GIs-Based Method in Developing Wildfire Risk Model: A Case Study in Sasamba, East Kalimantan, Indonesia
THESIS
GIs-BASED METHOD IN DEVELOPING
WILDFIRE RISK MODEL:
A CASE STUDY IN SASAMBA, EAST KALIMANTAN,
INDONESIA
Jaruntorn Boonyanuphap
99?39/MI~
Graduate Program
Bogor Agricultural University
2001
Thesis
GIs-BASED METHOD IN DEVELOPING
WILDFIRE RISK MODEL:
A CASE STUDY IN SASAMBA, EAST KALIMANTAN, INDONESIA
Jaruntom Boonyanup hap
997391MIT
A thesis
Submitted to
Graduate School of Bogor Agricultural University, Indonesia
In Fulfill of the requirement for the degree of
Master of Science in Information Technology for
Natural Resources Management
Graduate Program
Bogor Agricultural University
July, 200 1
: GIs-Based Method in Developing Wildfire Risk Model:
Thesis Title
A Case Study in Sasamba, East Kalimantan, Indonesia
Student Name : Jaruntorn Boonyanuphap
Student ID
: 99739
Study Program : Master of Science in Information Technology for
Natural Resources Management
This thesis proposal is approved by the Advisory Board:
Prof. Dr. Ir. F. Gunaman Suratmo, MF
Supervisor
Dr. Ir. Neneah Surati Java, MAP^
Co-supervisor
Chairman of Study Program
vDr. Ir. Handoko, M.Sc
Date: July 30,2001
Dr.-Ing. Fahmi Arnbar
Abstract
GIs-Based Method In Developing Wildfire h s k Model:
A Case Study In Sasamba, East Kalimantan, Indonesia
by
Jaruntorn Boonyanuphap
Master of Science in Information Technology for Natural Resources Management
Bogor Agricultural University, Indonesia
July 200 1
The modem CIS technology is capable of assigning and weighing the vulnerability value
of each wildfire risk factor in order to develop the wildfire risk model and map the wildfire risk
class. The study area is located in SASAMBA, East Kalin~antanprovince, Indonesia. Physicalenvironmental and human activity factors were the two major groups of wildfire risk factors: the
physical-environmental factors consist of the average daily rnaxin~umtemperatures, the total daily
rainfall, the average daily 1300 relative humidity, and the average daily maximum wind speed,
agro-climatic zone, slope, and aspect. The village center, road network, river network, and
vegetation and land cover types were considered as the human activity factors.
The results showed that all factors had a relationship to the probability of fire starting.
Only the average of daily maximum wind speed, aspect, and river network did not influence the
start of fire in the study area. The relative humidity factor and the location of the village center
were the most important of all physical-environmental and human activity factors for starting
wildfire, respectively. Because small agncultural fires caused most of the wildfires in the study
area from fanners or shifting cultivators, it was evident that the human activities had more
influence than physical-environmental factors for a wildfire to start.
The range of wildfire vulnerability value was classified into three severity classes of
wildfire risk zone (low, moderate and high). The high wildfire risk class covered the largest area
while the low wildfire risk class was the smallest area. Additionally, the calculation of the
coincided area between each class of the wildfire risk and the actual burnt area in 1997198 was
used to assess the accuracy of the wildfire risk model. The high wildfire risk class had the highest
coincided value at 76.05 %, which was the value used to represent the accuracy of the model.
This research expands the basic function of GIs technology integrated with
environmental modeling to model the wildfire risk zone, which provide the information of the
area at risk to wildfire in terms of different severity levels. This map could be an effective early
-warning system for developing wildfire prevention activities in the study area.
Summary
Jaruntorn Boonyanuphap
Master of Science in Information Technology for Natural Resources Management
Bogor Agricultural University, Indonesia
Supenlisor: Prof. Dr. Ir. F. Gunarwan Suratmo, MF
Co-supervisor: Dr. I Nengah Surati Jaya. MAgr
: Dr.-Ing. Fahmi Amhar
Forest and land fire is not a new phenomenon in Indonesia. It has been reported
that within the past two decades, the largest tropical wildfire episode ever recorded in
Indonesia occurred in 1982/1983, where an estimated 3.7 million hectares of land was
burned in East Kalimantan. Other serious fires have occurred in 1986, 1991, 1994, and a
prolonged and extremely severe fire season occurred during the last ENS0 (El-NifioSouthern Oscillation) of 1997198 in Southeast Asia. Therefore, the prevention of fire
should be considered as a high priority because it is key to solving the fire problem. The
fire risk map is an alternative information showing locations where activities and
improvements might increase the chance of a fire to start. Likewise, the more effective
fire prevention, an understanding of the causes of fire, the risk factors and interactions
among them is necessary.
