Identification of vulnerable areas in agricultural production centers serves as a warning to farmers and public authorities
Research developed at Unicamp reveals that the State of Bahia has been experiencing an increase in the aridity index and a decrease in rainfall. Studies indicate that the situation is likely to worsen over the next 30 years, causing an increase in areas at risk of desertification in the region. The conclusions are from the doctoral thesis “Areas at risk of desertification: current and future scenarios, in the face of climate change”, defended by Camila da Silva Dourado at the Faculty of Agricultural Engineering (Feagri) at Unicamp.
Desertification is the degradation of land in arid, semi-arid and sub-humid to drought areas, as a result of climate variations and anthropogenic actions, that is, changes caused by humans in the environment. This phenomenon transforms fertile and arable land into unproductive land, causes environmental impacts such as the destruction of biodiversity, decreased availability of water resources and causes the physical and chemical loss of soil. In this case, the research shows that the mesoregions that most expanded areas at risk of aridity are the largest agricultural centers in Bahia. “A more in-depth analysis of desertification in these areas is still needed, but the data shows that these observed agricultural hubs are now considered high-risk areas”, explains Camila.
The work was carried out under the guidance of Stanley Robson de Medeiros Oliveira, researcher at Embrapa Informática Agropecuária and co-supervision of Ana Maria Heuminski de Avila, researcher at Cepagri (Center for Meteorological and Climate Research Applied to Agriculture). The authors warn of the need to adopt preventive measures now so that the predictions do not become consolidated.
In the scenario of national agricultural production, Bahia stands out in the Brazilian Northeast as a major grain producer, in addition to being responsible for 12,2% of the value of fruit production, occupying second place in the national ranking. Cotton cultivation in the state represents 25,4% of national production, second only to Mato Grosso with 64,1% of production, according to data from the 2016 harvest released by IBGE (Brazilian Institute of Geography and Statistics). The two main agricultural centers in Bahia are in the west, in cities such as Luís Eduardo Magalhães and Barreiras, for example, where cotton and grain production is strong, especially soy. Another hub is in the north of the state, the largest fruit producer in Bahia, with highlights being the cities between Juazeiro (BA) and Petrolina, in Pernambuco.
“The scenarios of an increase in areas at risk for agriculture due to desertification threaten several economic and social sectors in the region, mainly agriculture”, explains Camila. Therefore, one of the recommended alternatives is the development of intelligent tools and systems capable of capturing, organizing and quantifying data and information, which assist in agricultural production planning and the decision-making process, with the aim of reducing environmental impacts.
According to the results obtained through the analysis of climatic data (rainfall, temperature and evapotranspiration), edaphic data, terrain slope, soil fragility to erosion and vegetation (extracted from satellite images), between the years 2000 and 2014, the Bahian territory already showed a drop in the level of precipitation (rains), a decrease in areas of vegetation cover native, and an increase in the aridity index and areas at risk of desertification.
For the future, that is, between the years 2021 and 2050, the forecast is that the State will face an increase in temperature of approximately 1 °C and a decrease in precipitation, in relation to the current climate. Forecasts also point to an increase in areas considered arid and an expansion of lands with “high” and “very high” risk of desertification. “This research displays the future scenario; So, if we want to minimize these risks, we have to make decisions and actions now or it will be too late to make corrections. We cannot wait until 2050”, warns Stanley Oliveira.
According to the research advisor, data mining techniques associated with remote sensing techniques in orbital images address the challenge of capturing patterns and processes, and provide a spatiotemporal diagnosis of changes in the landscape, also allowing for monitoring and diagnosing the degree of degradation. of land. These techniques facilitate the analysis and manipulation of data in large areas, with less cost than conventional methods, allowing an assessment of changes occurring in the environment, in the past, present and with simulations of the future.
“Depending on the agricultural practice adopted today, productive land will be transformed into unproductive. There is no point using inappropriate practices that do not aim at the sustainability of that soil and natural resources. It is necessary to alert large and small producers about forms of production that alleviate this situation; It's a matter of raising awareness. Public policies are also necessary to encourage new forms of production and use of land and natural resources”, highlights Camila.
