Open Journal of Forestry
2013. Vol.3, No.4, 129-137
Published Online October 2013 in SciRes (http://www.scirp.org/journal/ojf) http://dx.doi.org/10.4236/ojf.2013.34021
Recent Trends of Fire Occurrence in Sumatra
(Analysis Using MODIS Hotspot Data): A
Comparison with Fire Occurrence in Kalimantan
Nina Yulianti1, Hiroshi Hayasaka1, Alpon Sepriando2
1Graduate School of Engineering, Hokkaido University, Sapporo, Japan
2Meteorological, Climatology and Geophysical Agency of Central Kalimantan, Palangkaraya, Indonesia
Received June 21st, 2013; revised August 6th, 2013; accepted August 21st, 2013
Copyright © 2013 Nina Yulianti et al. This is an open access article distributed under the Creative Commons
Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the
original work is properly cited.
MODIS (Moderate Resolution Imaging Spectroradiometer) hotspot and precipitation data for the most
recent 11-year period (2002 to 2012) were analyzed to elucidate recent trends in the seasonal and spatial
fire occurrence in Sumatra and the relationship with precipitation. Using a latitude line of S 0.5˚, Sumatra
was divided into two regions, N. (north) and S. (south) Sumatra. Different trends in seasonal fire occur-
rence were discussed and further defined by considering two different precipitation patterns. Analysis of
hotspot (fire) data was carried out using 0.5˚ × 0.5˚ grid cells to evaluate recent trends of spatial fire oc-
currence. Analysis results of hotspot and precipitation data were also tallied every 10-day to find the rela-
tionship between seasonal fire occurrence and the dry season. Standard deviation (SD) and variance (V)
were then used to evaluate fire occurrences in Sumatra and Kalimantan objectively. The relatively mild
fire occurrence tendency in Sumatra compared to Kalimantan could be the result of different stages of
forest development or the high deforestation rate in Sumatra compared with Kalimantan. This paper also
shows that the two different seasonal fire activities in N. and S. Sumatra were closely related to the two
different dry season types: a winter and summer dry season type (WD & SD) in N. Sumatra, and a summer
dry season type (SD) in S. Sumatra. Extreme fire occurrences in the Dumai region in 2005 and Palembang
region in 2006 could be partially explained by a severe drought occurrence enhanced by two different
kinds of El Niño events.
Keywords: Drought; Dry Season; El Niño; MODIS Hotspot; Peatland Fire
Introduction
During 1997 and 1998, areas of forest measured approxi-
mately 45,000 km2 were destroyed by extreme fires in Kali-
mantan and Sumatra, releasing more than 2.6 gigatonnes of
CO2 to the atmosphere (Tacconi, 2003), and causing a trans-
boundary haze event in South East Asia. An estimate of the size
of the area burned in East Kalimantan alone, between January
and April 1998, was 52,000 km2 (Siegert et al. 2001). In addi-
tion to this vast amount of damage, the fires caused serious
health and economic problems within the region (Dennis, 1999).
Since then, Sumatra and Kalimantan have experienced frequent
fire occurrences, but as discussed in our preliminary paper,
(Yulianti et al., 2012), the number has diminished. It is notable
that the frequent fire occurrence in Sumatra coincides with the
rapid rate of deforestation occurring over the past 10-years
(2000-2012), which has been at a rate of about 23.7%/(10-yr)
for all forests and 5.2%/(yr) for peat swamp forest (Miettinen et
al., 2011a).
Since 1990, deforestation in Sumatra has contributed to a
loss of 7.45 million hectares in an area of primary forest meas-
ured 75,000 km2 (Margono et al., 2012); a faster rate than in
Kalimantan. In other words, the present area of forest in Suma-
tra (120,000 km2) is smaller than that of Kalimantan (250,000
km2), as officially reported by the Ministry of Forestry of In-
donesia (MoF, 2012). It is also evident that most fires are re-
lated to agriculture activity, and this is consistent with Sumatra
being home to the largest industrial plantations in Indonesia
(15,280 km2 Joosten et al., 2012), as well as the most prolific
cultivation of palm oil (10,467 km2, Miettinen et al., 2012).
During development of the Tidal-coast Development Project,
which began in the late 1960s in Sumatra (Furukawa, 2004), for
instance, the swamp drainage channels caused a noticeable
reduction in the dry season water table (Fearnside, 1997) and
provided a low ground water level environment, which is very
suitable for the occurrence of peat fires. This also occurred in
the MRP area in Kalimantan (Putra & Hayasaka, 2011). Slash-
ing and burning activities associated with the rapidly increasing
number of plantation crops and industrial forests, have de-
stroyed the peat swamp forests in both Kalimantan and Sumatra.
This has caused serious environmental problems, and long-
burning fires occur from January to October every year (Yuli-
anti et al., 2012). For instance, vegetation fires in the southern
peatland of Sumatra, often occur in relation to burning activities
within the plantations during the dry season (Stolle et al.,
2003).

N. YULIANTI ET AL.
Once a fire begins, its behavior is highly dependent on the
prevailing environmental factors, such as the weather condi-
tions and the desiccated state of the material to be burned.
Rainfall conditions are an extremely important factor in the
tropical zone, and rainfall patterns can be affected by active fire,
particularly on peatland areas (Putra & Hayasaka, 2011). The
characteristics of fires in the Sumatra peatland areas are also
influenced by the rate of decomposition of the peat, as de-
scribed in Saharjo (2005). Under dry conditions, peat that is
more highly decomposed (mature) is very flammable and re-
sults in a longer period of smoldering fire, as discussed by Usup
et al. (2004). Thus, the emissions of carbon solely from peat
fires in Southeast Asia, (mostly from the Sumatran and Kali-
mantan peatlands) are very similar to emissions from the whole
of Sub-Saharan Africa (Harris et al., 2012). However, vegeta-
tion fires related to forest clearing activities tend to be more
widespread in Sumatra, even outside peatland or undeveloped
areas which are far from sea coast. Recent frequent fires related
to deforestation in the Riau areas of North Sumatra Province
have also become a global environmental issue, due to the
amount of CO2 emitted (3.6 Giga tons CO2; Uryu et al., 2008).
