Hazard map assessment of Mount Merapi, Central Java, Indonesia using remote sensing

Author : David Harris, IGES department, Aberystwyth University, Wales

As the global population is predicted to reach 7 billion people by 2012, land pressures and rapid population growth is resulting in many more communities living within danger zones of natural hazards, a pattern particularly seen around volcanoes. This thesis will emphasise the impact of volcanoes on populations using the example of Mount Merapi, Indonesia. Mount Merapi is the most active volcano in Java and has persistent minor eruptions, but according to volcanologists Mount Merapi is heavily overdue a large-scale eruption which could potentially put over 1.1 million people at risk. This thesis begins with a review of published papers and a description of Mount Merapi’s geological history, with a particular focus on its recent eruptions. To better assess Mount Merapi’s threat to the communities that are in close proximity to the summit the thesis uses GIS software to produce a risk map. The risk map is then used as a basis for further analysis on the potential impact in the event of a major eruption. The thesis specifically examines the risk on the basis of current population (e.g. Kemiren, a town with 103,777 people at a risk value site of 10.5) and social vulnerability (e.g. Ngablak, which has a Social Vulnerability Index value of 0.5 – 1.5 and a risk value varying between 7.5 and 28.5). Lastly, the thesis considers the impact of Merapi’s most recent eruptions in October and November 2010 and recommends some considerations for the future in terms of risk reduction by changes in response times and evacuation procedures.

1. Introduction

The world’s population is predicted to exceed 7 billion people by 2012 and is set to increase further in the next decade and beyond (Gilbert 2005). This is leading to increased pressures on land use and is forcing people to live in areas that are within danger zones of natural hazards. The risk associated with such a population rise is greater than ever before and is making more and more people vulnerable to geohazards.

As population rises, the threat rises proportionally. The largest population rises in the 21st Century are occurring in the Lesser Economically Developed Countries (LEDC’s) (i.e. Indonesia and China) which further increases the vulnerability of the population as more and more people live in danger zones with low education and poorly constructed buildings (due to less money being available within the country for education and for infrastructure) (Chester et al 2001). In a country that is already verging at or below the poverty line (e.g. the LEDC’s) the cost impact of hazards generally falls heavily on life loss and economic cost for LEDC’s but in More Economically Developed Countries (MEDC’s) only on economic cost. For example, Donovan (2010) states that between 1991 and 2005 over 90% of deaths resulting from natural hazards occurred in developing countries.

There are several equations to express the concept of risk (Beck 1992; Glade et al 2005; Granger et al 2003) however Blaikie et al (1994)’s equation whereby ‘Risk = Threat x Vulnerability x Cost’ emphasises exactly what this thesis is highlighting – threat, vulnerability and the cost of potential natural disasters. Risk in itself also has a contested definition with confusion residing between the concepts of risk and uncertainty, although Knight (1921) claimed that uncertainty is incalculable, while risk is calculable and hence knowable. With risk now being inferred as a known unit, mitigation against risk becomes possible and therefore also becomes calculable enough for mapping.

The impact of risk and natural hazards, in modern culture, has been subdued or even entirely mitigated in some areas throughout the world for example strengthening buildings against earthquake damage (i.e. the Yokohama Landmark Tower, Yokohama City, Japan) and technological advances on buildings that can position structures on landslide-prone hillsides with deeper foundations (e.g. Pacific Palisades, California, United States Of America). Unfortunately, not all areas on Earth have been protected against all the risks (and some areas which are protected against one hazard may not be protected against another). It is this very theory in which areas that avoid finances and spotlight of the media for awhile become ever more vulnerable to the Earth’s forces, especially that of natural hazards (e.g. Haiti in 2010). The three factors of risk cited by Blaikie et al (1994) namely threat, vulnerability and cost, may vary considerably depending on exact location, so there is no generic solution and every natural hazard must be assessed individually.

Natural hazards come in different forms varying from; floods, droughts, volcanoes, earthquakes, tsunamis, landslides, extreme temperatures and hurricanes. Each hazard has different effects, timings and impacts (in the short term and long term) all depending on the area that was affected (LEDC or MEDC) and how much preparation the area or region had prior to the event. These factors plus the rapidly rising population of the world (especially in LEDC’s) makes natural hazards ever more dangerous year after year. Prime examples include the 7.0 magnitude earthquakes that struck Haiti on 12th January 2010 and Japan on 26th February 2010. The Haiti earthquake that struck 25km WSW from Port-au-Prince killed over 230,000 people and destroyed up to 90% of the buildings in some villages near the epicentre (i.e. Leogane and Jacmel), mainly due to the lack of preparedness that the nation had and also the poverty that the nation was already in. The Japanese earthquake struck offshore of Ryukyu Islands (80km ESE from Okinawa) and injured no-one and no buildings were damaged (USGS 2010).

The specific natural hazard that this thesis will focus on is the impact of volcanoes on populations. It is not the most recent eruption(s) of Eyjafjallajokull, Iceland from early April 2010 to June 2010 and Mt Pacaya, Guatemala on the 27th May 2010 that makes volcanoes so recognisable in the current media but more focused upon the dormant volcano type, the so called ‘sleeping giants’ (Duffield 1997) such as Mount Vesuvius, Italy or Mount St. Helens, United States of America.