The study area in this research represents approximately 249,532 hectares or 79
percent of the Integrated Economic Development Zone (KAPET) in SASAMBA region.
It is situated in equatorial area between 1 16' 43' 30" and 1 17" 18' 30" East Longitude
and -0" 33' 30" and -lo 13' 30" South Latitude. Approximately 219,487 hectares or
87.96 % of the study area were affected by fire, which showed that the effort to minimize
wildfires in the study area had been low due to lack of institutional capacity, personnel
and guidelines. It is highly important and necessary to have an effective and eficient
wildfire prevention because the El Nifio drought event may occur more frequently than in
the past, creating conditions that could trigger even more fires in the future.
The overall objective of this study is to develop a wildfire risk model through the
study of the spatial dimensions of interacting factors associated with the likelihood of
wildfires using GIS application. Furthermore, the specific purpose is to assign and
analyze the Physical-environmental and human activity factors that are associated with
the location of wildfire starting in the study area, and map the severity class of the
wildfire risk
The wildfire risk factors were grouped into two major groups: Physicalenvironmental and Human Activity factors. The physical-environmental factors consist of
the average daily maximum temperatures, the total daily rainfall, the average daily 1300
relative humidity, the average daily maximum wind speed, agro-climatic zone, slope, and
aspect. The human activity factors consist of village location, road network, river
network, and vegetation and land cover types.
The location of NOAA-AVI1RRIIot Spot in 1997198 was overlaid with all of the
wildfire risk factors in order to analyze the spatial Hot Spots pattern, which showed that,
except for the average of daily maximum wind speed, aspect and distance to river
network, all of the factors had reasonable relationship to the probability of fire starting.
The result of wildfire vulnerability value revealed that the relative humidity factor was
the highest important of all physical-environmental factors in the start of wildfires.
For the human activity factors, the location of village center was the most
important, followed by the road network and the vegetation and land cover types, which
had the same influence to causing a wildfire to start. Likewise, it was evident that the
human activity factors had stronger influence in causing a wildfire than physicalenvironmental factors.
The relationship of hot spots to some factors gave illogical result. These factors
include the average daily maximum wind speed, aspect, and river network. That is the
weakness of GIs-based model in comparison with the system analysis model.
Nevertheless, GIS has the advantage of manipulating, analyzing, and representing the
spatial dataset.
The composite wildfire vulnerability value was grouped into three severity classes
of wildfire risk zones. The low wildfire risk zone was 30,709 hectares or 12.3 1% of the
study area. This zone was mostly affected by physical-environmental factors, which
covered most of the mangrove forest and the protected area. The moderate wildfire risk
zone covered 59,422 hectares or 23.81% of the study area, which was mostly
disseminated in the lowland dipterocarp forest. The high wildfire risk zone covered the
largest area of 159,401 hectares or 63.88% of the study area. Open land, Alang-Alang,
bush land, secondary forest, and plantation forcst were the main land cover types of this
zone. This zone also evidently showed that the area of human activities has a very high
potential to start wildfire. Moreover, the weather condition of this zone evidently
contributed to wildfire during the abnormally long drought associated with the El-NiRo
phenomenon.
The calculation of the coincided area between each class of the wildfire risk and
the actual burnt area in 1997198 was used to assess the accuracy of the wildfire risk
model. It gave the coincided values of 76.05 %, 34.24 %, and 22.12 %, for the high,
moderate, and low wildfire risk classes, respectively. Another technique of modcl
assessment was the calculation of simple relationship between the actual burnt area and
the wildfire risk model. The area of high wildfire risk had the highest percentage, when
compared with the others wildfire risk classes.
This research report i s a requirement for the degree Master of Science in
Information Technology for Natural Resources Management at Bogor Agricultural
University. This research concluded a study on developing a wildfire risk rnodel using
GIs technology. Concerning the Indonesia Great Fire in 1997198, where approximately
87.96 percent of the study area was affected by fire. It evidently indicates that the efforl to
minimize wildfires had been low due to lack of institutional capacity, personnel and
guidelines. Therefore, this research concerns the integration of geographic information
system (GIs) and environmental modeling for developing a wildfire risk model. The
objective of this study was to assign and analyze the physical-environmental and human
activity factors that area associated to location of wildfire starting as well as to develops
the model of wildfire risk in term of the severity class of wildfire risk, which is concerned
with the probability of occurrence in future fire.