The use of satellite-based monitoring enables a more com-
prehensive understanding and evaluation of fire activities inside
forest areas, and also in large peatland areas. Since 2002, the
MODIS sensor on board the Terra and Aqua satellites has cov-
ered the whole of Indonesia. Several earlier studies using
MODIS hotspot data have focused on Asia (Vadrevu & Justice,
2011), Southeast Asia (Miettinen et al., 2011b), and Borneo
(Langner & Siegert, 2009). Our previous research analyzed the
recent fire trends during the MODIS era for Indonesia (Yulianti
et al., 2012), and Kalimantan (Borneo part of Indonesia, Yuli-
anti & Hayasaka, 2013). Peat fire activity in the MRP area was
discussed in Putra & Hayasaka (2011). However, there have
only been a few reports related to fires in Sumatra.
In this study, we carefully analyzed the seasonal and spatial
occurrence of fire using the last 11-year (2002-2012) of
MODIS hotspots (fire) and the last 12-year (2001-2012) of
precipitation data. We attempted to determine trends in the
seasonal and spatial occurrence of fire under severe drought
conditions in north and south Sumatra separately. Analysis of
seasonal fire occurrence was carried out using hotspot and pre-
cipitation data at 10-day intervals from January to December.
Fire (hotspot) distribution maps were drawn to identify active
fire areas in Sumatra for the analysis of spatial fire occurrence.
The results of similar research that was conducted on Kaliman-
tan fires by Yulianti and Hayasaka (2013), were compared to
the above-mentioned analyses, to observe typical trends in the
occurrence of peatland fires in Sumatra and to enable an effec-
tive, future peatland fire forecast.
Methodology
Study Area and Pe atland
The study area covers Sumatra Island from N 6˚ to S 6˚ and
from E 95˚ to 108˚ (Figure 1). Sumatra covers a total land area
of about 474,000 km2 (approximately 89.1% of the size of Ka-
limantan which covers 532,000 km2). The total peatland area of
Sumatra is about 72,000 km2, which is approximately 15% of
the land area of Sumatra; (Wetlands, 2003), (120% larger than
that of Kalimantan at 58,000 km2). (The distribution of peatland
is shown in a dark color (brown) in Figure 1). The main area of
peatland in Sumatra has developed in a large area of lowland on
Figure 1.
Map of peatland, weather stations, and the eight regions used for
analysis in Sumatra.
the east coast, covering a distance of roughly 1200 km from the
north of Dumai in Riau (~N 2.6˚) to the north of Lampung (~S
5˚), and covering a distance of nearly 200 km (maximum dis-
tance) inland from the coast. Other several small areas of peat-
land are independently located on the west coast and inland,
(west of Riau, Jambi and Palembang). North Sumatra has the
deepest area of peatland, at a depth of around 9 m (Jauhiainen
et al., 2011); in S. Sumatra the depth is shallower, at about 4 m
(Wetlands, 2003). Nowadays, major areas of peatland are
maintained by large plantation companies by the provision of
drainage systems. However, many of these use unsuitable sys-
tems for the preservation of peatlands, and cause a very low
ground water level).
In this paper, in order to clarify spatial and seasonal fire oc-
currences in Sumatra, the authors have given suitable names to
regions. The borders of these regions are different from those
used in conventional political and geographical maps. We
achieved this by dividing Sumatra into two regions, N. (north)
and S. (south) Sumatra, by using a latitude line of S 0.5˚ and
with consideration of the following: two different precipitation
patterns (Aldrian & Susanto, 2003), an historical climatic map
showing precipitation patterns in Sumatra (Oldeman et al.,
1979), and a previous study by Chang et al. (2005), (as shown
in Figure 1).
Analysis using a 0.5˚ grid cell provided approximately 105
and 95 cells for N. and S. Sumatra respectively. These cells in
N. and S. Sumatra were then grouped into regions, to clarify the
spatial and seasonal fire occurrence, particularly for areas of
peatland (shown in Figure 1). The borders of these regions
were constructed on the connecting lines from the rectangular
analysis cells (0.5˚ grid cell). We then allocated suitable re-
gional names, such as “Dumai + 14” (14 cells, a subset of
North Sumatra and Riau Province), “Pekan Baru + 12” (12 cells
for a subset of Riau Province) in N. Sumatra, “Jambi + 7” (7
cells, a subset of south Riau and Jambi Province) and “Palem-
bang + 16” (16 cells, a subset of South Sumatra and north
Lampung Province) in S. Sumatra. Borders of these four re-
gions are shown by the lines in Figure 1. We put “+” after each
130

N. YULIANTI ET AL.
region to show the expanded region of each local area name
(such as “Dumai+”), so as to avoid confusion with the conven-
tional administrative name. The number provided at the end of
the names is used to describe the number of cells that the region
occupies. The other two regions, and the remaining cells from
the above four regions, were simply named N. Others (79 cells),
and S. Others (72 cells). Finally, two special areas, Dumai 4 in
N. Sumatra and Palembang 6 in S. Sumatra, were introduced to
highlight the extreme fire activities in these areas in 2005 and
2006.
MODIS Hotspot Data and Grid Analysis
Daily MODIS hotspot data (Collection 5.1 active fire product)
from the most recent 11-year period (2002 to 2012) were auto-
matically extracted from the FIRMS website (Fire Information
for Resources Management System, http://earthdata.nasa.gov/
data/near-real-time-data/data/firms) and were then analyzed in
this paper. Approximately 250,000 hotspots were recovered from
the 11-year data for Sumatra, (which includes Insular Malaysia).
