Two research projects in the field of meteorology, funded by the Microsoft Research-FAPESP Institute for IT Research, have provided new data on and the possibility of understanding local micro climates, the interaction between forests and the atmosphere, and the effects of climate changes on agriculture. Researchers working on one of these research projects are developing geosensors that will be installed throughout the Amazon Rainforest to form a wireless network. This network will capture and transmit local environmental data, such as temperature and humidity, for example, of a three-dimensional slice of the environment. The other research project created software and mathematical models to analyze and predict the weather in specific regions.
The objective of the second project, named Agrodatamine, is to understand the correlation between various climate and agriculture analysis parameters and to improve agro climate models, evaluating and cross-checking terabytes of data collected by sensors placed in the soil, by meteorological radars, and by satellites. “Our goal is to create models and mathematical algorithms that will enable us to identify tendencies and establish correlations in this data to help agro meteorologists make more accurate forecasts and quick decisions,” says Agma Juci Machado Traina, a professor at the Mathematics, Computer, and Biosciences Institute of the University of São Paulo (ICMC-USP) in São Carlos, and coordinator of the research project.
She explains that, due to technological advances related to the collection of climate data, the resulting information is much heavier than the existing capacity to analyze it. This is why it is necessary to develop new computer techniques to explore this enormous volume of data and produce knowledge for agrometeorology. “Our project is based on the premise that the search for associations and exceptions related to meteorological data could help us to find correlations and identify seasonal and extreme behaviors, thus enabling a more accurate interpretation of the associated climate phenomena,” she says.
Theory of Fractals
The analysis of data-related behavioral variations over time also resorts to fractal theory, developed by the French mathematician Benoit Mandelbrot in the 1970s. The fractal theory measures or classifies complex situations that are not based on traditional geometry. In the case of the Agrodatamine project, the fractal theory supports the mapping of the distribution of a vast number of data. It also helps identify temporal patterns, especially when the focus is on the monitoring of multiple series, such as the integrated analysis of the behavioral evolution of rainfall measurements and the highest and lowest temperatures over a specific period of time.
The results of the research work include three computing tools. One is SatImagExplorer, a software program that is fed with satellite images. Based on these, the program enables researchers to measure and analyze events occurring in the monitored region over the period during which the images were collected. The ClimFractal Analyzer is another software program, to conduct dynamic analyses – based on the fractal theory – on climate data from real meteorological stations or on data generated by climate models. The third tool is a program called TerrainViewer, which presents and enables the manipulation of three-dimensional models of terrains by means of altimetry data and images obtained through remote sensors, with different spatial resolutions.
“All this information can be used to provide support for agrometeorology and climatology specialists, enabling them to conduct analyses quickly and more accurately,” says Agma. The research work was conducted in partnership with Embrapa Informática Agropecuária, the State University of Campinas (Unicamp), the National Space Research Institute (Inpe) and the Federal Universities of São Carlos and of the ABC.
The objective of the second research project is the development and installation of the geosensors network for the environmental monitoring of the Amazon Rainforest. This project is coordinated by meteorologist Celso Von Randow, from Inpe. “Our work will lead to a better understanding of how the forest interacts with the atmosphere, how it influences the micro climate, and how the micro climate affects the forest and the ecosystem.
The project will also develop computational tools for the mining and analysis of environmental data, as well as information transmission studies on wireless networks. In the field of environmental sciences, the goal is to understand aspects of the forest’s micro climate. To put these objectives into practice, the researchers installed a network of geosensors in Serra do Mar State Park, near the town of São Luís do Paraitinga, State of São Paulo. The geosensors are in the form of small boxes equipped with an electronic device for data collecting and an antenna, to which the sensors are attached. Four sensors in each box were used in the experiment – three temperature sensors and one humidity sensor. “They were developed at Johns Hopkins University, in the United States. This university is a partner of ours on this project,” says Randow. “We are now developing similar sensors at Inpe; ours are ceramic sensors, which are more resistant than the ones available in the market.”
Six towers – a central one plus five surrounding towers – were installed in the Mata Atlantica Rainforest for the pilot project. The towers are ten meters tall, which is higher than the tree canopies. The towers are interconnected with cables – that function like clotheslines – on which some of the boxes were hung. Other boxes were placed on trees, some of them one meter above the ground, while other boxes were placed at different distances, up to the forest’s canopy. Fifty two boxes were installed with a total of 208 sensors, covering an area of ten thousand square meters.
“The circuit boards record the data and transmit it to each other through the wireless network,” Randow explains. “This results in a type of three-dimensional picture of the region’s environmental conditions, showing how the temperature or the humidity of the air varies from one point to the next and from one altitude to another.”
Another innovation developed during the project was the technology to extend the life of the batteries in the sensors. The batteries are switched off and only “wake up” once a minute to check if there is any contact with the researchers’ notebooks. From time to time, the researchers go out to collect the data stored in the boards. “The next step is to install a pilot project over a bigger area – one square kilometer – in the Amazon Rainforest,” says Randow. “In the next experiment, we are going to test the sensors that are being developed.”
1. Agrodatamining: development of data mining methods and techniques to support research on climate changes with emphasis on agrometeorology (nº 2009/53153-3); Modality Regular Funding of Research Project; Coordinator Agma Juci Machado Traina – USP; Investment R$ 178,631.48 (FAPESP)
2. Development and application of a geosensor network for environmental monitoring (nº 2009/53154-0); Modality Regular Funding of Research Project; Coordinator Celso Von Randow – Inpe; Investment R$ 216,957.00 (FAPESP)
GONÇALVES, R.R. et al. Analysis of NDVI time series using cross-correlation and forecasting methods for monitoring sugar cane fields in Brazil. International Journal of Remote Sensing. v. 33, n. 15, p. 4653-72. 2012.