The Use of Low Spatial Resolution Remote Sensing for Fire Monitoring in Nicaragua: a survey of three successive burnings seasons.


The Use of Low Spatial Resolution Remote Sensing for Fire Monitoring in Nicaragua: A Survey of Three Successive Burnings Seasons


A. Jacques de Dixmude*, P. Navarro*, S. Flasse*, I. Downey*, C. Sear*, J. Williams*, P. Ceccato*, R. Alvarez**, F. Uriarte**, Z. Zúñiga**, A. Ramos**, I. Humphrey**

* Natural Resources Institute, University of Greenwich, Central Avenue, Chatham Maritime, Kent, ME4 4TB, UK
** Ministerio del Ambiente y los Recursos Naturales, Km 12½ Carretera Norte, Managua, Nicaragua


The Nicaragua Land Resources (Fire) Monitoring Project is an environmental monitoring project carried out by the Ministerio del Ambiente y los Recursos Naturales (MARENA) and the Natural Resources Institute (NRI). The Project was supported by the UK Government Department For International Development (DFID) until June 1998. A PC-based NOAA satellite receiver, installed at MARENA headquarters in Managua since its start in June 1995, has enabled daily observations to be made on vegetation fires in Nicaragua and Central America. These observations are used particularly to assist and support operational forest management activities in Nicaragua. The technical approach and project findings are presented with reference to forest fire monitoring in Nicaragua during the last three dry seasons (January-May, 1996 to 1998). A geographical information system (GIS) is used in order to attempt to identify fire activity patterns. Spatial and time distributions of hot spots are examined and assessed with regard to forest/land use type coverage and to population density and rural poverty levels. Some features of fire activity in Nicaragua are highlighted by considering some local cases in particular.


El Proyecto de Monitoreo de Recursos Terrestres (y Fuegos) en Nicaragua es un proyecto de monitoreo ambiental que fue ejecutado por el Ministerio del Ambiente y los Recursos Naturales (MARENA) y el Natural Resources Institute (NRI). El Proyecto fue auspiciado por el Department For International Development del Gobierno del Reino Unido (DFID) hasta Junio del 1998. Una estación receptora de datos de los satélites NOAA, funcionando en base a computadoras personales, fue instalada en la sede central del MARENA en Managua en Junio del 1995. Esto ha permitido hasta la fecha, que se realizaran observaciones diarias de los fuegos de vegetación en Nicaragua y Centroamérica. Estas observaciones son utilizadas particularmente para asistir y respaldar actividades de manejo de bosques en el país. A continuación se presenta los resultados conseguidos por el proyecto luego de las últimas tres temporadas de quemas en Nicaragua (Enero a Mayo, 1996 a 1998). Un sistema de información geográfica (SIG) es utilizado para tratar de identificar tendencias en la actividad pirómana. Las distribuciones espaciales y temporales de los puntos de calor son examinadas y evaluadas con respecto al tipo de ocupación del suelo, la densidad de población y el nivel de pobreza rural. Al considerar algunas situaciones locales en particular, se puede resaltar algunas características de la actividad de fuego.


Le Projet de surveillance des ressources terrestres et des feux au Nicaragua est un project de monitoring environnemental exécuté par le Ministerio del Ambiente y los Recursos Naturales (MARENA) et le Natural Resources Institute (NRI). Ce projet fut financé par le Department For International Development du Gouvernement du Royaume-Uni (DFID) jusque juin 1998. Depuis son commencement en juin 1995, une station réceptrice de données satellitaires NOAA, installée au siège principal de MARENA à Managua, permet de faire quotidiennement des observations sur les feux de végétation au Nicaragua et en Amérique Centrale. Ces informations sont utiles particulièrement pour assister et renforcer les opérations de gestion des forêts au Nicaragua. Outre une approche technique, les résultats obtenus après les trois dernières saisons sèches (de Janvier à Mai, pour 1996, 1997 et 1998) sont présentés ici. Un système d’information géographique (SIG) est utilisé pour tenter d’identifier des tendances du phénomène ‘feux’. Les distributions spatiales et temporelles des ‘points de chaleur’ sont examinés et évalués par rapport au type d’occupation de la terre, à la densité de population et au taux de pauvreté rurale. En prenant à part certaines situations locales, on peut aussi faire ressortir certaines caractéristiques de l’activité incendiaire.


1. Fire concern in Nicaragua

Fire is a useful and efficient tool for environmental management such as forest clearance, field preparation, regrowth for livestock, and reduction of fire hazard. But its misuse can have adverse consequences (e.g. Bond and Van Wilgen, 1996). Areas of concern include natural resource sustainability, biodiversity conservation, destruction of commercial wood and threats to village welfare. As a result, there is increasing pressure from international, national, regional and local communities to manage vegetation fires. However, developing countries are often primarily challenged by a lack of means to monitor fire on a national scale.

There has been a very long term trend of conversion of forest to agricultural land through the use of fire in Nicaragua. However, Nicaragua still retains a significant and valuable forest resource, including the most extensive natural broadleaf forest in Central America. In common with its neighbours, Nicaragua also contains significant coniferous forest, an environment that is particularly prone to forest fire. Available estimates of deforestation (FAO 1993) suggest that forest loss is occurring at around 2.3% p.a. – equivalent to around 125,000 ha per year.

