Authors: Kátia Fernandes, Douglas Morton

 

Fires in the Amazon are the result of human activities, including the use of fire for deforestation, land clearing and agricultural maintenance. Most fires begin in non-forest areas that have already been cleared for agricultural use, but these fires can escape into neighboring forests and burn through the understory. Even these low-intensity surface fires kill most trees, since Amazon forests are not adapted to fire.

Both climate and socioeconomic factors contribute to fire activity in the Amazon each year. Dry conditions predispose the region to larger and more severe fires, and drought increases the risk that human-started fires will spread out of control. Drought impacts on fires in the Amazon region are predictable, often months in advance, based on the influence of large-scale climate modes linked to the El Niño Southern Oscillation and tropical Atlantic Sea surface temperatures on precipitation. Climate variability sets the stage for regional increases or decreases in fire activity, as rainfall deficits at the start of the dry season increase the risk that fires will spread out of control. Forecasts of climate-driven fire risk use the long time series of active fire detections from the MODIS instruments on NASA’s Terra and Aqua satellites, combined with observations and models of changing sea surface temperatures.

Fires in 2021

The assessment of two regional forecast models suggest slightly below average to slightly above average fire activity across the Amazon in 2021, based on cooler sea surface temperatures in the Pacific (La Niña) and a mildly warm tropical north Atlantic. For Peru and the western Brazilian Amazon region, Fernandes et al. (2011) uses sea surface temperature (SST) in the north tropical Atlantic (NTA) forecasts to predict July-September (JAS) fire season anomalies. The NTA has been shown to be a good indicator of the general pattern of JAS fire anomalies as shown in Fig.1. The forecast is updated (initialized) monthly from March to June. June initialization provides the most robust forecast and it currently points to a mildly warm north tropical Atlantic during JAS (see it here). This SST pattern leads to a shift northward of the Intertropical Convergence Zone, which reduces moisture inflow into the Amazon and droughts can occur as a consequence. In years of cooler NTA the opposite behavior (rainier dry season in western Amazon) tends to be observed.

Fig. 1. July-September (JAS) North Tropical Atlantic (NTA) sea surface temperature index (previous years’ forecast) and observed standardized fire anomalies over a domain that includes the state of Acre in Brazil and the department of Ucayali in Peru. The blue bar corresponds to the current NTA forecast for the July-September 2021 season released in June.

Fig. 1. July-September (JAS) North Tropical Atlantic (NTA) sea surface temperature index (previous years’ forecast) and observed standardized fire anomalies over a domain that includes the state of Acre in Brazil and the department of Ucayali in Peru. The blue bar corresponds to the current NTA forecast for theJuly-September 2021 season released in June.

 

The NTA index is then used to predict JAS fire anomalies in western Amazon. JAS sea surface temperature forecast in the Atlantic (initialized in June) points to an average to slightly above-average JAS fire season predicted for western Amazon (shown as light pink shades in Fig. 2). Gaps in the map represent areas with no fires or with poor relationship to the NTA index.

Figure 2. July-September 2021 fire season forecast (June initialization).

Figure 2. July-September 2021 fire season forecast (June initialization). Source: https://firecast.cast.uark.edu/ 

 

A second statistical fire forecast model (Chen et al., 2011) covers the entire southern Amazon regions of Peru, Bolivia, and Brazil. This forecast relies on SST observations from the tropical Pacific and Atlantic oceans, in combination with 20 years of satellite fire detections from the MODIS instrument on NASA’s Terra satellite.  The 2021 Fire Season Severity Forecast for different southern Amazon regions shows similar results to the model by Fernandes and colleagues, pointing to below-average fire activity for most regions except Santa Cruz, Bolivia (Fig. 3). A new interactive website hosted by the Amazonia hub of SERVIR, a NASA-USAID effort to increase the use of satellite observations for regional awareness and decision making, provides further details about the model and historic performance.

Fig. 3. Forecast of the 2021 Fire Season Severity for regions in the southern Amazon

 

Even though the two models use different methodologies, both forecasts converge to an average climate-driven fire risk in 2021. It’s important to keep in mind that the climate determines only part of the variability of the fire season. In the absence of human activities, fires would be rare in the Amazon, even in dry years. Nonetheless, seasonal forecasts of fire risk can provide important situational awareness for how climate conditions may favor (droughts) or inhibit (rainy) widespread fire activity.  Using real-time fire detections from NASA and NOAA satellites, the SERVIR team and scientists in the Amazon region will continue to track new fire ignitions and fire spread throughout the fire season (http://globalfiredata.org/pages/amazon-dashboard/).

Fires in 2020

The year 2020 witnessed widespread fires in sectors of the Amazon, especially in areas of severe drought during JAS. Our 2020 seasonal fire forecast pointed to an active fire season in its western sector although the most severe fire events occurred in the southern fringes of the Amazon, especially in the Pantanal and Cerrado biomes (Fig. 4a).

July-September 2020 (a) standardized fire anomalies and (b) the Standardized Precipitation Index (SPI).

Figure 4. July-September 2020 (a) standardized fire anomalies and (b) the Standardized Precipitation Index (SPI)

A warm north tropical Atlantic forecasted in March 2020 was an early indicator of the July-September droughts shown in Fig. 4b as Standardized Precipitation Index (SPI). The areas of drought closely match the areas of anomalously active fire occurrence (Fig. 4a) such as southern Mato Grosso and parts of Amazonas and Pará (areas marked by #1 in Fig. 4b). In contrast, areas where precipitation was near or above normal had lower than average fire conflagrations as seen around the border of Bolivia and the Brazilian state of Rondônia (area marked by #2 in Fig. 4b).

SERVIR-Amazonia is working at various fronts to improve the robustness of seasonal fire forecasts as well as visualization tools that will allow interactive mapping. In addition, meetings are being planned to inform stakeholders and potentially allow for early preventive measures in the region.

Climate conditions leading up to the dry seasons of 2020 and 2021

 Forecasts for average fire risk in 2021 describes a very different outlook than that of JAS 2020, when the models pointed to droughts and an active fire season.  Another interesting aspect that distinguishes these two years concerns the climatic condition leading up to the dry season. In 2020, the month of May registered mostly near normal to above normal precipitation which likely dampened the fire season severity in some sectors (Fig. 5). In contrast, the month of May 2021 has seen unusually dry conditions in the western Amazon, especially in the Department of San Martín in Peru, the Bolivian Amazon, and the Pantanal biome, which could accelerate vegetation water stress and create favorable conditions for fires early in the season.

Standardized precipitation index for May 2020 and May 2021

Fig. 5. Standardized precipitation index for May 2020 and May 2021

Authors

 

Kátia FernandesKátia Fernandes, Climate Scientist (University of Arkansas)

 

 

 

 

 

Doug Morton, Physical Scientist (NASA’s Goddard Space Flight Center),
Adjunct Assistant Professor (University of Maryland, College Park)