The National Space Research Institute (Inpe) is developing an ambitious computer program for climate modeling, which is being coordinated by the climatologist Carlos Nobre. This program, for the first time ever in Brazil, includes information about the flow of humidity, heat and carbon dioxide in the world’s atmosphere and oceans and the earth’s surface. To use this global climate modeling program to generate climate change scenarios, among other things, a supercomputer manufactured by the American company Cray, capable of performing 224 trillion operations per second, has been bought for R$50 million (R$15 million from FAPESP and R$35 million from the Ministry of Science and Technology). The new supercomputer arrived in Brazil in October, in 84 boxes, and this month should be fully assembled on the INPE campus in Cachoeira Paulista, a town in the Paraíba Valley, in São Paulo state. The processing capacity of the new computer is 50 times bigger than that of the computer in use at INPE?s Weather Forecasting and Climate Studies Center.
The new machine should house a computer program officially called the Brazilian Global Climate System Model, whose main function is to become the basis for research into the effects of climate change on human activities. Therefore, because it can predict plausible climate scenarios in 10 (with a spatial resolution of up to 10 x 10 km) or 100 years time (with a spatial resolution of up to 50 x 50 km), it will be useful for farmers to plan the design of Brazilian agriculture in the future, for economists to anticipate the risk to their business and for public managers to adjust transport and the occupation of cities and coastal areas.
“We hope the conclusions of the new climate model will help reduce the damage caused by climate change. Are epidemics going to emerge that we could have prevented? Do we need more hydroelectric power stations? Or should we reinforce the production of wind or solar energy?” says Paulo Nobre, an INPE researcher and coordinator of the Brazilian global climate system model. “The recent extreme weather episodes are a warning, reinforcing the need for all of us to come up with more reliable climate forecasts to avoid even greater damage”. He is referring to the drought that damaged the wheat crop in Rio Grande do Sul at the beginning of the year, the one that made the bed of the Amazon and Negro rivers retreat like never before and floods, such as those in Pakistan, which left 4 million people homeless.
The construction of the new weather forecast program involves some 50 researchers, mobilized by the Brazilian Research Network on Global Climate Change – Climate Network, and the National Institute of Science and Technology on Climate Change (INCT-MC), based at INPE. Early results indicate the worrying dimensions of the climate changes that are under way in the Brazil. A preliminary version of the surface module, being prepared at the Federal University of Viçosa (UFV) and at INPE, showed that deforestation in the Amazon and the Cerrado may increase and then reduce the volume of water in the region’s main rivers, because of changes in the circulation of air masses in the lower atmosphere; the flow in the Araguaia River, for example, has increased by 25 percent (see Pesquisa FAPESP nº 164). Paulo Nobre and other researchers at INPE, using an intermediate version of the new climate model that was fed with data from the last 20 years of climate variations on the continent and in the oceans, showed that the hypothetical loss of the entire Amazon Rainforest would reduce the amount of rainfall in the Amazon, due to an increase in the frequency of El Niño in the equatorial Pacific.
The project for the Brazilian model of the global climate system involves close international collaboration. An example is the program that evaluates the impact on climate of the smoke from forest fires, which is being developed by researchers Karla Longo and Saulo Freitas, from INPE, and that deals with forest fires, the result of joint work of team from INPE and from the Hadley Centre in the United Kingdom. A preliminary version of the modules that assess the impact on the global climate of plumes of smoke and forest fires in Brazil is already being tested at the Hadley Centre’s climate system model. According to Nobre, these two programs will represent the original Brazilian scientific contribution to the formulation of climate change scenarios that the Intergovernmental Panel on Climate Change (IPCC), the world’s most influential scientific institution in the area, intends to present in 2014. “Climatologists from South Africa are also developing a computer model of the global atmosphere, which will use the surface module developed by the UFV and INPE”, Nobre comments. This is not the first time that Brazilian climatologists have created programs with bold purposes. In 2007, a team from INPE, coordinated by José Marengo, concluded a regional climate model that detailed information from global models and showed an increase in mean annual temperature and rainfall intensity in Latin America (see Pesquisa FAPESP nº 130).