In this paper, 0.5˚ grids were used to clarify the spatial trend
of fire occurrence. Hotspots were tallied depending on their
longitude and latitude. For a hotspot density, units of “hot-
spots/cell,” and “hotspots/km2,” were used. The approximate
area of these grid cells at the equator is 3070 km2, and this was
used for the conversion to hotspot density, “hotspots/km2”.
Weather and Precipitation Data
To determine the relationship between fire activity on peat-
land and weather patterns, we attempted to obtain weather data
from many places in Sumatra, but were only able to purchase
daily precipitation data from six stations. In this paper, we ana-
lyzed precipitation data for the most recent 12-year period
(2001 to 2012) measured at three airports (see Figure 1) in
Medan (Polania, N 3.66˚, E 98.60˚, North Sumatra), Pekan
Baru (Sultan Syarif Kasim II, N 0.47˚, E 101.45˚, Riau), and
Palembang (Sultan Mahmud Badaruddin II, S 2.90˚, E 104.70˚
South Sumatra). All data belongs to the Indonesia Meteoro-
logical, Climatology and Geophysical Agency (BMKG).
Precipitation data measured at three weather stations in Su-
matra were tallied every 10-day to define the dry season period
and to explain the different trends in fire occurrence in N. and S.
Sumatra. Data processing from every 10-days was applied in-
stead of using other data smoothing periods of a 10-day (week)
and 30-day (month), because the smoothing of data over a pe-
riod of 10-day is less difficult than over other time periods. In
addition, the 10-day period will enable us to label seasons by
using expressions such as, “early June,” instead of “DN = 152
to 161”.
Results and Discussions
Dry Season Type, Period, and Lowest Daily Mean
Precipitation
Precipitation patterns for three weather stations (Medan,
Pekan Baru (known as P. Baru), and Palembang) were calcu-
lated using the recent 12-year interval from 2001 to 2012 (see
Figures 2(a)-(c)). In total, 35 mean daily precipitation values
for every 10-day interval, were calculated for each station, and
are shown by the solid lines in the graphs in Figure 2. Each
daily mean precipitation value was placed on its representative
Figure 2.
Dry season period in: (a) Medan; (b) Pekan Baru; and (c) Palembang
derived from 12-year (2001-2012) data.
point from DN = 5 to DN = 355 within the 10-day interval. In
addition, simple smoothed curves (dotted lines) have been
added to the graph, representing the daily mean precipitation.
With the help of these two lines, we were able to ascertain the
dry season period and the low precipitation values during each
dry season, for each station.
The definition of a dry season period for each area in this
paper, follows previous analysis of Kalimantan by Yulianti &
Hayasaka, (2013), and the daily mean precipitation of 5
mm/day was used as a threshold value for the dry season. By
using the line for daily mean precipitation in our graph in Fig-
ure 2, we can see that threshold value is apparent in Medan (N
3.66˚) in the Northern Hemisphere, revealing an approximately
two month winter dry season in January and February (WD(2))
and a summer dry season (SD(2)) in June and July. In Palembang
(S 2.90˚) in the Southern Hemisphere, there was a relatively
long dry season, about 4-month SD(4) from June to September,
and in P. Baru (N 0.47˚) near the Equator, (located between
Medan and Palembang, see Figure 1), there was an intermedi-
ate dry season, or a one-month quasi-winter dry season with
rain (daily mean precipitation in February higher than 5.5
mm/day), and a dry season of about 2-months SD(2) in June and
August.
The lowest values of daily mean precipitation during the

N. YULIANTI ET AL.
above-mentioned dry seasons were an indication of fire activi-
ties near each weather station. In Medan, the lowest values of
daily mean precipitation in WD(2) and SD(2) found in Figure 2(a)
were about 2.1 mm/day in February and 4.3 mm/day in July
respectively. In Palembang, the lowest value in Figure 2(c) was
1.7 mm/day in SD(4) in August. In P. Baru, the lowest value was
3.4 mm/day, from only one period in each of June and August,
as shown in Figure 2(b). Moreover, the annual mean precipita-
tion of P. Baru (about 3200 mm) was higher than the amounts
of about 2500 and 2600 mm from the other two weather sta-
tions. Reasons for this were that we could estimate that the
heavy precipitation in P. Baru would mainly be due to the effect
of the mountain behind the area (see Figure 1).
Fire Prone Areas and Peatland
Fire prone areas (>100 hotspots/(yr. cell) in the most recent
11-year period (2002 to 2012) are highlighted using different
colors in Figure 3. Most of these areas are located on the east-
ern side of Sumatra, but some are in the northern part. There
were 12 cells showing a very high hotspot density (>400 hot-
spots/(yr. cell) = 0.129 hotspots/(yr. km2)). These cells were
named H1, H2, …, H12, in descending order of hotspot density.
The 7 highest hotspot density cells (H1~4, H8, H11 and H12),
located in the Dumai region or Dumai + 14p (Dumai4p: H1, H3,
H4, and H12), cover most of the northeast coast peatland in N.
Sumatra (Riau Province). According to the peatland map (Wet-
lands, 2003), the Dumai region contains a peat layer that is
relatively deeper (~8 mm) than other places in S. Sumatra and
Kalimantan (except the MRP area). It is of note that Dumai +
14p belongs to a different climate zone than most of S. Sumatra
and Kalimantan (see Figures 2(a) and 3).
The seventh highest cell, H7, is located on the eastern side in
Pekan Baru (capital of Riau Province) near the Equator and on
the coast. The other highest cells, H5, H6, H9, and H10, are
located in the region of S. Sumatra. H5 and H6 are located on
the eastern side at Palembang (capital of the South Sumatran
Province), and two cells in Palembang 6p. H9 in the south in
Figure 3.
Map of the 12-highest hotspot cells, fire prone cells, and peatland in
Sumatra.
