Scientists combine remote sensing and field experiments to study the effects of fire on Amazon savanna

Using experimental fires set at different times of year in the Campos Amazônicos National Park, Brazilian researchers found that lags in acquiring and assessing information obtained from satellite imaging can lead to misinterpretation. The study points to the need for adjustments to monitoring algorithms.

By Cristiane Paião  |  Agência FAPESP – The algorithms used to detect and monitor fires in savanna areas of the Amazon region should take into account the time of year at which the fires occur in order to estimate their effects on biodiversity more accurately.

This is one of the conclusions of a study by researchers at São Paulo State University (UNESP) and the Federal University of Rondônia (UNIR) in Brazil, in collaboration with environmental analysts at Chico Mendes Institute for Biodiversity Conservation (ICMBio), an agency of the Ministry for the Environment (MMA).

The study combined analysis of remote sensing and field data collected before and after experimental fires set at different times of the year in selected areas of Campos Amazônicos National Park (PNCA), a conservation unit located in the states of Amazonas, Mato Grosso and Rondônia. The controlled fires were managed in partnership with PNCA’s administrators and the local fire brigade. The aim of the analysis was to investigate the effects of fire on the vegetation in this area of savanna.

The findings of the study are reported in an article published in the Journal of Applied Remote Sensing.

The study was part of the postdoctoral research of Daniel Borini Alves supervised by Alessandra Tomaselli Fidelis, a professor at UNESP, and supported by FAPESP.

“In the Amazon savanna, as in other tropical savannas, fire scars observed by satellite correlate with the amount of combustible matter consumed by fire [technically termed fuel load]. If a fire occurs at the start of the dry season, which in the study area is April-June, the vegetation is still moist, and less combustible matter is consumed than by fires occurring in the period July-September. This entails varying degrees of complexity for the detection and monitoring of the effects of fire by airborne sensors,” Borini said.

The article reports only some of the results of the study, which is still in progress. The stage funded by FAPESP was conducted between January 2020 and April 2022. Articles with more results will be published shortly. The project is called Campos Amazônicos Fire Experiment (CAFE) and began in 2018 as a partnership with Antônio Laffayete Pires da Silveira, a professor at UNIR, and Bruno Contursi Cambraia, an environmental analyst responsible for management of fires and firefighters at ICMBio.

“All fires were directly supervised by the PNCA’s fire brigade to assure safe burning, and duly authorized under SISBIO license 67210-5,” Borini said.

CAFE is located in the largest enclave of tropical savanna in southern Amazonia and is home to many endemic species that frequently venture into the nearby Amazon Rainforest. The vegetation consists mainly of grass, shrubs and small trees with twisted branches. It is similar to the Cerrado, although species diversity is lower.

According to the latest report issued by the Brazilian Annual Land Use and Land Cover Mapping Project (https://mapbiomas.org/em-37-anos-amazonia-perdeu-12-de-florestas) from MapBiomas, a collaborative network run by NGOs, universities and tech startups, the Amazon contains 17.2 hectares of grasslands and savanna-type natural formations, equivalent to about 70% of the area of São Paulo state. They are located mainly in Mato Grosso, Pará and Roraima.

“MapBiomas classifies the PNCA as a natural grassland formation, since the term savanna is broader and includes areas technically classified as campo limpo [grassland savanna], campo sujo [grassland with scattered shrubs], campo cerrado [low tree and shrub savanna], cerrado sensu stricto [relatively dense woodland and shrubland] and cerradão [dense closed-canopy forest]. Given the scant knowledge we have of the plant species that live in these areas and the effects of fire on them, studies like ours contribute significantly to biodiversity conservation efforts in the Amazonian savanna,” Borini said.

According to Fidelis, remote sensing has played a major role in detecting and measuring the size of the burned areas and the number of fires in the region. This information combined with data obtained in the field from controlled fire experiments will help decision makers design integrated fire management plans in future.

“Remote sensing has recently proved to be a powerful tool for the detection of large-scale patterns and processes, which can’t be done on the ground,” she said. “The article is important precisely because it details the results of the combined use of remote sensing and controlled experiments in a region with a type of vegetation that hasn’t been studied in depth before: the Campos Amazônicos, an enclave of tropical savanna in the middle of the Amazon Rainforest.”

Experimental fires

The experimental part of the project involved the monitoring of 30 one-hectare plots, 12 of which were burned at the start of the dry season in May 2019 and 12 in the middle of the dry season in August 2019. Six others were used as controls and not burned. 