The pace of forest cover loss has undoubtedly accelerated in recent years. High unemployment (estimated at 60%), resettlement of rural populations, rural landlessness and poverty have all contributed. This makes Nicaragua increasingly more vulnerable to natural hazards. Deforestation played a relevant role in the huge damages due to flooding and mudslides caused by Hurricane Mitch, which violently struck the north-western part of Nicaragua in end October 1998.

Fire is a major risk to forest resources and is itself a cause of deforestation. This is because forest ravaged by fire is left open to agricultural conversion. Many prescribed ‘agricultural’ fires escape out of control and burn into surrounding forest because farmers and ranchers do not take proper precautions, such as construction of effective fire breaks (Alvarez and Travisany, 1993). Other human causes of wildfire include burning to drive game, careless practices such as tossing cigarette ends into dry vegetation, abandoned campfires or intentional burning by arsonists (Ciesla, 1997).

Fire occurrence is therefore a prime indicator of human-induced land use change, particularly land conversion from forest to agriculture. Fire maps can be used to define ‘agricultural fronts invading forested areas’. The forest is also suffering from the territorial ambitions of large-scale cattle ranchers. It is believed that in order to facilitate forest clearance and so maintain a plentiful supply of cheap land, ranchers supply ‘campesinos’ with chain-saws to facilitate clearing of land for agriculture which is then ‘sold’ to the graziers. Burning the cut vegetation is an integral part of land clearance (Downey, 1997).

In the coniferous forest, although fire is a ‘natural’ feature, its incidence has increased remarkably with human encroachment. But broadleaf forest is paradoxically most exposed to the dangers of fire, particularly at its margins or where it has already been divided into a mosaic of remnants (Quan, 1995; Ciesla, 1997).

It follows that monitoring of forest conditions and fire activity in particular is essential for the sound management of Nicaragua’s important areas of forest resource and for the rational allocation of limited resources to meet fire threats and outbreaks. Until recently, the available information on the occurrence, extent and impact of fires (and changes to the national forest estate generally) was limited in quality, quantity and timeliness. This presented a major handicap to forest management. Effective fire detection reporting has been carried out in only a few areas. For example, three local projects, two along the Northwestern volcanic cordillera and one in the Northeastern Caribbean pine plain, use a network of observation towers, and have fairly well organized fire detection and reporting procedures, complementary to fighting teams operating on field (Ciesla, 1997). But these are spatially very limited, and suffer from frequent logistical restrictions, which prevent them from working properly.

While comprehensive analysis of fire data can improve understanding of fire activity and enable better management decisions to be made, such an approach may be a real challenge for a budget constrained government. Ahead of the need to enhance political and environmental awareness, assistance is required to improve management of Nicaraguan natural resources and environment. Decrepit observation networks, poor communications, underpaid and demoralised staff and urban-orientated political systems do not improve prospects. Under such circumstances, systematic and reliable data on fires over large remote areas are most unlikely to be collected, processed, assimilated or incorporated into the decision-making process. As a consequence, the basic information required to analyse the situation (for example, time and location of fires) is often the primary challenge: there is a need to find sustainable and reliable ways of documenting events to allow scarce fire management resources to be used more efficiently (Flasse et al., 1998).


2. Monitoring Fire Activity through Remote Sensing

The most practical, feasible, relatively objective and cost effective means to quantify and monitor these fire events at national to regional scales is to utilise Earth observation remote sensing and GIS technologies. Many studies have demonstrated the potential usefulness of remote sensing techniques to monitor the Earth’s surface and fire related information in particular (Kaufman et al., 1990; Prevedel, 1995). Clearly, the major benefit of remote sensing is that it permits the observation of large areas of territory, on a regular basis.

Originally conceived for meteorological applications, NOAA series satellites have proved useful for Earth observation applications and especially for fire monitoring (e.g. Malingreau, 1990). Equipped with the Advanced Very High Resolution Radiometer (AVHRR), the NOAA satellite series allow monitoring of active fires at different scales compatible with the request for fire management (from local to national and global scales). The use of NOAA-AVHRR to monitor fires has been developed in recent years with the development of techniques to detect active fires automatically. Measurements provided by the thermal infrared channels of the AVHRR sensor are used to detect fire in vegetation through the effect of combustion on radiative temperature (Malingreau, 1990).

NRI has recently developed a contextual fire detection algorithm (Flasse and Ceccato, 1996) to improve the reliability of automated fire detection. Compared to previous techniques, where threshold values are set to detect hot spots and need constant updates according to time and spatial conditions, the contextual algorithm allows the automatic and reliable detection of fires with minimum input from the user (Eva and Flasse, 1996). The algorithm developed at NRI now forms the basis of operational fire detection projects in Nicaragua as well as in other countries in Africa, Southeast Asia, and global fire monitoring projects such as IGBP-DIS-Global Fire Product.