Pekan Baru, and H10 in southeast Jambi are two cells included
in Jambi + 7p (see Figure 3).
From the distribution of these highest hotspot cells, it is evi-
dent that the most recent fires in Sumatra have occurred mainly
on the coastal peatland. Many of the fires on peatland can be
explained by the history of development as in the MRP area of
Kalimantan (Furukawa, 2004). Areas in the MRP with dense
hotspots were related to high human activity such as deforesta-
tion, slash and burn clearing, and the maintenance of planta-
tions (Yulianti et al., 2012).
Recent Fire Occurrence Trends in Sumatra
Annual and Average Seasonal Fire Occurrence in
N. Sumatra
The stacked bar graph in Figure 4 shows the number of fires
that have occurred in three regions in N. Sumatra (from bottom
to top: Dumai + 14, P. Baru + 12p, and North Others 79). The
annual mean numbers of fires in these three regions and those
in S. Sumatra, (and the annual fire in Kalimantan), are shown in
the two bars on the far right in Figure 4. The unit of the Y-axis
is the number of hotspots in each region.
The annual mean number of hotspots in N. Sumatra over the
past 10-years was about 8600, and 60% of these fires occurred
in peatland areas (Dumai + 14p and P. Baru + 12p). A com-
parison of the two bars in Figure 4, shows that the number of
fires in 2005, (about 27,000 hotspots/yr.), is larger than the
average number in Kalimantan (23,000 hotspots/yr.). About
80% of the extreme fires in 2005 occurred mainly in peatland in
the Dumai and P. Baru regions; areas responsible for approxi-
mately 16,000 and 5400 fires (60.3% and 20.4%) respectively.
Fire occurrences in both peatland areas were 4.2 times larger than
in the North Others 79, which is predominantly non-peatland.
The occurrence of fires in most years (6 out of the 10 years)
in N. Sumatra was in the range of +1σ to −1σ. The variance of
fire occurrence (V) of about (3900)2 was smaller than that of
about (16,000)2 in Kalimantan. This relatively stable fire oc-
currence in N. Sumatra may suggest that the occurrence of so-
called routine fires on plantations and developed land are not so
strongly related to the weather conditions, and the yearly fire
Figure 4.
Recent trends in annual fire occurrence in N. Sumatra.
132

N. YULIANTI ET AL.
occurrences could be related to agriculture activities such as
agricultural residue burning. Small-sized fire occurrence could,
therefore, imply that large land clearing or deforestation activi-
ties are no longer taking place in N. Sumatra.
The extreme fire occurrences in Dumai + 14p in 2005 was an
exceptional case, and the main cause of these severe fires was
due to the expansion of fires lit for land clearing and agricul-
tural activities. On inspection of weather data measured at
Medan, (located about 400 km northeast of Dumai), these fires
may also have occurred during severe drought conditions. In
Medan, the drought lasted for about two months, from
mid-January to mid-March in 2005, and overlapped with the
dry season in N. Sumatra.
In this paper, to clarify the average seasonal fire occurrence
tendency in N. Sumatra, the average number of hotspots every
10-days (1/3-month) were calculated. In doing so, one active
fire area named “Dumai 4p” was extracted from a part of Du-
mai + 14p, to illustrate the seasonal fire occurrence clearly.
From Figure 5, it can be seen that the seasonal fire occur-
rence trend in N. Sumatra displays two major fire seasons: one
fire peak in February and one in March, and another from May
to August. Such seasonal fire peaks can be partially explained
by the two precipitation patterns and the dry season periods for
Medan (N 3.5˚) and P. Baru (N 0.5˚) in Figure 2. We have
observed that, firstly, the winter fire peak in February (in Fig-
ure 5) is mainly due to fires occurring in Dumai 4p (N 1.8˚)
and P. Baru + 12p. These fires correspond to the winter dry
seasons WD found in Figures 2(a) and (b). The highest fire
peak for Dumai 4p (mid-February) corresponds to the lowest
daily mean precipitation (about 2 mm/day) in Figure 2(a).
Secondly, the early summer peak in June in Figure 5 is mainly
due to fires occurring in North Others 79 and Dumai + 14p.
These fires can be mainly explained by the summer dry season
in P. Baru in Figure 2(b), because the low latitude 32 cells (<N
2.5) in North Others 79 are responsible for about 80% of the
fires, and these cells have a similar rainfall pattern to P. Baru.
Thus, most fires in N. Sumatra tend to occur under a rainfall
condition of about 3.5 mm/day in mid-June, as seen in Figure
2(b). Thirdly, the highest fire peak in the whole of N. Sumatra
occurs in early August. This peak in August is mainly due to
the increased number of fires in P. Baru under the lowest daily
rainfall of about 3.5 mm/day. Such a tendency may, therefore,
imply the occurrence of peat fires in P. Baru+.
0
20
40
60
80
100
120
0 60 120 180 240 300 360
Average
number
o
f
hotspots
(hotspots/day)
Day
Number
North Others ( 79 cells)
P.Baru+ 12p
J
F
M A M
J
J
A
S
O
N
D
Average fire occurrence
in N. Sumatra 10‐yr data:
2002‐2012 (except 2005)
:Total Number
in North
: Dumai+ 14p
: Dumai+ 4p
Areas mostly
on peatland
Figure 5.
Recent trends of average seasonal fire occurrence in N. Sumatra.
Annual and Average Seasonal Fire Occurrence in
S. Sumatra
The annual mean number of fires in S. Sumatra is shown in
Figure 6 (using a similar graph format to Figure 4). The
stacked bar graph in Figure 6 shows the number of fires in
three regions in S. Sumatra, from bottom to top: Palembang +
(16p), Jambi + (7p), and South Others 72. The annual mean
number of fires in the three regions (excluding 2006), and the
annual number of fires in Kalimantan, are shown by the two
bars on the far right in Figure 6.