In three years, when the new global climate model is ready, Brazil will enter the limited group of countries that can deal first hand with information about the behavior of the climate in their own territory and check how what happens here has repercussions in the world, thus reconciling local and global phenomena. So far, few countries – the United States, the United Kingdom, Germany, France, Japan, Canada and Australia – have their own climate models, seen as strategic because they help identify local problems, such as the formation of storms, droughts or heavy rainfall. “We cannot expect scientists from other countries to solve our problems”, says Marcos Heil Costa, a professor from UVF who is coordinating the development of the surface module. “In Brazil”, he says, “the impact of changes in land use, such as deforestation in the Amazon region and the Cerrado savannas, may be more serious than that caused by rising levels of carbon dioxide in the atmosphere”.
As one of the projects of the Fapesp Research Program on Global Climate Change, the Brazilian climate computer model should bring together three distinct modules that analyze the weather phenomena that occur in the atmosphere, in the ocean and on land. According to Nobre, a part of the new program, which combines an atmospheric component of the CPTEC global model and the ocean component of the ocean model (MOM4) of the Geophysical Fluid Dynamics Laboratory, of the National Atmospheric and Oceanic Administration (NOAA) in the United States, is already functioning on INPE’s NEC supercomputer, which provides global forecasts of the surface temperature of the sea and of events such as El Niño, in the Pacific Ocean, months in advance. This computer module that links climate forecasts in the atmosphere and oceans should shortly begin working on the Cray supercomputer at INPE, which is expected to come into operation in early 2011 in Cachoeira Paulista.
This is just the beginning. Gradually, the ocean-atmosphere core should be integrated with other parts of the program that deal with the effects of vegetation, forest fires, rivers and the carbon cycle on the earth’s surface and of sea ice on the climate in Brazil and worldwide. “Until three years ago, none of us believed that it would be possible to make a computer model in Brazil that was capable of running on high performance machines, with hundreds of integrated processors working simultaneously”, says Costa. Widely reported news about the likely impact of climate change has alerted public entities and mobilized researchers around a pioneering project.
The new climate program did not start from scratch. The 1.0 version, adapted from a program on weather in the atmosphere, was already running on CPTEC’s computers. Complementary programs for the 2.0 version were already half-finished or ready, having been developed at INPE (as is the case of the atmospheric chemistry and aerosols model) or released free of charge by institutions such as the United States Space Agency (NASA), the National Center for Atmospheric Research (NCAR), the Woods Hole Oceanographic Institute, and Princeton University and the University of Wisconsin, also in the United States. Partnerships with the Indian Institute of Science (IISc) and the Council for Scientific and Industrial Research (CSIR), from South Africa, are also part of the project. Of course, the programs alone would be insufficient if the researchers did not know to handle them. In addition to organizing the groups and setting common directions, it was necessary to tailor the programs to meet Brazilian needs and to understand – sometimes adjust – the mathematical equations that govern the information and that indicate, for example, the intensity of rainfall in the north or south of the country.
Another challenge was to create appropriate methodologies to take advantage of already existing databases, such as that of the Large-Scale Biosphere-Atmosphere Experiment in the Amazon Region (LBA), an international research program that started in the late 1990s, and of the research network with buoys anchored in the tropical Atlantic Ocean (Pirata), initiated in 1998 in a partnership between INPE, the Board of Hydrology and Shipping, NOAA in the United States and the IRD and MetéoFrance in France. Something that is equally difficult for any research group to do is to integrate the information of groups that are working with different approaches, emphasizing vegetation, the oceans, rivers or human activities, like deforestation, and crop and livestock farming, and on different geographical scales, with a more global or more detailed view, according to their own needs.