July, 2001
Jaruntom Boonyanuphap
Acknowledgment
First and foremost, I wish to express my sincere appreciation and gratitude to my
supervisor, Prof. Dr. Ir. F. Gunanvan Suratmo, and my co-supervisors, Dr. Nengah Surati
Jaya and Dr. Fahmi Amhar for their meaningful support and generous time.
My entire study in Bogor Agricultural University was made possible with the
financial support given by the SEAMEO Development Education Fund (SEDF). It is
gratefully acknowledged. I would also like to extend my special thanks to Dr. Handoko
for giving me the opportunity to study in the MIT program.
I would like to gratefully acknowledge the Integrated Forest Fire Management
(IFFM) project, GTZ Samarinda Office for providing the weather and fire occurrence
data. I am also indebted to Drs. A. Ari Dartoyo and Ir. Sumaryono, Center for Survey of
Natural Resources (PUSSUSDA), BAKOSU RTANAL, for providing the thematic vector
data as well as Mr. Guswanto, BPP Teknologi, for the weather data.
Special thanks to Mr. Redhahari, the great person from Mulawarman University,
who gave me the opportunity to meet so many nice people and have the chance to build
up a great friendship. This is include, Ms. Lenny Christy, junior expert of IFFM Project,
who kindly gave me a valuable discussion and wonderful time during my visit in
Samarinda.
Special thanks to Rona Dennis, the remote sensing and GIs analyst of the
Center for International Forestry Research (CIFOR), as well as Pak Johnnie Hadi
Prakoso, early warning and detection system specialist of the Forest Fire Prevention
Management Project (JICA), for sharing their valuable forest fire knowledge.
I am also especially grateful to Ms. Arum Rahayu Nusawana Azhar, Faculty of
Forestry, Bogor Agricultural University (IPB), for Bahasa Indonesia-English translation
of this report, and the other invaluable support during my research time.
1 wish to acknowledge and thank all my instructors who gave the very important
information technology knowledge. I would also like to thank my classmates and all MIT
students, who were all hard working, pleasure of working, with gave me a happy time to
treasurc! the time we spent together as well as our friendship through these years. A lots of
thank, of course, extended to all my friends who helped out when the pressure become
too much.
To Arporn Khunnachak, who has always been by my side and make me feel
comfortable wherever I am. Especially, you have been able to bring out the best in me.
Finally, the biggest thanks go to my parents, Chin and Fongchan, my sister and
brother, Napattaraporn and Chanon, for their support and encouragement through the
good times - words cannot fully express "How much I Iove you"
GIs-BASED METHOD IN DEVELOPING
WILDFIRE RISK MODEL:
A CASE STUDY IN SASAMBA, EAST KALIMANTAN,
INDONESIA
Jaruntorn Boonyanuphap
99?39/MI~
Graduate Program
Bogor Agricultural University
2001
Thesis
GIs-BASED METHOD IN DEVELOPING
WILDFIRE RISK MODEL:
A CASE STUDY IN SASAMBA, EAST KALIMANTAN, INDONESIA
Jaruntom Boonyanup hap
997391MIT
A thesis
Submitted to
Graduate School of Bogor Agricultural University, Indonesia
In Fulfill of the requirement for the degree of
Master of Science in Information Technology for
Natural Resources Management
Graduate Program
Bogor Agricultural University
July, 200 1
: GIs-Based Method in Developing Wildfire Risk Model:
Thesis Title
A Case Study in Sasamba, East Kalimantan, Indonesia
Student Name : Jaruntorn Boonyanuphap
Student ID
: 99739
Study Program : Master of Science in Information Technology for
Natural Resources Management
This thesis proposal is approved by the Advisory Board:
Prof. Dr. Ir. F. Gunaman Suratmo, MF
Supervisor
Dr. Ir. Neneah Surati Java, MAP^
Co-supervisor
Chairman of Study Program
vDr. Ir. Handoko, M.Sc
Date: July 30,2001
Dr.-Ing. Fahmi Arnbar
Abstract
GIs-Based Method In Developing Wildfire h s k Model:
A Case Study In Sasamba, East Kalimantan, Indonesia
by
Jaruntorn Boonyanuphap
Master of Science in Information Technology for Natural Resources Management
Bogor Agricultural University, Indonesia
July 200 1
The modem CIS technology is capable of assigning and weighing the vulnerability value
of each wildfire risk factor in order to develop the wildfire risk model and map the wildfire risk
class. The study area is located in SASAMBA, East Kalin~antanprovince, Indonesia. Physicalenvironmental and human activity factors were the two major groups of wildfire risk factors: the
physical-environmental factors consist of the average daily rnaxin~umtemperatures, the total daily
rainfall, the average daily 1300 relative humidity, and the average daily maximum wind speed,
agro-climatic zone, slope, and aspect. The village center, road network, river network, and
vegetation and land cover types were considered as the human activity factors.