The annual mean number of hotspots in S. Sumatra during
the last 10-years was about 8900, and, approximately 47% of
the fires occurred in peatland areas (Palembang+ and Jambi+).
Fire data from 2006 were excluded from the statistics calcula-
tion, due to severe fire occurrence (+5.3σ, +2.6σ if the data
from 2006 are included) under the drought enhanced by El
Niño in 2006. A comparison of the two bars in Figure 6, shows
that the number of fires in 2006, (about 29,000 hotspots/yr.),
was larger than that of the average number in Kalimantan
(23,000 hotspots/yr.). About 57% of these extreme fires in 2006
occurred mainly in peatland areas in the Palembang+ and
Jambi+ regions, which are responsible for around 13,300 and
3300 fires (45.5% and 11.4%), respectively. The occurrence of
fires in both these peatland areas was 1.3 times larger than that
in South Others (mostly non-peatland).
The fire occurrence in the majority of years in S. Sumatra, (6
out of the 10 years), was within the range of +1σ to −1σ and the
variance of fire occurrence (V) of an area of around (7200)2
was smaller than that of an area of about (16,000)2 in Kaliman-
tan, but larger than that of an area of about (5900)2 in N. Suma-
tra. The relatively stable fire occurrence in S. Sumatra may also
suggest that continual and small-size fire occurrences are not
strongly related to weather conditions, as in N. Sumatra. It is
evident that the occurrence of small fires in heavy rain condi-
tions during 2010 could have made the variance of fire occur-
rence in S. Sumatra larger than that in N. Sumatra.
The extreme fire occurrence in Palembang+ in 2006 was an
exceptional case and the main cause of these severe fires was
due to an expansion of fires for land clearing and agricultural
Figure 6.
Recent trends in annual fire occurrence in S. Sumatra.

N. YULIANTI ET AL.
134
Dumai, and two cells of H5 and H6 are near Palembang. A
comparison of the seven highest cells from the top 12 highest
hotspot density cells in Kalimantan (Yulianti & Hayasaka, 2013)
are listed in Table 1.
activities, and which occurred under severe drought conditions
lasting three months from early August to late October in 2006,
and which overlapped with the dry season in S. Sumatra.
Figure 7 is similar to that of Figure 5 for N. Sumatra, and
was used to clarify the average seasonal fire occurrence ten-
dency in S. Sumatra. By using this graph, one fire active area
named “Palembang 6p,” was defined and extracted from one
part of Palembang + 16p, to clearly illustrate the seasonal fire
occurrence.
From Table 1, the fires in H1 near Dumai could be referred
to as “extreme” fires, because the average and the maximum
number of hotspots in H1 were 941 and 4494, respectively. The
average number in H1 was almost the same as the value of 971
in the H1k cell in Kalimantan. However, the greatest value in
H1 (4494) was the highest overall. SD values of 2.77 - 2.94 for
the four highest cells in Sumatra were higher than those values
(1.45 - 2.4) in Kalimantan (seen in Table 1). These differences
could imply that the extreme fires in Sumatra with higher SDs
were incidental fires. In contrast, the extreme fires in Kaliman-
tan with lower SDs were routine fires lit for the purpose of
development. The seasonal occurrence of these extreme fires in
Sumatra is discussed in the sections below.
From Figure 7, the trends of seasonal fire occurrence in S.
Sumatra show one fire season from May to late October, with a
few peaks in September. This one-fire-season pattern in S. Su-
matra almost corresponds to the one-summer-dry-season pat-
tern in Figure 2(c). The difference in the fire peak in Septem-
ber of about one month in Figure 7 and the low of precipitation
in August in Figure 2(c) can be estimated using the author’s
previous research related to Kalimantan, where we found there
was a time lag of approximately one month between precipita-
tion and a rising ground water level (Putra & Hayasaka, 2011).
This result was obtained from measurement results in peatland
in Block C, the MRP, in Central Kalimantan (Takahashi et al.,
2007).
The Extreme Seasonal Fire Occurrence and Drought in N.
Sumatra (2005)
In Figure 8, seasonal fire occurrence and drought in N. Su-
0
20
40
60
80
100
120
0 60 120 180 240 300 360
Average
number
o
f
hotspo ts
(hotspots/da y)
Day
Numbe
r
: Total Number
in South
South Others
( 72 cells)
Jambi+ 7p
: Palembang+ 16p
J
F M A M J
J
A S O
N
D
Average fire occurrence
in S. Sumatra 10‐yr data:
2002‐2012 (except 2006)
:Palembang 6p
Areas mostly
on peatla nd
From Figure 7 the highest peak can be seen in late Septem-
ber for all three regions in S. Sumatra. Palembang 6p had the
same values as the MRP area in Central Kalimantan, (Yulianti
& Hayasaka, 2013). This could suggest that peat fires in S.
Sumatra and south Kalimantan occur under a similar weather
condition created by the southern monsoon during the summer
months (between May and October).
Recent Extreme Fire Occurrences in Sumatra
Highest Hotspot Cell in Sumatra
Extreme fire occurrences were observed in both N. and S.
Sumatra in 2005 and 2006 respectively. The cells with the top
12 highest hotspot densities (>400 hotspots/yr.) in Figure 3
were mainly due to the occurrence of these extreme fires. The
top four highest hotspot cells from H1 to H4 are located near
Figure 7.
Recent trends in average seasonal fire occurrence in S. Sumatra.
Table 1.
Comparison of the highest hotspot cells in Sumatra and Kalimantan.
