Fire Review of Sumatra Island in 1997

 

Fire Review of Sumatra Island in1997


Introduction

During the extensive fire period that ravaged through Indonesia in 1997 and 1998, fire was attribute to different actors and many estimates for the amount of forest and land burnt were given. In this review we want to give a short overview of the 1997 fires in Sumatra. The aim of this is not to give a new estimate of area burnt, but review the evidence we can get by combining remote sensing data and existing maps. In this article we like to give an overview of what remote sensing can contribute to the understanding of the cause, impact and effects of fires in Indonesia.

To do this we first have to give some considerations about the limitations of the satellite sensors we use in this study. The satellite used in this study is the NOAA satellite (National Oceanic Atmospheric Administration) Advanced Very High Resolution Radiometer (AVHRR), with a ground resolution of 1 km. Although this satellite sensor has some serious drawbacks, it represents one of the few options for monitoring fires over such a large area as Indonesia.

The fire points detected by the satellite are called ‘hot spots’. The satellite sensor however was developed for weather and oceanic purposes both of which have temperatures below 400° C. The sensor measures the average temperature of 1 km2. This does not mean that a fire has to be this size, since a small hot fire can influence the average temperature of the 1 km2 pixel considerably. An intense fire of 50×50 m can be detected as a hot spot but also a fire of several hectares can ban be depicted as one hot spot. Therefore an area calculation is very difficult or impossible with only hot spot information. Another drawback in detecting is the false positives or fires not recorded by the satellite. Detecting hot spots is done by using the infrared temperature sensitive band. Giving a threshold value of the general temperature, all the spots hotter than this threshold will be recorded as hot spots. Unfortunately, this is not flawless. Bare soil, corrugated iron roofs, and low vegetation (e.g. grass) can have a very high temperature in the sun and can easily be wrongly assigned as hot spot (false positive). On the other hand, low-intensity fires burning under a thick canopy of forest will not produce a hot spot signal.

69909 Byte

Photo 1. Fire under canopy

The mis-classification of false positives in a small test area was more than 50% (Legg 1997). Furthermore, hot spot cannot be detected in areas with thick haze or smoke cover, as the sensor cannot penetrate haze, smoke or cloud.

Although so there are some problems with the NOAA satellite it is still a widely used tool to assess fire occurrence over a large area.

Facts and Figures

Indonesia is one of the three countries with the world largest remaining tropical rain forest. According to a 1996 survey conducted by the Food and Agricultural Organization of the United Nations (FAO) and the Government of Indonesia, 57% of the nation is forested, equalling 99.2 million ha. This is down from 127 million ha forest cover in 1980. The rapid disappearance of forest is caused by logging concessions (43 million ha under concession right in 1980) and expansion of agriculture by smallholders and large holders. Tree crop estates increased very rapid. The land area converted to the four main tree crops (rubber, oil palm, cacao and coconut) increased from 4.3 million ha in 1970 to 9.6 million ha in 1996. One of the fast expansion crops is oil palm, which in 1970 covered about 100,000 ha and expanded to almost 2 million ha in 1996.

7615 Byte

Fig. 1. Indonesia

Population increased from 120 million in 1970 to 195 million in 1995. The main population growth during 1970-1990 were in Central and South Sumatra and East Kalimantan.

The Sumatra island has a wet tropical climate, and consists of five major ecological zones, the lowlands of Western Sumatra, a mountain range, the footslopes, the peneplains, and the lowlands of East Sumatra. (Fig.2a). The zones have clear different soils, topography and climate (for details: see Scholz [1983]).

In Sumatra commercial logging concessions started in the 1970s and reached its peak in the 1980s. From the total area of Sumatra 30% is under active or passive logging concession today. Part of the increase in population in Sumatra is by transmigration. Transmigration started already in the beginning of this century and is still continuing today. The government transmigration programme has so far resettled 220,000 families (ca. 1 million people) to Sumatra (RePPProT 1990). The land allocated to transmigrants is well mapped and totals for Sumatra of 6% of land surface.

click to enlarge (62 KB)

Fig. 2a. Agro-ecological zones, provinces and population of Sumatra

click to enlarge (43 KB)

Fig. 2b. Hot spots in 1997

In this paper we want to review the fire situation in respects to these and other characteristics. The presence of forest concessions and timber estates, transmigration areas, agro-ecological zones, broad land use types, population density and closeness to roads or towns are all factors thought to have influence on the presence of fires. The data for Land use comes form the Digital Chart of the World (DCW, scale 1:500,000) and from The World Conservation and Monitoring Council (WCMC, scale 1:500,000).

