Detection of spread fire at the mesoregion of Paraíba backwoods

The spread fire is an early agricultural technique widely used in Brazil; such technique causes serious harm to environment as well as damage to human health. Knowledge of spread fire cause and their localization is important for taking preventive measures. In Brazil, the National Institute of Spatial Research (INPE) manages the spread fire by detecting heat spots. The objective of this research was to identify the main heat spots over the mesoregion of Paraíba backwoods’ at the year of 2014. It was used heat spots’ data captured by the MODIS sensor, which is onboard of the AQUA_M-T satellite. It was registered heat spots in 56 of 83 cities inside the backwoods’ mesoregion, summing up to 304 heat spots. The occurrence of spread fire huddled mainly between July and December, months of drought in the studied area, therefore, it was concluded that the phenomenon is a product of the climate features which act simultaneously together with human deed related mainly to agriculture, where the fire is used by farmers as a tool for handling animal husbandry.


Introduction
Spread fires consist of a widely used agricultural equipment in Brazil, according to Miranda et al. (2006) and used both in primitive ways by farmers as well as in production systems highly intensified as the sugarcane.According to Vasconcelos et al. (2005), this practice causes various damage to the environment as emissions of greenhouse gases, reduction of biodiversity and the depletion and soil erosion.
Brazil is the leader among the countries of South America spread fires, and this phenomenon is concentrated in the Midwest, North and Northeast respectively (INPE, 2015).According to the Brazilian Agricultural Research Corporation (EMBRAPA, 2010) human actions are the main causes of these fires.
In northeastern Brazil, fires arise with greater intensity during the dry season, especially in years that is the El Niño phenomenon.In Africa, the effects of this phenomenon anticipate agricultural spread fire, extending the traditional period, increasing the occurrence of fire outbreaks throughout the year (Brumatti, 2012).
Knowing what the causes are and where fires occur with greater intensity is important to create plans to prevent and fight the fire.For Fonseca and Ribeiro (2003) preventive activities are most effective when based on information such as which areas catch fire more often and which favored its happening in these areas.
Through this perspective, Santos et al. (2011) state that geotechnology emerge as an important resource for identifying spread fires allowing locate, quantify and make studies of areas of spatiotemporal analysis where there are fires.An important tool for this identification is the Remote Sensing allowing cross georeferenced information and know the amount and location of heat sources (Tomzhinski et al., 2011).
Since 1980 in Brazil, the National Institute for Space Research (INPE) has been perfecting a system using satellite-sensing images for the detection of fires.They are called hot spots which according Gontijo et al. (2011) are geographical points obtained by space borne sensors when detected a temperature above 47 °C in a minimum area of 900 m².
Thus, the objective of this research was to identify the main hot spots in the middle region of Paraiba backlands in 2014, aiming to understand the factors that determine the causes and intensity of this phenomenon in the area.

Study Area
The middle region of the Hinterland has an area of 22,720 square kilometers, is located in the state of Paraiba between parallels 6 ° and 8 ° S and between the meridians 37 ° and 38 ° W (Figure 1).In 2015 presents 83 municipalities, all inserted in the area considered semi-arid.According to Köppen the climate is Bsh (hot semi-arid) (Rodriguez, 2012;IBGE, 2015).

Choose the statistical method for the selection of municipalities
For the selection of municipalities and identification of critical areas in terms of spread fire, a test was performed between two statistical measures.The arithmetic mean is which was added to the data and divided by the total number of observations.The median observation that divides the data set in half when the data are arranged in ascending order.The best fitted was used as a criterion.
To calculate the arithmetic mean was used to equation 1: (1) Where: Σ is a symbol indicating the sum in mathematics, n is the number of elements of the series or sample, i = 1 indicates that the sum begins with the first element of the series, x_i are the values of each element of serious and X indicates the value of simple arithmetic average.Equation 2: used for median: (2) Where: Li is the lower limit of the class containing the median; n is the number of set of data elements, Fa is the sum of the frequencies of the above classes to containing the median, fmd is the frequency of the class containing the median and hmd is the amplitude of the class containing the median.
After the tests, the median was chosen as selection criteria.It observed that the arithmetic mean failed to draw the correct profile of the analyzed group, due to the data set to present outliers, that is, very low and very high numbers, making the result was influenced could divert the selection of municipalities which in fact did not present such high numbers of outbreaks, thus leading to a false result.

