An alert system for Seasonal Fire probability forecast for South American Protected Areas. Issue 1 (20th October 2021)

• Record Type: Journal Article
• Authors: Anderson, Liana O.; Burton, Chantelle; dos Reis, João B. C.; Pessôa, Ana Carolina M.; Bett, Philip; Carvalho, Nathália S.; Junior, Celso H. L. Silva; Williams, Karina; Selaya, Galia; Armenteras, Dolors; Bilbao, Bibiana A.; Xaud, Haron A. M.; Rivera‐Lombardi, Roberto; Ferreira, Joice; Aragão, Luiz E. O. C.; Jones, Chris D.; Wiltshire, Andrew J.
• Other Names: Brown Foster guestEditor.; Peters Wouter guestEditor.; Koren Gerbrand guestEditor.; Koven Charlie guestEditor.
• Is Part Of: Climate resilience and sustainability. Volume 1:Issue 1(2022)
• Page Start: n/a
• Page End: n/a
• Publication Date: 2021-10-20
• Subjects: conservation -- disaster risk reduction -- mitigation -- wildfires; Climatic changes -- Periodicals; Sustainability -- Periodicals; Climatic changes; Sustainability; Climate change mitigation$2fast$0(OCoLC)fst01749583; Periodicals; 338.927
• DOI: 10.1002/cli2.19
• Languages: English
• ISSNs: 2692-4587

Abstract

Abstract: Timely spatially explicit warning of areas with high fire occurrence probability is an important component of strategic plans to prevent and monitor fires within South American (SA) Protected Areas (PAs). In this study, we present a five‐level alert system, which combines both climatological and anthropogenic factors, the two main drivers of fires in SA. The alert levels are: High Alert, Alert, Attention, Observation and Low Probability. The trend in the number of active fires over the past three years and the accumulated number of active fires over the same period were used as indicators of intensification of human use of fire in that region, possibly associated with ongoing land use/land cover change (LULCC). An ensemble of temperature and precipitation gridded output from the GloSea5 Seasonal Forecast System was used to indicate an enhanced probability of hot and dry weather conditions that combined with LULCC favour fire occurrences. Alerts from this system were first issued in August 2020, for the period ranging from August to October (ASO) 2020. Overall, 50% of all fires observed during the ASO 2017–2019 period and 40% of the ASO 2020 fires occurred in only 29 PAs were all categorized in the top two alert levels. In categories mapped as High Alert level, 34% of the PAs experienced an increase in fires compared with the 2017–2019 reference period, and 81% of the High Alert false alarm registered fire occurrence above the median. Initial feedback from stakeholders indicates that these alerts were used to inform resource management in some PAs. We expect that these forecasts can provide continuous information aiming at changing societal perceptions of fire use and consequently subsidize strategic planning and mitigatory actions, focusing on timely responses to a disaster risk management strategy. Further research must focus on the model improvement and knowledge translation to stakeholders. Abstract : We considered five variables that represent the anthropogenic and climatological conditions that increase the probability of fire occurrence. The anthropogenic components of the forecast are the (i) trend in the number of active fires over the past three years and (ii) accumulated number of active fires over the same period. The number of active fires and their trend can be measured using the "hot pixels" in satellite‐based data sets, described below. A positive trend in the number of hot pixels can be interpreted as a recent increase in the use of fire in that region, which can be associated with ongoing land use/land cover change. The accumulated number of hot pixels was used to identify the areas with high fire occurrence in the period analysed. The climatological conditions that we selected for the forecast are: i) the length and duration of the dry season for the period studied, and probabilistic forecasts of (ii) below‐average rainfall and (iii) above‐average temperature. These latter two variables were prioritized due to an adequate skill on the probabilistic forecasts. This product is currently operational, updated at a monthly scale.