Rank Ave.* Max. SD (σ) for Max.Year for Max. Region
H1 941 4494 2.94 2005 Dumai 4p (east)
H2 686 2385 2.77 2005 Dumai (northwest)
H3 624 2231 2.85 2005 Dumai4p
H4 483 772 1.35 2006 Dumai4p (south)
Sumatra
H5 477 2966 2.95 2006 Palembang (east)
H1k 971 2417 1.76 2006 MRP (C south)
H2k 804 2289 1.80 2009 MRP (C north)
H3k 678 1846 1.65 2002 MRP (C middle)
H4k 629 629 1.45 2006 MRP (B&D)
H5k 561 1443 1.51 2002 MRP (A)
Kalimantan
H7k 506 2298 2.40 2006 Sampit
*
Ave.: 11-year (2002-2012) for Sumatra, 10-year (2002-2011) for Kalimantan.

N. YULIANTI ET AL.
Figure 8.
Fire occurrence and accumulated precipitation in 2005.
matra are found in the various curves of hotspots and precipita-
tion. In the figure, N. Sumatra, fire occurrence in three regions,
Dumai + 14p, P. Baru + 12p, and north others (79), are ar-
ranged from bottom to top using a variety of lines. The number
of hotspots in Dumai 4p is shown independently by a dotted
line. The difference between Dumai + 14p and Dumai 4p shows
only the fire occurrence in the 10 grid cells in Dumai + 14p,
and not that in Dumai 4p or “Dumai + 10p”. Average seasonal
fire occurrence over 10-years is also shown with a dotted line to
highlight the severe fire occurrence in 2005. Two accumulated
curves for precipitation using two solid lines of different colors,
and using daily data for P. Baru and Medan, are also included
so as to evaluate the severity of the drought period.
From Figure 8, it is evident that the extreme fires of 2005 in
N. Sumatra occurred in different regions and in different sea-
sons. The most distinctive feature is that the 2005 extreme fires
occurred during the dry seasons. The first extreme fires in Du-
mai 4p (a part of Dumai + 14p) and P. Baru + 12p occurred
during a drought in the winter dry season WD or between
mid-January and mid-March. The value of more than 100 hot-
spots/day was considerably higher than that of 50 hotspots/day
(the 10-year average value for the whole of N. Sumatra). The
following extreme fires occurred in late June in Dumai + 14p
(mostly in Dumai north in Table 1). The third extreme fire peak
was mostly due to fires in Dumai north and North Others (79),
and occurred during early and mid-August.
The extreme fires could therefore be explained by the
drought conditions (using the accumulated precipitation curves
in Figure 8) or by the flat part of the lines for both Medan (~2
mm/day) and Pekan Baru (~1 mm/day). The drought conditions
in 2005 are likely to have been caused by the active winter
boreal monsoon occurring under El Niño Modoki or quasi-El
Niño conditions (Ashok et al., 2007). This abrupt, wide area
drought affected the north of Southeast Asia, and caused a ris-
ing by about +0.5 of the average EMI (El Niño Modoki Index)
values between late 2004 and early 2005, as recorded by
JAMSTEC (2013).
Fires in the Dumai region, (except in Dumai4), showed an-
other two peaks in late June and in early and mid August. These
fires can be explained by the short, a devastating summer
drought (SD) in Medan occurring in June and August, giving
the lowest precipitation rate over the past 12-years. Corre-
spondingly, fires in North Others (mostly areas of non-peatland)
became very active in late June and early August (see Figure 8)
under drought during SD.
The Extreme Seasonal Fire Occurrence and Drought in
S. Sumatra (2006)
In Figure 9, the seasonal fire occurrence and drought in S.
Sumatra are found using the various curves for hotspots and
precipitation. In S. Sumatra, fire occurrence in three regions,
Palembang + 16p, Jambi + 7p, and south others (72), are dis-
played from bottom to top using a variety of lines (in Figure 9).
The number of hotspots in Palembang 6p is independently
shown by a dotted line. The difference between Palembang +
16p and Palembang 6p shows only the fire occurrence within
the 10 grid cells in Palembang + 16p, and not that in Palembang
6p or “Palembang + 10p”. In Figure 9, the average seasonal
fire occurrence over 10-years is also shown with a dotted line to
illustrate the severe fire occurrence in 2006. An accumulated
curve for precipitation using the daily data for Palembang was
also added to evaluate the severity of the drought period (using
solid lines).
From Figure 9, we can see that the 2006 extreme fires in S.
Sumatra occurred in different regions and in different seasons.
Of particular note, however, is that the 2006 extreme fires in S.
Sumatra also occurred during the dry season. The extreme fires
in Palembang 6p (a part of Palembang + 16p) and Jambi + 7p
occurred either during a drought in summer dry season SD or
from mid September to mid October. Their numbers (more than
200 hotspots/day) are considerably higher than that of the 100
hotspots/day of the 10-year average values for the whole of S.
Sumatra. A catastrophic fire (about 700 hotspots/day), can be
ascertained by one sharp peak in early October (twice as large
as in Sampit). The worst fires in the Palembang and Jambi re-
gion coincided with the fire peak in the MRP area. About 60%
of these fires in early October occurred in Palembang 6p
(mostly in eastern Palembang in Table 1). Fire occurrences on
non-peatland (or in South Others) were in early July and Au-
gust, and we are thus able to refer to these fires as “warning
fires” for the areas of peatland in S. Sumatra.
The extreme fires occurred in Palembang, under a devastat-
ing long-term drought involving a period of no rain lasting
about 3-months. About 3-month prior to this drought, a pre-
drought with very low rainfall (3.6 mm/day or the same rate as
the annual rate in a dry season) was observed (see Figure 9).
These conditions are likely to be related to the El Niño event
with ONI (Ocean Niño index) values in NDJ of about +1, as
discussed in relation to Kalimantan (Yulianti & Hayasaka,
2013).
Figure 9.
Fire occurrence and accumulated precipitation in S. Sumatra (2006).

N. YULIANTI ET AL.
Conclusion
Analysis results using MODIS hotspot data of the most re-
cent 11-year period, clearly show trends of seasonal and spatial
fire occurrence in Sumatra. The recent trends in fire occurrence
are:
1. Two extreme fires occurred both in N. and S. Sumatra in
2005 and 2006 respectively under enhanced drought or
rainless conditions related to two different types of El Niño
events.