3. NOAA Satellite data reception in Nicaragua

The UK Department For International Development (DFID) supported a joint environmental monitoring project between the Nicaraguan Ministerio del Ambiente y los Recursos Naturales (MARENA) and the Natural Resources Institute (NRI). MARENA is the government agency charged with responsibility for a sustainable management of renewable natural resources in forestry, and in other areas of environmental damage.

The overall purpose of the Nicaragua Land Resources (Fire) Monitoring Project was to encourage more integrated and sustainable environment monitoring methods in Nicaragua towards improved management of natural resources, particularly forests. The particular objective was to assist with appropriate and cost effective forest fire and environmental monitoring through utilisation of real-time local reception of data from environmental satellites and to evaluate their relevance and sustainability in the context of Nicaraguan institutions. The project ran from June 1995 to June 1998.

The project installed a PC-based NOAA satellite receiving ground station at MARENA headquarters in Managua to enable daily observation of fires and other environmental phenomena in Nicaragua. These are used to assist forest protection, fire control and natural resource management activities in Nicaragua. Information products are generated on a routine basis and supplied to a number of Nicaraguan institutions at local, provincial and national level. Regional level information can also be generated and this is attracting a wider base of end users. This also was encouraged by two participatory workshops organised by the project, which were aimed at raising awareness and adapting the service (the information supply) to the particular needs of local users.

During the project implementation, two NOAA satellites provided reliable data (NOAA-12 and NOAA-14). This gave a nominal overpass frequency (time resolution) of four data captures every 24 hours. In practice, the project essentially focused captures on NOAA-14 afternoon passes, since this is the most representative time for monitoring fire activity in Nicaragua.

The contextual algorithm used by the project addressed a problem specific to the Western Coast region of Nicaragua. During day time and when the sky is clear, the darkness of the volcanic soil and the fact that it is mostly bare during the dry season, result in temperature reaching such levels that image pixels covering those areas are saturated just as if there were fires present. Simple threshold methods would lead to an overestimation of the number of hot points. By taking the ‘context’ of each hot pixel into account, the contextual algorithm reliably avoids recording these hot soil areas as ‘false’ fires.

The information on fire activity is delivered to departmental authorities in charge of fire control in the form of lists of co-ordinates with references to the 1:50,000 topographic map index and to observation towers where appropriate. At a local scale, this provides an early-warning tool for fire fighting.

At a national scale, the data are analysed in their context (land cover, forest type, administrative divisions) by using GIS, to provide thematic information which can be used for example to locate possible deforestation fronts, helping to raise political awareness, or to direct extension programmes to promote adequate strategies, such as alternative land use. This paper is aimed at sharing the most recent findings obtained after monitoring three complete fire seasons: 1996, 1997 and 1998.


4. Results and Discussion

General issues

In Nicaragua, most fire activity occurs during the dry season, which stretches from the end of December up to the end of May. Three successive seasons were monitored by the project and the ‘first sight’ results were widely promoted and distributed, either as simple lists, as tables, or as national or regional fire maps. This greatly strengthened the public awareness about fire concern; as, before then, there was no alternative way to gather this information so quickly and so widely.

For the longer term, such a system really becomes useful when it is integrated with other institutional concerns and enables the forest resource managers to improve their understanding of fire issues. Typically, this is an iterative learning process for fire management officers using remote sensing and GIS tools. Through repeated use across several fire seasons, technicians and managers become familiar with the data and able to analyse statistics in more detail, such as:

  • the distribution of fires within the political territories (departments or municipalities)
  • the proportion of fires within the forest
  • the correlation between fire occurrence and population density
  • the relationship between fire occurrence and social issues (for example poverty)

These analyses are aimed at identifying and classifying the detected hot spots with regard to that ancillary information. Thus, every fire, according to the type of land use where it occurs, the moment it is set, the vulnerability of the particular ecosystem, the potential risk for human settlements and the potential impact on soil erosion, could be crudely qualified as a ‘good’ or ‘bad’ fire. Once this ‘judgement’ can be made with an acceptable reliability, a step forward will have been made towards the goal of using this technique in real decision-making support. To date, this is still constrained by a limited availability of good quality thematic data. This present study is thus only a first attempt at putting fire activity back within its context in Nicaragua.

In every report and in every analysis, it has to be assumed that those satellite-derived data are underestimated with regard to the actual number of fires, because of some limitations:

  • only the fires active at the time of satellite pass are detected
  • a certain amount of information is lost as it may technically not be possible to capture a valid image every single day due to orbit variations
  • cloud cover prevents satellite observation of fires underneath, which also decreases the amount of valid information available with respect to the situation on the ground
  • the contextual algorithm always acts as a conservative system; i.e., when, after the successive tests within the algorithm, a hot pixel doesn’t appear clearly enough as a fire, this is not retained; for example, when there can be confusion with hot soils
  • the limited spatial resolution of the AVHRR sensor (1.1 km pixel at nadir) enables us to detect fire fronts as small as 50 x 100 metres (Belward et al., 1993). This means that smaller fires may not be detected, or that several fires may fall within an area covered by the same pixel (the elementary unit of an image), resulting in only one registered hot spot. Conversely, a hot pixel can also be one of several covering a large fire front

Consequently, all the results we present are expressed in terms of numbers of hot spots (or hot pixels) selected by the contextual algorithm. These are the closest approximation that can be made to reality. It has been seen however that the trends and the relative amplitudes of fire occurrence remain the same if compared with older data retrieved from national statistics (Ciesla, 1997).