Population and Infrastructure

The correlation between population pressure (as expressed in population densities) per province and hot spot density for the southern provinces is not very strong (corr coef. 0.42). Population pressure per province by itself so does not explain the hot spot occurrences. Fire by land clearing or by accident is more likely to occur where the land is easy accessible than further away form population centers or infrastructure. We therefore test this by looking at the amount of hot spots in relation to distance to roads, rivers and towns. Since many hot spots were recorded near the coast (Fig.2b), we also look at the distance to coastline and hot spot density. The data used here on rivers, roads, towns and coastline is from DCW.

59916 Byte

Photo 2. Easy road access and use of the land

 

8018 Byte

Fig. 3. Amount of hot spots in relation to distance rivers, roads, towns and coastline

In Figure 3 it can be seen that density near the coast is higher than average and drops to average further than 35 km away. This is possibly due to the swap area, which showed many hotspots, near the coast (see next paragraph). Rivers, roads and towns do not seem to affect hotspots as was expected. All objects show an increase in hotspot density further away from the object, to peak at a certain distance and then drop again. The peak of the hotspot densities is at different distance categories because of the density of the objects. The drainage network in Sumatra is very dense. This means that the zone of 0-20 km from river covers almost the entire island of Sumatra, while the same area is covered at zone 0-50 for the road network. This means that the river zones 20-25, 25-30 and further are almost empty with almost no land and no hotspots. Figure 3 indicates so that remote areas are more likely to burn than less remote areas.

Land Use

In Indonesia forest concessions for logging and plantations are a well-known land use changer. In Table 1 the hotspots are shown in the different land uses and eco-zones for Sumatra. This table shows the hotspot densities (number of hot spots per km2 * 100) and the overall contribution to the total amount of hotspots. Since the land use categories overlap each other the contribution of the different parts is more than 100.

The data in Table 1 show the density of hotspots in 1997 and 1998. The average hotspot density for Sumatra is 6.4. This figure can be used as an overall average fire density for all land uses and ecological zones in this year. All covers or zones with a higher density can be regarded as favouring burning while all land use with lower densities can be consider to mitigate burning. The areas with more than double the average densities are: Inland swamp, wetlands and Coastal eastern strip. These are all different terms in different dataset (DCW, WCMC) for the same area, the eastern lowlands of Sumatra (Fig.2). This eco-zone so is very susceptible to fire. Other areas with higher fire densities than average are Transmigration, Peneplain and HpH&HTI (HPH&HTI is forest concession and timber estates). The areas with less fire densities are: “Mountain forest (with different names in the different data), Protected areas, lowland Rainforest, and Alang-alang fields (Imperata cylindrica grasslands which invade degraded soils).

Tab. 1. Hot spots per land use zone in Sumatra Land use/cover

Hotspots

Area
(in sqkm) Hotspot density
(Hotspot/Area) Contribution
(Htspt/tot.htspt) Montane rainforest

272

31551

0.9

1.4

Western Coastal Strip

466

22246

2.1

2.4

Mangrove

628

10031

6.3

3.3

alang-alang

998

27225

3.7

5.2

Protected area

1064

59606

1.8

5.6

Piedmont

1220

63966

1.9

6.4

Sumatra

19053

465000

4.1

100.0

Transmigration

1307

25943

5.0

6.9

Mountainous area

1822

151521

1.2

9.6

Lowland rainforest

4805

122047

3.9

25.2

Inland swamp

5450

64183

8.5

28.6

HpH&HTI

6312

140481

4.5

33.1

Peneplain

6855

103473

6.6

36.0

Eastern coastal strip

7613

81865

9.3

40.0

Wetlands

9585

112138

8.5

50.3

 

28269 Byte

Photo 3. Fire in a forest

The hotspots in the swamp area contribute 28% of the total hotspots in Sumatra. Since the different ecological zones have very different fire densities, we analyze the fire densities of human made changes (transmigration, HPH&HTI’s) per ecological zone (eastern lowlands, mountain zone, peneplains, piedmont zone, west coastal strip) and the contribution the fires had to the total amount of hotspots.