2. The highest hotspot cell found near east of Dumai in N.
Sumatra contained an annual maximum number of hotspots
of 4494 in 2005. The 11-year average number is 941/yr.,
and this average number is slightly smaller than the 971
hotspots observed in one of cells in the MRP region (Block
C south) in Kalimantan (where a maximum number of 2417
was observed in 2006). However, the annual maximum
number of 4494 within the Dumai cell is the highest re-
corded.
3. The fifth highest hotspot cell found near eastern Palembang
in S. Sumatra had a maximum number of 2966 hotspots in
2006, when the average number was 477/yr. This maximum
number was also higher than that of 2417 in the cell in the
MRP referred to above.
4. Extreme fire occurrences in N. Sumatra in 2005 and in S.
Sumatra in 2006 could be partially explained by an en-
hanced drought occurrence due to El Niño events. But their
relatively high standard deviation (σ) values of 2.7σ in N.
Sumatra, and 2.6σ in S. Sumatra (higher than 1.8σ in Kali-
mantan in 2006 which was the worst recent fire year for
Kalimantan), suggest that both extreme fire occurrences
could be classified as accidental fires. The origin of extreme
fires could be from intentional fires related to practices such
as land clearing and plantation development.
5. The recent fire occurrence in Sumatra was not as intense as
in Kalimantan (the annual mean number of fires was about
18,000 in the whole of Sumatra and smaller than that of
23,000 in Kalimantan). Nevertheless, the peatland area in
Sumatra (about 70,000 km2) is larger than that of Kaliman-
tan (about 60,000 km2).
6. With the exception of data from the extreme fire years,
most of the annual fire occurrences in the recent 10-years
lay between +1σ and −1σ, except for 2010 in S. Sumatra.
7. The results of spatial analysis using 0.5˚ × 0.5˚ cells for N.
Sumatra, clearly show one extreme fire occurrence cell
(centered at N 1.75˚ and E 101.75˚) in the southeast of Du-
mai (one of the cells in Dumai + 4). Within this cell, the
11-year average number of fires was 944 hotspots/year.
This result was much higher than the result of the second
highest cell, at 686 hotspots/year (centered at N 2.25˚ and E
100.25˚) in the northeast of Dumai (one of the cells in Du-
mai + 14).
8. In 2005, (an extreme fire year in N. Sumatra), a few clusters
of hotspots were found in the southeast of Dumai (centered
at around N 1.55˚ and E 101.7˚, 30 km and 115 deg. from
Dumai, N 1.7˚ and E 101.5˚). From these hotspot clusters,
the burnt area in the southeast of Dumai was roughly esti-
mated. During January to March (DN = 1 - 90) in 2005, the
burnt area of these few clusters was estimated to be more
than 700 km2.
9. In late June 2005 (DN = 170 - 180) and in August 2005
(DN = 210 - 230), fires in the northwest of Dumai (centered
at around N 2.0˚ and E 100.4˚, 120 km and 290 deg. from
Dumai) also became active. The burnt area of these few
clusters was estimated to be about 400 km2.
10. In 2006, (an extreme fire year for S. Sumatra), a few clus-
ters of hotspots were also found in eastern Palembang (cen-
tered at around S 2.85˚ and E 105.43˚, 76 km and 78 deg
from Palembang, S 3˚ and E 104.8˚). The total burned area
of these hotspots clusters was estimated to be about 500
km2.
11. The above-mentioned recent fire occurrences in Sumatra
were partially explained by using various precipitation pat-
terns and dry season periods for both N. and S. Sumatra.
Acknowledgements
This research was partially supported by the JST-JICA Sci-
ence and Technology Research Partnership for Sustainable De-
velopment (SATREPS) project on “Wild Fire and Carbon Man-
agement in Peat-Forests in Indonesia.”
REFERENCES
Aldrian, E., & Susanto, R. D. (2003). Identification of three dominant
rainfall regions within Indonesia and their relationship to sea surface
temperature. International Journal Clim a to log y, 23, 1435-1452.
http://dx.doi.org/10.1002/joc.950
Ashok, K., Behera, S. K., Rao, S. A., Weng, H., & Yamagata, T. (2007).
El Niño Modoki and its possible teleconnection. Journal of Geo-
physical Research, 112, C11007.
http://dx.doi.org/10.1029/2006JC003798
Badan Pusat Statistik (2010). Trends of the selected socio-economic
indicators of Indonesia. Jakarta: BPS-Statistic Indonesia.
Chang, P., Wang, Z., McBride, J., & Liu, C. H. (2005).Annual cycle of
Southeast Asia-Maritime Continent rainfall and the asymmetric mon-
soon transition. Journal Climate, 18, 287-301.
http://dx.doi.org/10.1175/JCLI-3257.1
Dennis, R. (1999). A review of fire projects in Indonesia (1982-1998).
Jakarta, Indonesia: Center for International Forestry Research (CI-
FOR).
Fearside, P. M. (1997). Transmigration in Indonesia: Lessons from its
environmental and social impacts. Environmental Management, 21,
553-570. http://dx.doi.org/10.1007/s002679900049
Furukawa, H. (2004). The ecological destruction of coastal peat wet-
lands in Insular Southeast Asia. In H. Furukawa, M. Nishibuchi, Y.
Kono, & Kaida, Y. (Eds.), Ecology destruction, health, and devel-
opment: Advancing Asian paradigms (pp. 31-72). Nagoya: Kyoto
University Press.
