, Michelle Picolli and Rolf Simões. The training aimed at demonstrating, in a practical way, the use of the SITS package, using MODIS and Sentinel images. Members of the PRODES project team, graduate students in Remote Sensing and INPE researchers attended the trainning.
This was the first of two trainings planned for this month, to disseminate the use of SITS to process the data cubes generated in the BDC project.
The SITS package allows the retrieval of time series of satellite images using geographic services and methods for their visualization, analysis and processing. Different methods of classifying time series, using machine learning, are available in the package.
Between July 20-24, 2020, collaborators of Brazil Data Cube Project carried out field work to collect samples in field to support project activities.
Lucas Oldoni, doctoral student in Remote Sensing and José Guilherme Fronza, research associate of the project, master in Remote Sensing, both from Brazil’s National Institute for Space Research (INPE), traveled about 2904 km, leaving from Brasília, DF to the Luis Eduardo Magalhães region, in the west of the state of Bahia.
In the five days of work, points and photos of areas cultivated with cotton, corn, sorghum, millet, coffee, brachiaria, among others, were collected.
The work is a continuation of the survey carried out in march/2020 to understand the dynamics of the cultures that will be analyzed later by time series.
The collected samples will be used to train images classification algorithms and integrate sample database of the project that is being built with a system to facilitate researchers and students access, analysis and use this samples in automatic classifications processes using data cubes of the project.
Below we can see some images and photos of the field work:
Route traveled from Brasília, DF to Luís Eduardo Magalhães, BA.
Sentinel-2 image of 07/13/2020, false color composition, RGB: NIR, SWIR1, Red.
Bales of hay in a brachiaria area cut for haymaking.
Harvester carrying the cotton harvest.
Bales of freshly harvested cotton, ready for transport.
The researchers presented the project and participated in technical discussions with leaders of similar initiatives and projects in other countries to build Earth observation data cubes.
The session was mediated by Marie-Francoise Voidrot, Europe Director, Innovation Program – Open Geospatial Consortium, OGC and Prof. Thierry Ranchin, Director of the Observation, Impacts, Energy Center in MINES – ParisTech – PSL University.
Besides the Brazil Data Cube project, other initiatives as the “Swiss Data Cube”, “Data Cubes for Australia and Africa” and “Euro Data Cube” were presented by the researchers Gregory Giuliani (Head/Project Leader – Digital Earth Unit/Swiss Data Cube), Brian Killough (Head, CEOS Systems Engineering Office – NASA Langley Research Center), Fang Yuan (Assistant Director – Digital Earth Australia) and Grega Milcinski (CEO – Sinergise).
The GEO Data and Knowledge Week symposium, which would take place in Pequin, China, in February 2020, was postponed due de Coronavirus pandemic. It gave way to the GEO VIRTUAL SYMPOSIUM 2020 that will take place between the 15th and the 19th of June 2020 through virtual discussions and interactive webinars.
Dr. Karine Ferreira and Dr. Gilberto Ribeiro, represent the Brazil Data Cube project in a presentation on Data Cubes, along with other global initiatives, on June 16, 2020.