The fires had two main impacts. The data shown in Figure 4 reveal the impact of the different land uses on fire occurrences and land-use changes. In Figure 5, however, the impacts of fire on the total contribution to trans-boundary smoke as function of number of hotspots is shown. The hotspot density is shown per eco-zone. From the last category Sumatra, the all island data, it is clear that transmigration areas and HpH&HTI areas have a higher hotspot density than other land uses. The Sumatra category, however, only shows the overall trend. The different ecological zones provide a quite different trend.

6678 Byte

Fig. 4. Hot spots densities in different agro ecological zones of Transmigration and HpH&HTI

From Figure 4 it is clear that in all agro-ecological zones the hotspot densities inside transmigration and HpH&HTI are lower than the average hot spot density outside these areas. Except in the areas with few hotspots, the mountains and west coast, here transmigration has a higher density than the non-transmigration and HpH&HTI category. It is also clear that the non-trans-and-HpH and transmigration and HpH categories show similar overall trends with high densities in the eastern lowlands and lowest densities in piedmont and mountain zone. It can be concluded from this graph that the presence of forest concessions and/or transmigration areas do not necessarily increase the hotspot occurrence in an area.

In Figure 5 the contribution of the ecological zones and land conversion is shown. From this it is clear that although transmigration areas can have sometimes a very high hotspot density they do only contribute very little to the overall hotspot numbers.

8951 Byte

Fig. 5. Hot spot contribution per zone

HPH&HTI’s are a large contributor to the total amount of hotspots. However in most ecological zones other land use are the main contributor of hotspots except in the Eastern lowlands. 40% of the total hotspots detected during the 1997 fire episode were in the eastern lowlands zone, with 25% of those in HpH&HTI.

Timing

The delivery of hotspots started at 11 September 1997. As can be seen in Figure 6, the number of hotspots rose to over 10,000 recorded in one week. Remarkable is the very dense time spacing. From the 20,000 hotspots recorded from 11 September until 31 December 1997 more than half of all the hotspots were recorded in a period of one week (12 to 18 October 1997) – a remarkably short time period.

9076 Byte

Fig. 6. Hot spots in time

Conclusions

By analyzing the fire episode of 1997 in a spatial way some interesting facts can be found and the episode can be put in a broader perspective of what happened where and when.

Most hotspots were detected in only a limited amount of time. In one week time more than 50 % of all hotspots of 1997 were registered. This indicates that fires must have had numerous ignition points. This seems to be very unlikely if this would come from natural causes. It seem likely that fires were human-made. The question arises then if the fires were indeed forest fires, as the term used during the fire episode of 1997, or where other cover types burned. The result of the analyses carried out here shows that fires were not in natural forest areas. Protected forest, lowland and mountains forest were areas with the lowest fire densities and lowest contribution of hotspots. This seems to be good news for the forest environment.

The next question is than if the fires were not natural and not in forested areas were and why were they set then? It seems that most fires occurred in remote areas far away form cities, roads and rivers. The cover type with the heaviest burning was in the wetland zone. This zone contributes more then 50% of the total amount of fires in Sumatra. In this zone more than 60% of the fires were in logging concessions and timber estate areas. Logged forest is well known to be very susceptible to fire since it has much dead material that easily ignites. This explains how the fires could burn such large areas, but does not explain why it was set on fire since it happened at so many places at the same time.

There could be two explanations. As mentioned in the Sumatra overview paragraph, the pressure on land is lately increasing in Indonesia. Especially the demand for oil palm plantation has converted much land to plantation and the demand is still increasing. The change of land use from small holder to large estates has brought also conflicts in land use between these parties. One explanation so is that although fires were found on concessionaires lands, large holder did not set these fires themselves, but that these fires were set by others to settle a land use conflict.

The second explanation is that concessionaires burn their land after logging to start a plantation. This is known to occur in Indonesia. Especially oil palm plantations are increasing rapid in Indonesia since the profit margins are high in this trade. Until now 2.4 million ha of land are already converted to oil palm and the target for the year 2000 is set on 4 million ha. If all hotspots were a sign of total land clearance for conversion to oil palm and we assume the fire extent was half the pixel size for each hot spot and no pixel was double counted than the land clearing this year would just be enough (1.5 million ha) to reach the target. This shows the enormous amount of land use change happening today in Indonesia.

The transmigration areas do have higher fire density than other land uses. However the transmigration fires were few compared to fires in other land uses and only contributes to less than 10% of the total hotspots in 1997. The transmigration so although can be a driving factor of land use change and land use conflicts the importance for haze and green house gases is limited.

The analyses by different eco-zones show that the fire situation in different provinces and different eco-zones are very different. It seems logical that the underlying driving forces so are different in different provinces and zones.

