Researchers of the University of Barcelona have worked on a new system to improve the predictions of fire risks related to climate causes. The study, published in the journal Nature Communications, uses –for the first time- the seasonal forecast of variables such as temperature and rain to predict the extension of the burnt area during the different seasons of the year at a global scale. Results of the study complete the existing prediction systems and could help to develop a seasonal prediction system to work globally and to develop strategies on fire management.
Marco Turco, Juan de la Cierva researcher in the Department of Applied Physics of the UB, is the first author of the study, in which the Professor Maria del Carme Llasat, from the same department, takes part too. Other experts participating in this study are Sonia Jerez, from the University of Murcia; Francisco J. Doblas Reyes, from the Barcelona Supercomputing Center (BSC); Amir AghaKouchak, from the University of California (USA), and Antonello Provenzale, from the Institute of Geosciences and Earth Resources (IGG) of the National Research Council of Italy (CNR).
The climate-fire model and seasonal forecasts
The new model is based on a standardized precipitation index which quantifies the conditions of a lack or excess of rain in a certain place for a specific time gap. The study combines empirical models that link the burnt area with previous data on precipitation, and seasonal climate forecast are added to this information. “These seasonal predictions do not tell the specific day of a higher temperature over a specific range or a significant precipitation but they can predict climatological anomalies”, says Marco Turco, researcher in the Group of Analysis of Adverse Meteorological Situations (GAMA) of the UB.
In particular, the model is mostly based on merging the observed climate evolution over the previous months to the fire season with seasonal forecast. “Combining observations with climate forecasts is a special feature of our system which contributes to increase the predictability of fires, benefiting from the information as much as possible. First we develop an empiric model to quantify the burnt area depending on whether the summer, for instance, is likely to be dry or not, and then, we use the seasonal predictions to determine the predicted drought risk and we add that information to our model”, says the researcher.
According to the study, with this information, the ability to predict seasonal fires is significant regarding a great part of the planet. “Areas where significant correlations include extratropical areas, such as Mediterranean Europe and the areas in the center and north of Asia, whose seasonal climate forecast models are more limited in terms of prediction. The prediction skill is generally higher in tropical areas, mostly due the influence of worldwide phenomena like El Niño. In other areas, our model takes the information of observations to increase the predictability of fire risks”, says Marco Turco.
Transnational system to predict fires
Researchers note that studies that assess the capacity of seasonal climate predictions to predict fires are still a few and limited to an only season or region. However, “one of the keys of this model is its prediction capacity at a global scale and for all seasons of the year”, notes Marco Turco.
In many occasions, the development and evolution of big fires depend on phenomena like droughts and heatwaves that affect large areas, further from national borders. “The study provides the necessary scientific base to develop a global system on seasonal prediction of fires that allows a transnational management of risks. However, developing such a prototype is still a challenge due the lack of quality data in areas that are less historically monitored, such as Africa and South America”, concludes the researcher.
Turco, M.; Jerez, S.; Doblas-Reyes, F. J.; AghaKouchak, A.; Carmen Llasat, M.; Provenzale, A. “Skilful forecasting of global fire activity using seasonal climate predictions”. Nature Communications, 2018. Doi: 10.1038/s41467-018-05250-0