Coordinated by DPE/COAGRO, the project of modernization of the agricultural statistics aims at improving the use of administrative records and including Remote Sensing (RS) as a tool for agricultural statistics, reducing the subjectivity of some surveys and thus fostering new approaches. The Coordination has made efforts to assemble a team and create conditions to meet these goals.
SR modeling studies of the national agriculture require robust computer resources in its initial phase, which is a limitation for the IBGE. To do this, Ian and Octávio Oliveira, COAGRO´s researchers, gathered in October 2023 with a representative from the LNCC to know the Call for Use Proposals - 2023/01, and submitted a proposal in the end of November 2023 a research proposal called “Representational Learning for the Segmentation of Culture in Brazil”, to use Santos Dumont supercomputer for developing the project. On January 15, 2024, the proposal was approved by the MCTI´s (Ministry of Science, Technology and Innovation) LNCC (National Laboratory of Scientific Computation), the office responsible for the equipment, based in Petrópolis.
External view of the installation of Santos Dumont supercomputer (left) and of the LNCC in Petrópolis (RJ). Source: LNCC website.
This is the initial step of the project that will incorporate the use of remote sensing data into the production of agricultural statistics at the IBGE. In the first phase of the modernization project, the free access to the computer capacity available in Santos Dumont will allow to infer planted areas of selected crops.
In a second step, methods developed and tested in the LNCC infrastructure should be run in the IBGE to allow the incorporation of the information produced, in a continuous and permanent form.
We are very happy to share the beginning of this new phase with all of our IBGE colleagues, hoping to produce processes and results that might encourage and support the adoption of data science and machine learning techniques in modernization projects of official production of information.