Spotlights heat
Data on heat sources used for this research refer to the year 2014 and were obtained through spread fires Database (BDQUEIMADAS) INPE, available in shapefile format free access on the internet.These data are generated from various methodologies using satellite images in polar orbit of NOAA, EOS (TERRA and AQUA) and satellites in orbit geo seasonal GOES and METEOSAT (INPE, 2015).
For the development of this study were chosen data detected by the sensor Spectrum Imager Radiometer Moderate Resolution (Moderate Resolution Imaging Spectroradiometer) or MODIS who is aboard the AQUA_M-T.Goddard Space Flight Center Aeronautics developed the MODIS and National Space Administration (NASA) in 36 spectral channels operates at a wavelength ranging from 0.4 to 14,4μm and spatial resolution of 250 to 1000 m.For identifying spread fires scientific team responsible for creating the MODIS sensor, developed the MOD14 fire detection algorithm, which provides the location of hot spots based on energy emitted by spread fire at wavelengths of 4 micrometer (channel 21:22 ) and 11 micrometer (channel 31).In addition to the monitoring of fires, the MODIS sensor is also used for the measurement of properties of clouds, radiant energy flow, aerosol properties, changes in the use and coverage of land, volcanic activity, among others (Piromal et al., 2008;EMBRAPA, 2013).

Results and discussion
Orbital monitoring of fires recorded between the months of January to December 2014 hotspots in 56 of the 83 municipalities that make up the middle region of the Hinterland, totaling 304 outbreaks (Figure 2).Of the municipalities that make up the area, they had 28 number above the median outbreaks, which was 3.5 (Table 1).The high concentration focuses on the number of municipalities 33.7%, these possibly related to the characteristics of semi-arid climate.According to Moreira (2002) the rains in this area occur from December to April with prolonged drought, which can reach seven months, annual rainfall totals ranging around 800 to 1,000 mm, the relative humidity is around 70% the annual temperature range is about 10°C and the average temperature is 27°C with maximum around 34°C.Analyzing Figure 3, which brings the temporal distribution of detection of hot spots throughout the year, found that between the months of February to June was not registered the occurrence of outbreaks, which rules out the possibility caused fires by lightning because this is usually the wettest period of the middle region.
The occurrence of outbreaks concentrated in January and between July and December, months that comprise the dry season in the area, this factor associated with dry vegetation, low relative air unit and high temperatures facilitate the development and spread fires.Spread fire is a phenomenon that occurs throughout the state of Paraiba from climate characteristics that act in conjunction with human actions practiced mainly through agriculture.However, analyzing the Figure 4, we find that the Hinterland was the middle region with the highest concentration of hotspots in relation to other meso state, 69% of outbreaks, followed by Mata with 16% Agreste with 10% and last Borborema 5% of the total.

Final considerations
Spread fire is a state phenomenon, however through the orbital monitoring of hot spots was found that this phenomenon occurs with greater intensity in the middle region of the Hinterland.By analyzing the temporal distribution of outbreaks detection showed that these occur more frequently during the dry season, which shows the influence of dry weather in the emergence of this phenomenon.
Climate characteristics favor the spread fire; however, most of the fires are caused by human activities resulting from the tradition of using fire as agricultural management tool to create pastures, agricultural areas and or to facilitate harvest.The selection of critical areas in terms of spread fire and the indication of the causes that lead can be used to create preventive actions in order to reduce the use of fire.For this it suggests the creation of awareness campaigns, mainly farmers and cattle ranchers, showing them techniques of control and use of fire, schedules and more suitable for spread fire periods and alternatives to replace the use of fire.