In this respect and also because of at least 50% of the fires not being in forested area, the term forest fire does not describe adequate the area under fire. Another term used during the fire event was wildfire. This term neither seem to be adequate since this would indicate the fires happened by accident which seem highly unlikely. The best term suggested by the authors is land fire, since this neither implicates which cover burned nor what the reasons were.

The Indonesian economy is on the move. This has many changes in itself. Also land use change when small holders are replaced by large plantation. With this large land conversion is taking place. When the targets have to be met only for oil palm it seems that the El Niño drought was not the reason for the extreme fire situation but a convenient circumstance to convert in one season what would have been converted anyhow in the seasons to come. It is therefore the question if the El Niño was an ecological disaster or convenient for planned land use.

The figures presented here suggest that although El Niño surely set the stage for the smoke episode of 1997-1998 it did not affect entire Sumatra nor were the same causes in different zones and provinces. When the swamps caught fire, the disaster of the smoke and haze problem was born and attention focuses on the forest fire in Indonesia. In this paper the figures suggest that actually a small area in a limited time frame was the main source for fires. This suggests that regulating land preparation on certain dates in certain areas could prevent another nation wide disaster in the future.

The contribution of remote sensing during the fire event can not be underestimated. It alarmed the people and government of Indonesia and revealed which actors and places played important roles. In this study remote sensing and digital data it is used to give a review of what happened in 1997 and to point out what the possible causes are and give some idea of where to focus on the next time when such a disaster happens.

References

Tomich, T.P., A.M.Fagi, G.Michou, D.Murdiyarso, F.Stolle, M.Van Noordwijk. 1998. Indonesia’s fires: smoke as a problem, smoke as a symptom. Agroforestry Today 10 (1) January-March 1998.

RePPProT. The land resources of Indonesia. A national overview. 1990. Land resources department, Natural Resources Institute, Overseas Development Administration, Foreign and Commonwealth Office, London, England, and Direktorat Bina Program, Direktorat Jenderal, Penyiapan Pemukiman, Departemen Transmigrasi, Jakarta, Indonesia.

Scholz, U. 1983. The Natural regions of Sumatra and their agricultural production patterns. A regional analysis. Ministry of Agriculture, Republic Indonesia, Agency for Agricultural Research and Development, Central Research Institute for Food Crops (CRIFC).

GTZ-SFMP. 1998. Forestry highlights from the Indonesian Press, April/May 1998.

Uhl, C., and J.B.Kauffman. 1990. Deforestation, fire susceptibility, and potential tree responses to fire in the eastern Amazon. Ecology 71, 437-449.

Goldammer, J.G., and B.Seibert. 1990. The impact of droughts and forest fires on tropical lowland rain forest of Eastern Borneo. In: Fire in the tropical biota. Ecosystem processes and global challenges (J.G.Goldammer, ed.), 11-31. Ecological Studies 84, Springer-Verlag, Berlin-Heidelberg-New York, 497 p.

Lennertz, R., and K.F.Panzer. 1984. Preliminary assessment of the drought and forest fire damage in Kalimantan Timur. DFS German Forest Inventory Service Ltd., for the German Agency for Technical cooperation (GTZ), Ltd. Report of the fact finding Mission Nov/Dec. 1983.

Legg, C.A. 1997. A preliminary report on fires in Sumatera, Kalimantan, Sulawesi and Irian Jaya during September 1997. Paper presented at the International Workshop on National Guidelines on the Protection of Forest Against Fire, Bogor, December 8-9. 1997.

ESRI, Digital Chart of the World. For use with Arc/Info software, data dictionary. ESXR,Inc,1993

 

From: F.Stolle and T.P.Tomich and R.Dennis
Address:

International Centre for Research
in Agroforestry (ICRAF)
P.O. Box 161
Bogor, 16001
Indonesia

Fax: ++62-251-625-416
Tel: ++62-251-625-415
e-mail: F.stolle@cgiar.org
           T.Tomich@cgiar.org
Center for International Forestry Research
(CIFOR)
P.O.Box 6596 JKPWB
Jakarta 10065
Indonesia

Fax: ++62-251-622-100
Tel: ++62-251-622-622
e-mail: R.dennis@cgiar.org


Top
Back
Projects & Programmes in SE Asia

 

Print Friendly, PDF & Email
WP-Backgrounds Lite by InoPlugs Web Design and Juwelier Schönmann 1010 Wien