The challenge of feeding a growing population
The results that the research points out are important for the search for solutions to the main challenge of world agriculture: meeting the goal of feeding nine billion people by 2050, according to a forecast from FAO, the Food Organization of the United Nations (UN). and Agriculture. Studies of agency indicate that, to feed an extra population projected at an additional 2,3 billion people, o The world will need to produce 70% more food. Nonetheless, the expansion of arable land it will have to occur on around 120 million hectares over the next 40 years in developing countries, mainly in Latin America and sub-Saharan Africa.
Arid regions and desertified lands hinder and impede food productivity. Previously arable land becomes unproductive due to semi-aridity, aridity and desertification processes. It is estimated that a large part of the land located in areas with a climate prone to desertification will have its process of becoming unproductive accelerated. Therefore, the results are important to guide the work of managers and support the formulation of public policies focused on the region.
Historically, the northern region of the Bahian territory is part of the drought polygon, an area of more than 1 million km² between the States of Alagoas, Bahia, Ceará, Minas Gerais, Paraíba, Pernambuco, Piauí, Rio Grande do Norte and Sergipe, which faces repeated drought crises. Therefore, fruit growing in the north is maintained through irrigation systems. However, another problem highlighted by the research is that regions previously considered to have a low risk of desertification become moderate and high, as is the case in the western region. This situation would change the entire agricultural production scenario in the state.
In recent years, researchers have also been concerned about the influence of climate change on the advancement of the desertification process. “With the increase in temperature estimated at 1 ºC and the decrease in precipitation, there is the occurrence of another indicator that we use, evapotranspiration, which is a subsidy for another indicator, the aridity index. Putting these variables together, and with the new projections from the climate change model, it is confirmed that there is a very large expansion of desertification areas”, explains Camila.
“If we consider the results, this is a very drastic and frightening scenario, but the objective of this research is not to scare but to inform. It is time to create public policies so that people who live off the land can have a better quality of life, can stay and feed their families, because the great risk is that they will migrate to other regions and become marginalized. People need to continue producing food for subsistence and commerce”, says the research co-supervisor, Ana Avila.
Exclusive
To evaluate areas with potential risk of desertification in the state of Bahia, seven biophysical indicators of desertification were used: normalized difference vegetation index and enhanced vegetation index (NDVI and EVI, respectively), both generated by the sensor Modis; aridity index; soil data; precipitation; temperature and evapotranspiration. In the case of climate maps, the Empirical Bayesian Kriging geostatistical method was applied. Model maps of elevation, slope and soil classification were also created, with the aim of generating a soil fragility map, used as an indicator, with the edaphic characteristics of the region.
From the stacking of images of the seven desertification indicators, the classification task was applied, using the Support Vector Machines (SVM) algorithm on the product image, defining four levels of desertification risk: very high, high, moderate is low. The research used high spatial resolution images from the RapidEye satellite to validate the classification. The simulations of the impacts of climate change for the future scenario, 2021 to 2050, used Eta-MIROC5 climate models, which predict a decrease in precipitation, an increase in temperature and the displacement of areas with higher rates of potential evapotranspiration.
Two scenarios with different time periods were studied: the current climate scenario, covering the years 2000 to 2014, and the future climate scenario, for the period 2021 to 2050. The results showed that in 2014 there was a decrease in precipitation and areas of vegetation cover in relation to the year 2000, in addition to an increase in the aridity index and areas at risk of desertification. In the future scenario, there was an increase in temperature of approximately 1 °C and a decrease in precipitation in relation to the present climate. The aridity index points to an increase in arid areas for the future climate, and an expansion in areas at risk of desertification, mainly in areas of very high and high risk.
“This methodology is unprecedented and rich in that it uses remote sensing and data mining techniques, including an intelligent algorithm (SVM – Support Vector Machine), which learns interactively from a mass of data, discovers and presents the patterns found. This can be applied to other regions of Brazil, especially those most in need; the contribution is not restricted to the state of Bahia”, emphasizes Oliveira.