Harris, N. L., Brown, S., Hagen, S. C., Saatchi, S. S., Petrova, S., Salas,
W., Hansen, M. C., Potapov, P. V., & Lotsch, A. (2012). Baseline
map of carbon emissions from deforestation in tropical regions. Sci-
ence, 336, 1573-1576. http://dx.doi.org/10.1126/science.1217962
Indonesia Ministry of Forestry (2012). Forest statistics in 2011. Jakarta:
Directorate General of Forestry Plantology.
Japan Agency for Marine-Earth Science and Technology (2013).
MODOKI ENSO: A new phenomenon is found in the Tropical Pa-
cific.
http://www.jamstec.go.jp/frcgc/research/d1/iod/modoki_home.html.en
Jauhiainen, J., Hooijer, A., & Page, S. E. (2012). Carbon dioxide emis-
sions from an Acacia plantation on peatland in Sumatra, Indonesia.
Biogeosciences, 9, 617-630. http://dx.doi.org/10.5194/bg-9-617-2012
Joosten, H., Tapio-Biström, M., & Tol, S. (2012). Peatlands—Guid-
ance for climate change mitigation through conservation, rehabilita-
tion and sustainable use. Rome: The Food and Agriculture Organiza-
tion of the United Nations and Wetlands International.
Langner, A., & Siegert, F. (2009). Spatiotemporal fire occurrence in
Borneo over a period of 10 Years. Global Change Biology, 15, 48-
62. http://dx.doi.org/10.1111/j.1365-2486.2008.01828.x
Margono, B. A., Turubanova, S., Zhuravleva, I., Potapov, P., Tyukav-
136

N. YULIANTI ET AL.
ina, A., Baccini, A., Goetz, S., & Hansen, M. C. (2012). Mapping
and monitoring deforestation and forest degradation in Sumatra (In-
donesia) using Landsat time series data sets from 1990 to 2010. En-
vironmental Research Letters, 7, 034010.
http://dx.doi.org/10.1088/1748-9326/7/3/034010
Miettinen, J., Hooijer, A., Tollenaar, D., Page, S., Malins, C., Vernim-
men, R., Shi, C., & Liew, S. C. (2012). Historical analysis and pro-
jection of oil palm plantation expansion on peatland in Southeast
Asia. Washington DC: International Council on Clean Transportation
(ICCT).
Miettinen, J., Shi, C., & Liew, S. C. (2011a). Deforestation rates in
insular Southeast Asia between 2000 and 2010. Global Change Bi-
ology, 17, 2261-2270.
http://dx.doi.org/10.1111/j.1365-2486.2011.02398.x
Miettinen, J., Shi, C., & Liew. S. C. (2011b). Influence of peat-land and
land cover distribution on fire regimes in insular Southeast Asia. Re-
gional Environmental Change, 11, 191-2011.
http://dx.doi.org/10.1007/s10113-010-0131-7
Oldeman, L. R., Darwis, S. N., & Las, I. (1979). An agro-climatic map
of Sumatra. Bogor: Central Research Institute for Agriculture.
Putra, E. I., & Hayasaka, H. (2011). The effect of the precipitation
pattern of the dry season on peat fire occurrence in the Mega Rice
Project area, Central Kalimantan, Indonesia. Tropics, 19, 145-156.
http://dx.doi.org/10.3759/tropics.19.145
Saharjo, B. H. (2006). Fire behavior in Pelalawan peatland, Riau Prov-
ince. Biodiversitas, 7, 90-93.
Siegert, F., Ruecker, G., Hinrichs, A., & Hoffmann, A. A. (2001). In-
creased damage from fires in logged forests during droughts caused
by El Nino. Nature, 414, 437-440.
http://dx.doi.org/10.1038/35106547
Stolle, F., Chomitz, K. M., Lambin, E. F., & Tomich, T. P. (2003).
Land use and vegetation fires in Jambi Province, Sumatra, Indonesia.
Forest Ecology and Management, 179, 277-292.
http://dx.doi.org/10.1016/S0378-1127(02)00547-9
Tacconi, L. (2003). Fire in Indonesia: Causes, costs and policy impli-
cation. Bogor: Center for International Forestry Research (CIFOR).
Takahashi, H., Usup, A., Hayasaka, H., & Limin, S. H. (2007). Over-
view of hydrological aspects for recent 10 years in the basins of
River Sebangau and Kahayan. Environmental Conservation and
Land Use Management of Wetland Ecosystem in Southeast Asia.
Annual Report for April 2006-March 2007.
Uryu, Y., Mott, C., Foedad, N., Yulianto, K., Budiman, A., Setiabudi,
Takakai, F., Nursamsu, S., Purastuti, E., Fadhli, N., Hutajulu, C. M.
B, Jaenicke, J., Hatano, R., Siegert, F., & Stuwe, M. (2008). Defor-
estation, forest degradation, biodiversity loss and CO2 emissions in
Riau, Sumatra, Indonesia. Jakarta: World Wide Fund (WWF) Indo-
nesia.
Usup, A., Hashimoto, Y., Takahashi, H., & Hayasaka, H. (2004). Com-
bustion and thermal characteristics of peat fire in tropical peatland in
Central Kalimantan. Tropics, 14, 1-19.
http://dx.doi.org/10.3759/tropics.14.1
Vadrevu, K. P., & Justice, C. O. (2011). Vegetation fires in the Asian
region: Satellite observational needs and priorities. Global Environ-
mental Research, 15, 65-76.
Wetlands (2003). Maps of area of peatlands distribution and carbon
content in Sumatra. Bogor: Wetlands International-Indonesia Pro-
gramme.
Yulianti, N., & Hayasaka, H. (2013). Recent active fire under El Niño
conditions in Kalimantan, Indonesia. American Journal of Plants
Science, 4, 685-696. http://dx.doi.org/10.4236/ajps.2013.43A087
Yulianti, N., Hayasaka, H., & Usup, A. (2012). Recent forest and peat
fire trends in Indonesia, the latest decade by MODIS hotspot data.
Global Environmental Research, 16, 105-116.