Table 1 presents a synthesis of the number of fires (i.e. hot pixels) that were detected in Nicaragua over the three fire seasons monitored (between 1996 and 1998). This is in order to give a general idea of the magnitude of the phenomenon.


 Table 1 : Number of detected hot pixels on Nicaraguan territory over three dry seasons







Total Season























Fig. 1 illustrates the typical West-to-East movement of fire activity in Nicaragua. The earliest fires (in January and even December some years) are generally observed only in the pacific region (the driest, most densely populated and most farmed region of the country). Two months later, the central mountain region begins to suffer outbreaks of fire activity. Toward April, fire activity invades the rest of the country and especially increases in the Atlantic region, although this has the lowest population density and the wettest climate.


2031 Byte

January 1997


2130 Byte

February 1997


2232 Byte

March 1997


3747 Byte

April 1997

3867 Byte

May 1997

Fig. 1 : Monthly hot pixel maps for the 1997 fire season in Nicaragua. The inside boundaries represent the main geographical regions: Pacific Coast, Central Mountains, Atlantic Coast. The dotted line represents a likely ‘human settlement belt’ around one of the last large extensions of primary rainforest in central America.


Data weighting

For various reasons, it was not possible to ensure an uninterrupted coverage of the respective periods. The rate of valid data captured varied slightly over the respective months and between years. In order to compare data from different seasons, the monthly and seasonal hot pixel number figures were weighted. The numbers of hot spots were divided into the respective numbers of days of actual data capture and multiplied by the total number of days in every month or in the whole fire season, as shown in Table 2.


Table 2 : Number of days of actual captures every month (maximum number of days in brackets)







Total Season


17 (31)

18 (29)

24 (31)

22 (30)

18 (31)

99 (152)


13 (31)

13 (28)

23 (31)

19 (30)

19 (31)

87 (151)


16 (31)

17 (28)

17 (31)

25 (30)

18 (31)

93 (151)


In this correction, we assumed that fire occurrence increases linearly and that acquired data are evenly distributed within every month, which is close to the reality. The cloud coverage might also be taken into account through the same type of extrapolation. But in this study this consideration was left aside, because it is less certain whether there is also a linear relationship between fire occurrence variation and proportion of land hidden by clouds (Moula, 1996).

On this basis, all the tables and diagrams shown below are produced from the figures weighted by number of captures.


 Table 3 : Weighted numbers of detected hot pixels







Total Season*






















* The figures under ‘Total Season’ are not the sums of all the months, but are the ‘raw’ figures of Table 1 divided into the total number of captures and multiplied by the total number of days as shown in Table 2.


After three years of monitoring, a number of observations can be made, based on these data. It appears that fire activity can show a different pattern in each season. For example, it began rather early in 1996 and rather late in 1997. Unusually, in 1997, most of fire activity occurred in May. 1998 has been the most dramatic year in terms of fire occurrence by far; most of which was caused by an increased burning activity throughout the country. There is a likely cause-effect relationship with the exceptional level reached by El Niño phenomenon in late 1997 to early 1998, which is reported to have induced a very severe and long drought (CEPAL, 1998; WFP, 1998).

Analysing fire occurrence in its context

As indicated above, it is critical to analyse fires in their context. As a first approach, we compared the detected fires with available geographical information on:

  • administrative boundaries (140 municipalities, 17 departments, 3 main regions) (MARENA, 1995)
  • land use; dividing the country into 11 categories, including gross forest types, according to inventories dating from 1988 to 1992 (see Fig.2) (MARENA, 1995)
  • population and rural poverty rate in each municipality (Lacayo, 1998). (24854 Byte)

Fig. 2 : Land use and municipality map of Nicaragua (Marena 1995) For clarity the 11 original land use categories have been gathered into 4 new classes, which are:

  • Dark grey : broadleaf evergreen forests (>20m, 12 to 20m) and forest fallows
  • Middle grey :  mangroves, swamps and wet lands
  • Light grey : pine forests and tropical deciduous dry forests
  • White : crops and pastures, incl. perennial crops, other marginal categories and areas outside the study area


Fire occurrence vs. forests and regions

Tables 4 and 5 show the fire occurrence distribution according to the geographical regions, and to the gross land use categories, respectively.


