News

How do climate and oceans predict forest fires?

26.04.2015

Forest fires including in mountainous areas are a global problem that affect ecosystem services as much as the economy and human health.. The National Commission of Forest Fires (Comisión Nacional de Incendios Forestales – CONIFOR) and the National Meteorological Institute (Instituto Meteorológico Nacional – IMN) of Costa Rica were given the task of investigating the relationship among some atmospheric and oceanic variables, forest fires, quantity of burnt hectares and hot spots.

Materials and methods

The climatic and oceanic variables used were: a) the anomalies in the surface temperatures of the equatorial Pacific Ocean in the regions ofNiño1.2, Niño3 and Niño 3.4; b) the pattern of rainfall in the semester corresponding to fire season and the sixmonths preceding it; c) land temperatures at surface level and their relationship with other factors, namely: i) hot spots, ii) the quantity of forest fires and iii) the number of burnt hectares.

An analysis was carried out at national level as well as in the province of Guanacaste (in the northern Pacific part of the country).

Considered climate forcing, the values of the anomalies of the sea’s surface temperature between January and June for the period 1951-2014 were analysed, and they were statistically correlated with the quantity of burnt hectares in each of the conservation areas of the nation.

The difference of the air temperature at an atmospheric level of 925 hectopascals was calculated in the four years that recorded more than 40000 burnt hectares and the four that registered less than 10 000 in Costa Rica, too.

To establish the dominant climatic scenarios, from very dry to very wet , the variability of the semester rainfall scenarios in Guanacaste was determined – particularly in the years that were most or least active compared to the quantity of burnt hectares – by calculating the Standardized Precipitation Evapotranspiration Index (SPEI). To do so, the software used was that available on the “SPEI Global Drought Monitor” portal.

Fire seasons of greater or lesser intensity in Guanacaste were classified according to the phases of the El Niño-Southern Oscillation (ENOS) and the corresponding SPEI six-month period to determine any existing relation between the two.

Finally, at national level, in order to analyse whether a relationship may exist between climatic variable and fires, the relationship between the quantity of hot spots, the number of forest fires, the six-month period anomalies of the sea’s surface temperature in the regions of Niño1.2 and Niño3 and the quantity of burnt hectares were combined.

Results

With the aim of understanding whether a greater amount of burnt area corresponds to a greater number of fires, the statistical relationship between annual quantities of burnt hectares and the number of forest fires was analysed for 2000-2013 period. The Pearson Correlation coefficient (cP) between both time periods was 0.89, meaning that the greater the quantity of fires, the greater the number of burnt hectares.

The hot spots of the period 2004-2013 were determined using the data of the National Commission for the Understanding and Use of Biodiversity (Comisión Nacional para el Conocimiento y Uso de la Biodiversidad – CONABIO) aiming at understanding whether there was a statistical relationship with forest fires in the country. The Pearson coefficient between both sets is 0.78; the high values of the coefficient indicate that the more hot spots should present a greater quantity of forest fires.

The significant increase of both variables in the years 2012 and 2013 might be due to the extremely low rainfall that occurred in that period, see table 1 (download full paper).

The annual quantity of hectares burnt by fires in Guanacaste from January to June is shown in Figure 6 on page 97 of the paper. The years 2013, 1998, 2001 and 2002 show values above 40 000 ha and the years 1999, 2006, 2009 and 2011, below 10 000 ha.

Both the fire season of the first semester of the year and the semester prior, which is the rainy season, show a varying pattern over the years. The four fire seasons in the study area with less burnt hectares were associated with the phenomenon La Niña and were preceded (year -1) by semesters with heavier rain than usual, as is shown in the positive semester values of SPEI for each of these years; despite having dry scenarios in the year of the analysed period (year 0).

The four years with the greater quantity of burnt hectares were related with the different phases of the ENOS, but were always preceded by dry or very dry semesters such as in 1998 and 2013, independently of the phase of the dominant ENOS.

Conclusions

Based on the results obtained, the atmospheric and oceanic variables analysed can be used as climatic indicators or predictors for the implementation of an index of intensity of the fire season.