A program called a flow coupler that interacts with other programs solved this problem. “The flow coupler is like an online translator that gives the rainfall calculation concluded by the atmospheric module for the surface program. The surface program, in turn, is going to work with this information, which is going to indicate that the rivers are going to fill more, and then it notifies the ocean program that there’s more water on its way”, explains Nobre. In this way, each module can have a degree of independence of scale and programming language, provided they can interact with the flow coupler. This approach has diluted the differences between parts of the computer climate model. “Until a few decades ago, climate science did not look at the ocean, which is much more difficult to study than the atmosphere, in addition to being very large”, says Edmo Campos, a researcher at the Institute of Oceanography (IO) of the University of São Paulo (USP ), who is coordinating a group from USP that is helping to develop the ocean forecasting module. “Oceans take much longer than the atmosphere or the continent to heat up and cool down. Temperatures don’t change that suddenly, but afterward heat release may also take longer.”
The arrival of ocean climate prediction programs is timely, because various studies indicate that the temperature of the waters of the South Atlantic, which bathe the southern and southeastern coastline of Brazil, is rising. One reason for this, confirmed by the program of the team from USP, is that the Atlantic is getting more hot water from the Indian Ocean due to changes in wind currents. Warmer waters in the South Atlantic increase the risk of torrential rains, like those in April 2010, and hurricanes, like Catarina. An extremely rare phenomenon in Brazil, Catarina arrived in March 2004 after forming waves five meters high out at sea, and destroyed 100,000 homes and crops of rice and bananas in 40 towns in the states of Santa Catarina and Rio Grande do Sul.
However, computer programs alone are not enough to avoid tragedies like these. “Before Catarina reached the mainland, the staff from NOAA warned that the pressure was very low and the eye of a hurricane was forming, but we had no way of confirming it. We failed in our forecast and we had no monitoring instrument in the region. Strictly speaking, we still don’t have any”, says Campos. He said that to fill this gap, a team from the Climate Change INTC is working with engineers from a company in Rio de Janeiro to assemble a fiber glass buoy three meters in diameter that is likely to be anchored with train wheels 300 km off the Santa Catarina coast in water 4,000 meters deep. This should happen in June or July 2011 and will be to monitor variations in temperature and salinity of the upper layer of the ocean and atmospheric variables (wind, temperature, pressure, humidity and solar radiation) on the surface.
The integration of programs on the climate in the atmosphere, in the oceans and on land should provide a fuller picture of the relation between extreme weather phenomena in Brazil and the planet. “In September, in a lecture at the Agribusiness Fair in Londrina, I showed how deforestation in the Amazon could increase climate variability in Paraná”, says Nobre. “Then I asked if we shouldn’t include the cost of maintenance of the Amazon rainforest in each sack harvested in Paraná. At that moment I was not defending the Amazon Region, but food production, which is threatened by floods and worse droughts.”
The waters of the Amazon
Doubts about conceptual problems should be dispersed. One of them: on the computer, a line called the thermo-climate line, which separates hot and cold water according to depth, indicates that there is more cold water in the middle of the sea, rather than around the edges, which is what is observed in reality. “Ocean models are perhaps failing to reproduce the real thermal structure of the oceans, among other reasons because they disregard the discharge of fresh and colder water from the Amazon, Orinoco, São Francisco and Prata Rivers, whose effect we are now going to examine carefully”, says Nobre. “A recent study in Science showed that the discharge of water from the Amazon increases the consumption of CO2 in the Atlantic region near the mouth by a factor of ten, because the biological activity induced by nutrients washed down by river water absorbs more CO2. We never imagined this effect or this volume. We just thought about the thermal effect regarding the CO2 balance between the atmosphere and the sea”.
Paulo Nobre expects the number of users of the climate prediction program to grow rapidly. The possibility of each module being able to run on ordinary computers – even on laptops, with some effort – and being distributed free, with an instruction manual, should make their widespread use easier. “Our objective is to prepare hundreds of people to work with these programs in Brazil”, Costa says. “Maybe we can create a panel of experts on the same level as in other countries, since this is a strategic project for Brazil. At present, we’re where other groups were 10 years ago, but in a few years we want to be equal to countries that have their own global climate system models”.
Brazilian model of the global climate system (nº 09-50528-6); Type Thematic Project PFPMCG/Pronex FAPESP; Coordinator Carlos Afonso Nobre – Inpe; Investment R$ 579.200,0 e R$ 176.628,0