 Table 4 : Numbers of hot pixels in each geographical region, over three seasons (weighted figures)


Pacific Coast

Central Mountains

Atlantic Coast

















Ancillary information

Area (km2)





% forested *










* includes the forest cover categories as explained in the text below


Table 5 : Number of hot pixels by type of land cover in Nicaragua (weighted figures)

Forest types

Area (ha)


Hot Pixel Occurrence 1996

Hot Pixel Occurrence 1997

Hot Pixel Occurrence 1998











in the country

No. /
10 km2


in the country

No. /
10 km2


in the country

No. /
10 km2

Broadleaf evergreen forest > 20m












Broadleaf evergreen forest 12-20 m












Broadleaf deciduous forest 4-12 m












Pine forest
























Crop or pasture land












Perennial crop (e.g. coffee)












Swamp and wetland












Small islands












Forest fallow












Rocky outcrop

























Absolute figures draw attention to the lowland broadleaf evergreen forest areas, where the highest number of fires occurs. If those numbers are considered in relative terms, it can be seen that the percentage of fires in those forests with regard to the total number of fires in the whole country is not significantly different from the proportion of territory they occupy, at least as far as the two first years are concerned. But the increase observed in 1998 is significant (from around 35 to 37 % to more than 44 %). Both tables 4 and 5 make clear that the drought possibly caused by El Niño phenomena (and thus the consequent increase of fire activity) mostly affected the evergreen rainforest that lies across the wide Atlantic plain.

The density of ‘fires’ (hot pixels) was also investigated, rather than absolute numbers. In this way, the size of the respective areas is taken into account, since it is obvious that, whilst the greatest fire occurrence has been registered in the Atlantic Coast region (Table 4), this one also covers the largest territory, representing more than half of the country.

Observing fire occurrence density, expressed in number of hot pixels per 10km2, reveals that forest fallows as well as swamps and inundable lands have been the two most affected zones by fire activity over the three monitored seasons. Hot pixel density in swamps and wetlands nearly doubled between 1997 and 1998 (2.49 to 4.13 hot pixels/10km2). This is likely to be a direct effect of an El Niño-related drought, making those areas more prone to be burned than usual. This observation might mean that burning activity is exerting an increasing pressure particularly on wetlands. These are essentially located along the Atlantic coast and play an evident ecological role in the balance of the large catchments which drain the eastern half of the country. Most of these aquatic or semi-aquatic ecosystems are included among the highest priority areas in terms of biodiversity value (World Bank, 1997).

As for forest fallows, these are limited to two areas surrounding the broadleaf evergreen forest in the municipalities of San Carlos and Nueva Guinea (both in south-eastern Nicaragua). One might expect them to be far more represented throughout the country or at least in the Atlantic region, as they appear to correspond to land parcels temporarily left to forest recolonization inside complex forest-crop-pasture mosaics, at the margins of larger forest areas. The high fire activity observed there implies an acceleration of the traditional migrant agriculture cycle and then an increased reconversion to crop or pasture, which dramatically increases the fragility of soils that were already considered as marginally or not suitable for agriculture (Herrera, 1995).

Fire occurrence by population and municipalities

As it is assumed that most fires happen as a result of human action (Ciesla, 1997; Collins, 1992; Downey, 1997), one might deduce that human presence is a relevant factor in fire risk assessment and pressure on land. Hence we are interested in the relationship between hot pixel density and population density.


16165 Byte

Fig.3 : Relationship between hot pixel density and population density


The histogram in Figure 3 shows the general trend of fire occurrence density according to population density ranges at a municipal level. There is actually no significant correlation between both variables, according to each year (correlation coefficient going from r = -0.37 to r = -0.24). But it is clear that the outstanding numbers of fires reached during 1998 especially affected the least inhabited municipalities of the country (probably those with the largest areas of ‘burnable’ vegetation). This may also mean that there has been an unusual acceleration of the migration movements toward those isolated zones. This fact would not be included in the available figures that date from the last national census in 1995 (Lacayo, 1998).

To confirm the hypothesis above, one would need to compare fire (hot pixel) occurrence in forested areas with population density at the same time in those areas. This latter information is not available, however. Should this suggestion be confirmed, particularly in the Atlantic coast region, the increased fire occurrence in forested areas could be taken as a determining indicator of recent human settlement in areas of tropical rainforest.

Fire occurrence by forests and poverty

It is often said that tropical countries which are characterised by a high deforestation rate (which is in significant part due to land clearance by small farmers) have to cope with a large proportion of their rural population living in poverty (Collins, 1992). As an indicator of human influence on land use changes, an attempt was made to assess whether and how far fire activity is linked to standard of living in Nicaragua. In other words, to test the hypothesis that the more rural poor there are throughout the territory, the more fires occur in forests.

At this stage of the study, figures were obtained from the Municipal Poverty Map made by the ‘Fondo de Inversión Social de Emergencia (Lacayo, 1998). Besides the total population of every municipality of Nicaragua, that study provides the number of rural people who are considered as poor, i.e. living below a ‘poverty’ income threshold fixed at US$ 428.94/year.