The results showed that all factors had a relationship to the probability of fire starting.
Only the average of daily maximum wind speed, aspect, and river network did not influence the
start of fire in the study area. The relative humidity factor and the location of the village center
were the most important of all physical-environmental and human activity factors for starting
wildfire, respectively. Because small agncultural fires caused most of the wildfires in the study
area from fanners or shifting cultivators, it was evident that the human activities had more
influence than physical-environmental factors for a wildfire to start.
The range of wildfire vulnerability value was classified into three severity classes of
wildfire risk zone (low, moderate and high). The high wildfire risk class covered the largest area
while the low wildfire risk class was the smallest area. Additionally, the calculation of the
coincided area between each class of the wildfire risk and the actual burnt area in 1997198 was
used to assess the accuracy of the wildfire risk model. The high wildfire risk class had the highest
coincided value at 76.05 %, which was the value used to represent the accuracy of the model.
This research expands the basic function of GIs technology integrated with
environmental modeling to model the wildfire risk zone, which provide the information of the
area at risk to wildfire in terms of different severity levels. This map could be an effective early
-warning system for developing wildfire prevention activities in the study area.
Summary
Jaruntorn Boonyanuphap
Master of Science in Information Technology for Natural Resources Management
Bogor Agricultural University, Indonesia
Supenlisor: Prof. Dr. Ir. F. Gunarwan Suratmo, MF
Co-supervisor: Dr. I Nengah Surati Jaya. MAgr
: Dr.-Ing. Fahmi Amhar
Forest and land fire is not a new phenomenon in Indonesia. It has been reported
that within the past two decades, the largest tropical wildfire episode ever recorded in
Indonesia occurred in 1982/1983, where an estimated 3.7 million hectares of land was
burned in East Kalimantan. Other serious fires have occurred in 1986, 1991, 1994, and a
prolonged and extremely severe fire season occurred during the last ENS0 (El-NifioSouthern Oscillation) of 1997198 in Southeast Asia. Therefore, the prevention of fire
should be considered as a high priority because it is key to solving the fire problem. The
fire risk map is an alternative information showing locations where activities and
improvements might increase the chance of a fire to start. Likewise, the more effective
fire prevention, an understanding of the causes of fire, the risk factors and interactions
among them is necessary.
The study area in this research represents approximately 249,532 hectares or 79
percent of the Integrated Economic Development Zone (KAPET) in SASAMBA region.
It is situated in equatorial area between 1 16' 43' 30" and 1 17" 18' 30" East Longitude
and -0" 33' 30" and -lo 13' 30" South Latitude. Approximately 219,487 hectares or
87.96 % of the study area were affected by fire, which showed that the effort to minimize
wildfires in the study area had been low due to lack of institutional capacity, personnel
and guidelines. It is highly important and necessary to have an effective and eficient
wildfire prevention because the El Nifio drought event may occur more frequently than in
the past, creating conditions that could trigger even more fires in the future.
The overall objective of this study is to develop a wildfire risk model through the
study of the spatial dimensions of interacting factors associated with the likelihood of
wildfires using GIS application. Furthermore, the specific purpose is to assign and
analyze the Physical-environmental and human activity factors that are associated with
the location of wildfire starting in the study area, and map the severity class of the
wildfire risk
The wildfire risk factors were grouped into two major groups: Physicalenvironmental and Human Activity factors. The physical-environmental factors consist of
the average daily maximum temperatures, the total daily rainfall, the average daily 1300
relative humidity, the average daily maximum wind speed, agro-climatic zone, slope, and
aspect. The human activity factors consist of village location, road network, river
network, and vegetation and land cover types.
The location of NOAA-AVI1RRIIot Spot in 1997198 was overlaid with all of the
wildfire risk factors in order to analyze the spatial Hot Spots pattern, which showed that,
except for the average of daily maximum wind speed, aspect and distance to river
network, all of the factors had reasonable relationship to the probability of fire starting.
The result of wildfire vulnerability value revealed that the relative humidity factor was
the highest important of all physical-environmental factors in the start of wildfires.
For the human activity factors, the location of village center was the most
important, followed by the road network and the vegetation and land cover types, which
had the same influence to causing a wildfire to start. Likewise, it was evident that the
human activity factors had stronger influence in causing a wildfire than physicalenvironmental factors.
The relationship of hot spots to some factors gave illogical result. These factors
include the average daily maximum wind speed, aspect, and river network. That is the
weakness of GIs-based model in comparison with the system analysis model.
Nevertheless, GIS has the advantage of manipulating, analyzing, and representing the
spatial dataset.
