The technique merges two core components: Brazil's Data Cube architecture, developed by the National Institute for Space Research (INPE), and the Geographic Object-Based Image Analysis (Geobia) method. Leveraging satellite image time series from NASA's Moderate Resolution Imaging Spectroradiometer (Modis), the researchers accurately tracked seasonal crop rotations like soy and corn in the state of Mato Grosso.
By incorporating machine learning algorithms, the system achieved a 95% accuracy rate in land use classification. Geobia enhances interpretation by segmenting satellite images into geo-objects based on similarities in texture, shape, and reflectance. Meanwhile, data cubes organize spatial and temporal data, enabling efficient aggregation and visualization of features such as annual cropping patterns.
Current pixel-based analysis methods often encounter issues along the boundaries of different land types. "Scientific work has highlighted spectral confusion in border zones between different land uses as an area for improvement. So we decided to segment the images and evaluate the geographical object as the minimum unit of analysis, rather than the pixel. It's as if the image were broken down and classified according to each piece. In this way, we were able to reduce recurring edge errors and accurately identify the targets, even with moderate spatial resolution," said Michel Eustaquio Dantas Chaves, a professor at UNESP's Faculty of Science and Engineering and the paper's corresponding author.
Chaves has long applied data cube technology to assess land use change, particularly across Brazil's agricultural frontier in the Cerrado biome. He noted the adaptability of the method to data from other Earth observation satellites like Landsat and Sentinel, which his team is currently using to expand the research.
The methodology is detailed in the journal AgriEngineering's special issue on Agricultural Information Technology. The study was funded by FAPESP through three research grants: 21/07382-2, 23/09903-5, and 24/08083-7.
In practical terms, Mato Grosso provided an ideal testing ground due to its diversity in land cover and its role as a grain production leader, contributing 31.4% of Brazil's national output. Forecasts place the state's 2024/2025 harvest at 97.3 million tons, with soybeans accounting for nearly half.
The research team analyzed satellite imagery from the 2016/2017 harvest season to classify various land uses in the state, ranging from different crop combinations (e.g., soybean-corn, soybean-cotton) to sugarcane, urban development, and water bodies. The 95% classification accuracy confirmed the method's capacity to distinguish complex landscape features and support strategic planning.
"Since the approach manages to identify the targets in a consistent manner, the methodology can be applied to the estimation of areas within the same harvest, favoring productivity estimates; in territorial planning actions and anything that deals with land use and land cover for decision-making," explains Chaves.
He emphasized that the technique also enhances the detection of environmental disturbances. "It's quicker to detect deforestation than degradation. This method allowed us to detect these variations more quickly."
The publication includes a dedication to INPE researcher Professor Ieda Del'Arco Sanches, who passed away in January. "This article is a way of thanking her for her teachings and following her legacy. Ieda always worked to accurately assess the Earth's surface and to treat the data ethically and responsibly, showing how they can contribute to the construction of public policies," Chaves added.
Research Report:Mixing Data Cube Architecture and Geo-Object-Oriented Time Series Segmentation for Mapping Heterogeneous Landscapes
Related Links
Sao Paulo State University
Earth Observation News - Suppiliers, Technology and Application
| Subscribe Free To Our Daily Newsletters |
| Subscribe Free To Our Daily Newsletters |