To consider the proportion of forested area in each municipality, several categories from the land cover map legend (Table 5) were combined under a same term ‘forest’:

  • broadleaf forest > 20 m : typically corresponds to the low land tropical rain forest, which is assumed to be still mainly primary
  • broadleaf forest 12 to 20 m: corresponds either to secondary rain forest, or to montane tropical cloud forest
  • broadleaf forest 4 to 12 m: corresponds to tropical dry deciduous forest
  • pine forest: both montane Pinus Oocarpa forest, and Atlantic coast P. Caribbea forest, much of which is managed and exploited with a commercial purpose
  • mangroves
  • swamp and wetlands (although being largely ‘open’ landscapes, these one are included for their high biodiversity and their ‘wild habitat’ character)
  • forestal fallow

Should rural poverty represent a threat to forests (for instance, through an increased pressure on land conversion by fire), it would be important to know how ‘heavy’ is this poverty pressure on land. That means expressing it by territorial density values. Does rural poverty density have any influence on fire density occurring in forests? Would it be fair to affirm that the more rural poor live in a municipality, the more fires are detected in its forested areas ? The relationship between hot pixel density (in forested area only) and rural poverty density (in the whole area) in each municipality shows us that there is no evidence (Fig.4).

Although the hypotheses presented above may go some way to explain the social causes and impacts of fire activity, at the current stage of the study it is not possible to state any significant influence of the parameters used. More precise geographical information is required on population density.

Fire activity is a very complex phenomenon and assessing some situations on a case-by-case basis is likely to result in clearer interpretations. For example, the five municipalities which had the highest observed hot pixel density for every season were selected. These were examined with respect to the other recorded features (% forest, population density, % poverty rate) corresponding to each municipality. This is shown in Table 6.

In general, features such as a high proportion of forest, a low population density and a high proportion of rural poor, would seem to be typical. But there are some peculiarities. In 1997, the most affected municipalities are two densely populated ones in the Pacific Coast region (Jinotepe and La Paz de Carazo), with a fairly low proportion of rural poor population. One of them has even no forest coverage at all. It is possible that in that zone, fire activity is related to forest clearing, but only marginally. Fig. 2 locates those municipalities near the Pacific coast of Nicaragua, and Fig. 1 shows this particular concentration of fire activity around that area.


8739 Byte

Fig. 4 : Relationship between hot pixel density in forests and rural poverty density


The case of the municipality of Nueva Guinea (south-east of Nicaragua) is one of particular interest. In average over the three monitored fire seasons, it is third in terms of hot pixel density (3.19/10 km2), although its forest coverage rate is fairly low (7.73 %). This broad municipality has undergone a strong expansion of settlement and land clearing for a long while, since it has been particularly devoted to developing livestock production. Focused field studies might confirm that fire activity there is essentially aimed at refreshing pasture for cattle. Moreover, its forest coverage was particularly struck by Hurricane Joan which devastated the south-eastern coast of Nicaragua in 1988 and created an open ‘track’ within the original broadleaf evergreen forest (Ciesla, 1997).

Siuna, which also lies on the Atlantic coast region of Nicaragua but in the north-east, is very different. It also had a high average fire density over the three seasons but still has a high forest cover proportion. As this municipality is at the junction of the two highways linking the west to the east of the country, it is subject to increasing pressure from both sides of those roads. In addition, the main traditional economic activity there used to be mining, which is not as space consuming as livestock rearing or migrant agriculture. With the recent decline of this ‘industry’, lots of small miners lost their jobs and remained without resources (Stührenberg, 1996). A more intensive settlement process then began, invading the primary forest gradually. Hot pixels maps clearly show a ‘settlement belt’ around those large forest extensions, which include Bosawas protected reserve (see Fig.1).

Both Murrá, the most fire affected municipality on average over the three seasons, and Wiwilí, which has the fifth highest hot pixel density over the same period (Table 6) are in the Central Mountains region. Although the same consideration about land clearance pressure can be made (it is actually the ‘western’ settlement front of the same large forest area), another factor might influence fire activity. That part of Nicaragua is arguably the region most affected by the ‘desarmados’ problem that appeared at the end of the civil war in 1990. Indeed, when peace was signed, several thousands of ex-soldiers from both sides were given a little land in order to facilitate their return to a civil life after several years fighting. But most of them knew nothing of agriculture and in most cases they were left alone, with fire as their only clearance tool.


Table 6 : Features of the municipalities with the five highest hot pixel desities by season.





Fire Density /
10 km2

% Forest

% Forest

Pop. Density / km2

% Rural Poor

1996 39 Murrá Central Mountains 3.60 73.31 67.45 23 80.24 50 San José de Cusmapa Central Mountains 3.48 57.91 37.04 74 66.40 88 Nueva Guinea Atlantic Coast 2.30 7.73 7.73 28 37.54 85 El Rama Atlantic Coast 2.20 45.36 43.99 13 58.41 92 San Miguelito Atlantic Coast 2.13 76.05 76.05 12 72.66 1997 139 La Paz de Carazo Pacific Coast 3.87 0.00 0.00 223 41.61 134 Jinotepe Pacific Coast 3.86 56.14 56.14 194 29.52 39 Murrá Central Mountains 3.41 73.31 67.45 23 80.24 82 Prinzapolka Atlantic Coast 2.81 92.55 69.86 1 63.91 56 Wiwilí Central Mountains 2.68 55.35 55.35 18 70.20 1998 80 Siuna Atlantic Coast 5.91 80.15 80.15 12 76.97 88 Nueva Guinea Atlantic Coast 5.38 7.73 7.73 28 37.54 94 El Castillo Atlantic Coast 5.24 96.85 96.85 6 77.76 84 Cruz del Rio Grande Atlantic Coast 5.03 75.21 74.04 2 47.25 93 San Carlos Atlantic Coast 4.73 47.04 34.98 19 54.72 Average over the 3 seasons 39 Murrá Central Mountains 3.65 73.31 67.45 23 80.24 80 Siuna Atlantic Coast 3.32 80.15 80.15 12 76.97 88 Nueva Guinea Atlantic Coast 3.19 7.73 7.73 28 37.54 84 Cruz del Rio Grande Atlantic Coast 3.14 75.21 74.04 2 47.25 56 Wiwilí Central Mountains 2.90 55.35 55.35 18 70.20