The composite wildfire vulnerability value was grouped into three severity classes
of wildfire risk zones. The low wildfire risk zone was 30,709 hectares or 12.3 1% of the
study area. This zone was mostly affected by physical-environmental factors, which
covered most of the mangrove forest and the protected area. The moderate wildfire risk
zone covered 59,422 hectares or 23.81% of the study area, which was mostly
disseminated in the lowland dipterocarp forest. The high wildfire risk zone covered the
largest area of 159,401 hectares or 63.88% of the study area. Open land, Alang-Alang,
bush land, secondary forest, and plantation forcst were the main land cover types of this
zone. This zone also evidently showed that the area of human activities has a very high
potential to start wildfire. Moreover, the weather condition of this zone evidently
contributed to wildfire during the abnormally long drought associated with the El-NiRo
phenomenon.
The calculation of the coincided area between each class of the wildfire risk and
the actual burnt area in 1997198 was used to assess the accuracy of the wildfire risk
model. It gave the coincided values of 76.05 %, 34.24 %, and 22.12 %, for the high,
moderate, and low wildfire risk classes, respectively. Another technique of modcl
assessment was the calculation of simple relationship between the actual burnt area and
the wildfire risk model. The area of high wildfire risk had the highest percentage, when
compared with the others wildfire risk classes.
This research report i s a requirement for the degree Master of Science in
Information Technology for Natural Resources Management at Bogor Agricultural
University. This research concluded a study on developing a wildfire risk rnodel using
GIs technology. Concerning the Indonesia Great Fire in 1997198, where approximately
87.96 percent of the study area was affected by fire. It evidently indicates that the efforl to
minimize wildfires had been low due to lack of institutional capacity, personnel and
guidelines. Therefore, this research concerns the integration of geographic information
system (GIs) and environmental modeling for developing a wildfire risk model. The
objective of this study was to assign and analyze the physical-environmental and human
activity factors that area associated to location of wildfire starting as well as to develops
the model of wildfire risk in term of the severity class of wildfire risk, which is concerned
with the probability of occurrence in future fire.
July, 2001
Jaruntom Boonyanuphap
Acknowledgment
First and foremost, I wish to express my sincere appreciation and gratitude to my
supervisor, Prof. Dr. Ir. F. Gunanvan Suratmo, and my co-supervisors, Dr. Nengah Surati
Jaya and Dr. Fahmi Amhar for their meaningful support and generous time.
My entire study in Bogor Agricultural University was made possible with the
financial support given by the SEAMEO Development Education Fund (SEDF). It is
gratefully acknowledged. I would also like to extend my special thanks to Dr. Handoko
for giving me the opportunity to study in the MIT program.
I would like to gratefully acknowledge the Integrated Forest Fire Management
(IFFM) project, GTZ Samarinda Office for providing the weather and fire occurrence
data. I am also indebted to Drs. A. Ari Dartoyo and Ir. Sumaryono, Center for Survey of
Natural Resources (PUSSUSDA), BAKOSU RTANAL, for providing the thematic vector
data as well as Mr. Guswanto, BPP Teknologi, for the weather data.
Special thanks to Mr. Redhahari, the great person from Mulawarman University,
who gave me the opportunity to meet so many nice people and have the chance to build
up a great friendship. This is include, Ms. Lenny Christy, junior expert of IFFM Project,
who kindly gave me a valuable discussion and wonderful time during my visit in
Samarinda.
Special thanks to Rona Dennis, the remote sensing and GIs analyst of the
Center for International Forestry Research (CIFOR), as well as Pak Johnnie Hadi
Prakoso, early warning and detection system specialist of the Forest Fire Prevention
Management Project (JICA), for sharing their valuable forest fire knowledge.
I am also especially grateful to Ms. Arum Rahayu Nusawana Azhar, Faculty of
Forestry, Bogor Agricultural University (IPB), for Bahasa Indonesia-English translation
of this report, and the other invaluable support during my research time.
1 wish to acknowledge and thank all my instructors who gave the very important
information technology knowledge. I would also like to thank my classmates and all MIT
students, who were all hard working, pleasure of working, with gave me a happy time to
treasurc! the time we spent together as well as our friendship through these years. A lots of
thank, of course, extended to all my friends who helped out when the pressure become
too much.
To Arporn Khunnachak, who has always been by my side and make me feel
comfortable wherever I am. Especially, you have been able to bring out the best in me.
Finally, the biggest thanks go to my parents, Chin and Fongchan, my sister and
brother, Napattaraporn and Chanon, for their support and encouragement through the
good times - words cannot fully express "How much I Iove you"