According to Borini, the fire behavior data obtained in the field proved that burnings conducted in August (in the middle of the dry season) were 3.5 times more intense than those conducted in May (early in the dry season). The differences in fuel load consumption detected on the ground were confirmed with higher precision by remote sensing, using images from the European Space Agency’s Sentinel 2A and 2B and NASA’s Landsat 7 and 8.

“Our analysis of the variations in satellite sensitivity during a post-burning period of 60 days showed that modeling of the effects of fires depends on the existence of records for the first 35 days after each fire,” he said. “The spectral signals in the images are significantly affected by the onset of regeneration, potentially leading to underestimation of the effects of fire on the vegetation. Unfortunately, it isn’t always possible to obtain images close to the date of a fire owing to cloud cover while the satellite is overhead.”

The results met one of the project’s main aims, which was to demonstrate the extent to which satellite data and the data collected in the field experiments matched up. According to Borini, the algorithms need adjusting because the herbaceous stratum in savanna ecosystems tends to regenerate quickly in the first few weeks after a fire and this makes correct use of satellite data particularly complex.

“The images produced by optical sensors have significant potential in applications to detect and assess areas affected by fire, but advances are required in calibrating the analytical models used to connect the data observed in the field with the spectral data from the images in order to enhance the ecological significance of the results,” Borini said.

In addition to highly detailed information about the behavior of fire (including flame temperature and intensity, wind speed and atmospheric humidity, among other factors), the researchers also collected data on the vegetation in the plots’ arboreal, shrubby and herbaceous strata before the experimental fires. These inventories were performed before and after the experiments to assess the effects on the vegetation of the different fire treatments applied. This part of the research involving the flora in the PNAC is led by the team of scientists at UNIR.

“Our interest has focused on documenting the responses of the vegetation to fire at different times of year in order to understand its resilience mechanisms,” Silveira said. “As has been found in similar experimental studies of the Cerrado, we’ve observed the arboreal stratum to be more affected by fire in the middle of the dry season than at the beginning, whereas the herbaceous stratum appears to regenerate about 80% of its biomass two years after burning, regardless of the time of year at which the fire occurs.”

Discovery of a new endemic species

The results of the research also include the discovery of a new species of sub-shrub (a “dwarf” or short shrub), which was named Mabea dalyana and has recently been described in Acta Botanica Brasilica. The first author of the article in question is Narcísio Bigio, a professor at UNIR. The plant is endemic to the PNCA and was cataloged using specimens collected from the project’s experimental plots by scientists affiliated with UNIR’s herbarium (Herbário Rondoniense).

“We were struck by the large number of fertile individuals belonging to the species that appeared in August 2019 on experimental plots that had been treated with fire three months previously. The plant has underground woody structures called xylopodia that accumulate nutrients with which to survive dry periods and re-bud after the fire has passed,” said Silveira, a co-author of the article.

Although plants in the Amazon savanna have several mechanisms that resembles those of plants in the Cerrado to enable them to regenerate after fires, the researchers are concerned that the number of uncontrolled fires set by human beings in the region is increasing. “In contrast with the Amazon Rainforest, which is destroyed by fire, savannas in the same region have many plants equipped with mechanisms that make them relatively fire-resilient. However, that certainly doesn’t mean we don’t need to worry about anthropic fires in this environment. Resilience mechanisms evolved under a natural fire regimen, with occurrence or non-occurrence of fire depending mainly on lightning,” Silveira said. 

Precisely for this reason, he added, it is important to continue documenting how the vegetation responds to changes in the fire regimen that are being implemented in these areas, not least owing to the rising frequency of large anthropic fires. 

“All the climate change scenarios projected for savannas, from best case to worst case, point to longer annual dry seasons in the decades ahead. This will have significant effects on the region’s fire regimen and increase the likelihood of major fires,” Borini said. “It will be important to continue documenting the dynamics of fire as it affects the landscape, in order to have the information needed for land-use planning and management.

The article “Impact of image acquisition lag-time on monitoring short-term postfire spectral dynamics in tropical savannas: the Campos Amazônicos Fire Experiment” is at: www.spiedigitallibrary.org/journals/journal-of-applied-remote-sensing/volume-16/issue-3/034507/Impact-of-image-acquisition-lag-time-on-monitoring-short-term/10.1117/1.JRS.16.034507.short. 

The article “Mabea dalyana (Euphorbiaceae-Hippomaneae): a new subshrub with xylopodia endemic to the savannas of the Brazilian Amazon” is at: www.scielo.br/j/abb/a/9ctVQfb8xrQxRjjkHZZsTPh/?lang=en.