Another interesting fact is observed when comparing two neighbouring municipalities, in the extreme south-east of Nicaragua, along the San Juan River which makes the boundary with Costa Rica: El Castillo and San Juan del Norte. According to the land cover map, those territories have 97 % and 96 % covered by broadleaf evergreen primary forest respectively. But the difference between them in terms of hot pixel density is significant (see Table 7).


Table 7 : Comparison between two neighbouring municipalities



% Forests (all types)

% Broadleaf Forests

Population Density

% Rural Poor

San Juan del Norte

Atlantic Coast





El Castillo

Atlantic Coast





Hot pixel density (No. / 10 km2)

San Juan del Norte

Atlantic Coast









El Castillo

Atlantic Coast






An explanation can be found in the fact that San Juan del Norte is almost completely situated within the protected Indio-Maiz natural reserve, while El Castillo is mostly crossed by its buffer zone, which is subject to a high migrant agriculture pressure (Valerio, 1998). Population figures clearly show that in the latter, the population is generally rural and poor, thus dispersed throughout a large part of its territory. In the former, although the population is perhaps as poor, it is concentrated in the municipal capital, just on the Caribbean littoral. Along with Bosawas Reserve, Indio-Maiz Reserve, which is the core zone of the ‘Si-a-Paz’ protection area, is the last primary tropical rainforest area of relevant extension that still remains intact in Nicaragua. But despite all, that fire activity also increased notably there in 1998.


6. Conclusions

Results from three complete seasons clearly indicate fire incidence, severity and variation over time and between different regions. The availability of this time series information enables inter-year comparisons, initial studies of fire distribution within the country and analysis of seasonal trends.

Our interpretation demonstrates the potential of such data to increase understanding on extent and type of fire. It is critical to be able to discriminate detected fires between ‘good’ and ‘bad’ fires according to their impact on the environment, by including them in the ecological, social and economical contexts. This study also emphasises the importance of accurate and updated ancillary data to enhance appropriate interpretation of the information provided by the satellite imagery.

The satellite information should also be validated more systematically by field information. Remote sensing can obviously not substitute for direct ground observations completely. On the contrary, this technology, with its complementary input to existing knowledge, is expected to stimulate decision-makers at local, departmental and national policy level to reconsider the ways that fire information is assessed and incorporated into their activities.

What was achieved throughout this Project is very encouraging. There is now a greater appreciation and knowledge, within MARENA and on the part of forest managers and a number of local authorities, of the relevance, importance and use of NOAA/AVHRR data to assist monitoring and evaluating forest fires. A small scale remote sensing unit is now established and managed on a routine basis. In parallel, GIS capacity there has also experienced a recent and spectacular development, either in different divisions of MARENA or in other government agencies. It is hoped that, in the future, and provided there is a good collaborative spirit among institutions or departments, a wider collection of ancillary data (GIS layers) will be available, enabling more complete studies and at a wider range of scales.

Satellite information is but one tool to assist more informed decision-making. Up to now, Nicaragua has had no means to monitor the situation effectively and thus make informed decisions on natural resource pressures, effectiveness of policies, and areas to prioritise. Within the context of existing forest fire and natural resources management, Nicaragua now has a cost-effective mechanism to demonstrate the scale and nature of the problem and to adapt its policies accordingly. Nicaragua will also be able to use the same tool as a verifiable indicator to monitor the effectiveness of protection measures.

It is in this way that this low-cost and decentralised technology transfer will manage to bear fruit: to contribute to a better allocation of human and logistical resources, and to improve forest fire prevention and control strategies.


7. Acknowledgements

The authors acknowledge the UK Government’s Department For International Development (DFID), for their financial support to the Nicaragua Land Resources (Fire) Monitoring Development. Special thanks are also due to the British Ambassador and Deputy Head of Mission in Managua, for their open and collaborative support of the Project.

MARENA authorities and technical staff, who have been involved in the Project at any stage of its implementation and have made possible this output, are acknowledged.


8. References

Alvarez R. and G. Travisany. 1993. Plan nacional de protección forestal. Servicio Forestal Nacional, IRENA. Managua. 23 p.

Belward, A. S., J.-M. Grégoire, G. D’Souza, S. Trigg, M. Hawkes, J.-M.

Brustet, D. Serça, J.-L. Tireford, J.-M. Charlot and R. Vuattoux.

1993. In-situ, real-time fire detection using NOAA/AVHRR data. Proceedings of the 6th European AVHRR Data Users Meeting, 29th June to 2 July 1993, Belgirate, Italy. Eumetsat. Darmstadt, Germany, 333-339.

Bond W.J. and B.W. Van Wilgen. 1996. Fires and plants. Population and Community Biology Series 14. Chapman & Hall, London, 263 p.

Comisión Económica Para América Latina (CEPAL). 1998. El Fenómeno El Niño en Costa Rica durante 1997-1998. CEPAL-LC/MEX/L-363. Reliefweb.

Ciesla W.M.. 1997. Forest Fire Management: Assessment of Present Country Capacity and Needs for Additional Inputs. Natural Resources Institute, University of Greenwich, 69 p.

Collins M.. 1992. Les Forêts Tropicales. Published in collaboration with IUCN. Edts. Solar, Paris, 199 pp.

Downey, I.D., S.P. Flasse, J.B. Williams, P. Navarro. 1997. Overview of the Nicaragua Land Resources (Fire) Monitoring Project. Proceedings of a Technical Workshop of the Nicaragua Land Resources (Fire) Monitoring Project, Puerto Cabezas 19-20 nov. 1996. Natural Resources Institute, University of Greenwich, 8-16.

Eva H. and S.P. Flasse. 1996. Contextual and multiple-threshold algorithms for regional active fire detection with AVHRR data. Remote Sensing Reviews 14, 333-351.

Flasse S.P. and P. Ceccato. 1996. A contextual algorithm for AVHRR fire detection. International Journal of Remote Sensing 17, 419-424.

Flasse S.P., I.D. Downey, A. Jacques de Dixmude, P, Navarro, R. Alvarez, Z. Zuniga, I. Humphrey, F. Uriarte and A. Ramos. 1998. Cost effective operational remote sensing in support of forest fire monitoring : recent experiences from Nicaragua. Proceedings of the 27th. International Symposium on Remote Sensing of Environment, “Information for Sustainability”, 8-12 June 1998, ISRE, Norway, 773-776.

Food and Agriculture Organisation (FAO). 1993. Forest Resources Assessment 1990 – Tropical Countries. FAO Forestry Paper 112, Rome, 105 p.

Herrera L., N. González, A. Ponce. 1995. Recursos forestales nacionales. Ministerio del Ambiente y los Recursos Naturales, Direccion General Forestal, Managua, 29 p. Unpublished manuscript.

Kaufman, Y.J., A. Setzer, C. Justice, C.J. Tucker, M.G. Pereira, and I. Fung. 1990. Remote sensing of biomass burning. Fire in the Tropical Biota. Ecosystem Processes and Global Challenges, J. G. Goldammer (Ed.), Springer-Verlag, Berlin, 1-10.

Lacayo C.. 1998. Mapa de pobreza municipal de Nicaragua. Fondo de Inversión Social de Emergencia (FISE), Managua, 7 p.

Malingreau, J.P.. 1990. The contribution of remote sensing to the global monitoring of fires in tropical and subtropical ecosystems. Fire in the Tropical Biota. Ecosystem Processes and Global Challenges, J. G. Goldammer (Ed.), Springer-Verlag, Berlin, 337-369.

MARENA – Ministerio del Ambiente y los Recursos Naturales. 1995. Mapa forestal de Nicaragua. Raster layer in Idrisi for Windows. Unpublished manuscript.

MARENA – Ministerio del Ambiente y los Recursos Naturales. 1995. Mapa de muncipios de Nicaragua. Vector layer in Idrisi for Windows. Unpublished manuscript.

Moula M.. 1996. Modélisation des feux de biomasse en savane africaine et évaluation des émissions dans l’atmosphère de constituants en trace. Doctoral thesis. Université Paul Sabatier. Toulouse, France. Order Number 2262.

Prevedel, D.A.. 1995. Project Sparkey: a strategic wildfire monitoring package using AVHRR satellites and GIS. Photogrammetric Engineering and Remote Sensing, 61(3), 271-278.

Quan, J.. 1995. Visit report to the Nicaragua Land Resources (Fire) Monitoring Project. Natural Resources Institute, University of Greenwich, 11 p. Unpublished manuscript.

Stührenberg M. and P. Maître. 1996. Nicaragua, l’adieu aux armes ?. Geo – Un Nouveau Monde: La Terre, No. 209, Paris, 48-69.

Valerio L.. 1998. La producción de información elaborada a través de un sistema de información geográfica (SIG) a nivel local. Proceedings of the 2nd Workshop on Forest Fire Monitoring, Ocotal (Nueva Segovia), Nicaragua 27-29 January 1998. Natural Resources Institute, University of Greenwich, 60-64.

Williams, J.B. and J. Rosenberg. 1993. Operational reception, processing and application of satellite data in developing countries : theory and practice. Annual Conference of the Remote Sensing Society. Chester College 1993, UK, 76-83.

World Bank (WB). 1997. Nicaragua Atlantic Biological Corridor Project. Project Document. Washington, 133 p.

World Food Program (WFP). 1998. Emergency food assistance for families affected by El Niño in Central America. Regional EMOP-NIC-5949, Progress Report. Reliefweb.



Print Friendly, PDF & Email