While it would appear that these examples provide clear warnings to communities there is still a lack of perception of the risk that face so many people on Earth living near these potentially destructive volcanoes. Around 9% of the world’s population live within 100km of a historically active volcano and around 12% of the world population within 100km of a volcano which is believed to be active within the last 10,000 years (Small and Naumann 2001). The distance from large cities to nearby volcanoes that have been active in the Holocene can be seen in Figure 1.

Figure 1: A selection of large cities plotted according to relative distance to the nearest volcano, with population data. (Chester et al 2001)

For the populations that surround the volcanoes, volcanic activity can be hard to predict. Exact time and magnitude of volcanic activity has a very large variation on a precise measurement or date. This is due to the subterranean aspect of volcanoes and having no exact naked eye measurements, but purely based on scanners and other technologies (if they are available) as well as the varying geology and lava type of volcanoes (i.e. basaltic, andesitic or rhyolitic). This is very different from other natural hazards such as hurricanes or droughts as they can be visually shown either via the naked eye or via infrared technology and although shorter term, they build up via easy to see visual impacts so the magnitude of the event can be assessed. In most cases countries have enough time to make populations in the danger zone(s) to evacuate, such as: Hurricane Katrina in 2005 that struck New Orleans, America which evacuated approximately one million people (Litman 2006).

The damage that can be caused by a volcanic eruption can vary from; pyroclastic flows, lahars, lava, ash, lava bombs, tephra and possible landslides as well (Figure 2). These impacts can destroy buildings, scorch the surrounding land areas and in most cases cause health problems in the long term via the ash in the atmosphere and, ultimately, put lives at risk.

Figure 2: Potential volcanic hazards that could damage or harm the surrounding area (USGS 2010)

The facts are conclusive that populations living in close proximity to volcanoes are living in an area that can potentially cause major damage to their homes and even to their lives. However even though the disadvantages are reasonably clear, there are in fact some particular advantages for living near volcanoes. For example lava can be cut into blocks and used as stone for buildings and fine-grained volcanic ash can be used as a polishing compound (FON 2000). Volcanoes also attract large amounts of tourism, natural beauty and produce highly fertile soils generated by the volcanic minerals that rise to the surface (e.g. Alluvium).

Sometimes it is not about the advantages and disadvantages of settling near volcanoes, but on the reality that there is nowhere else to settle, which is unfortunately the case in many of the rapidly growing nations such as: Indonesia and Japan. This settlement factor plus the lack of education on the science (i.e. past eruption extents) and the risk of volcanoes caused by the poverty of the region creates a vulnerable area (or perhaps even a nation) which is sure to have many more natural disasters rather than just natural hazards (Blaikie et al 1994).

This brings the attention to the mapping of natural hazards. Hazard maps have been around for several decades, originally created by documenting old eruptions and plotting them together to form a sketched map (e.g. Crandell et al 1984; Hewitt 1997; Lavigne 1999 and Naranjo et al 1987). Most of these are now becoming out of date due to more recent eruptions by volcanoes which are generating new figures and different damage extents. Due to this reason some hazard maps have been replaced by technological advances such as remote sensing and Geographical Information Systems (GIS). However, this has only been done with volcanoes that are highly active and/or have high risk sites, such as Xiaojiang Basin, China (He et al. 2003), Mt Popocatepetl, Mexico (F. Goff et al 2001) and Mt. Ruapehu, New Zealand (Joyce et al 2009).

The aim of this thesis was to produce a new hazard map of Mount Merapi, Central Java, Indonesia using data from historic eruptions of the volcano then assessing the relative risk imposed on the area surrounding the summit using Geographic Information Systems (GIS) techniques.

This thesis concentrates on Mount Merapi because it is a volcano that has relatively persistent activity and could potentially put over 1.1 million people at risk especially considering the growing land pressures and global population rise. Also very little journals have concentrated on mapping Mount Merapi’s eruptions from a remote sensing perspective, there has only been sketched risk maps produced 10 or more years ago due to the lack of technology at the time (e.g. Thouret et al 2000 and Voight et al 2000).

The objectives of this thesis are:
·    Assessing the risk of the area by using historic data of lahar flows, pyroclastic flows and ash emissions by inputting the data on to a Geographical Information System (GIS).
·    Analyse these risk areas in association with Google Earth imagery and collected population data.

This thesis is organised into the following chapters:
·    Chapter 2 provides an overview of Mount Merapi including area of interest, geological record and any discrepancies there in and a background on remote sensing.
·    Chapter 3 covers an introduction of Mount Merapi, emphasising on local geography and social context. Also looking into Mount Merapi’s activity – recent and old, the populations at risk and monitoring strategies that are currently in place.
·    Chapter 4 covers the methods behind the creation of the hazard risk map via GIS software.
·    Chapter 5 shows the results produced from chapter 4 and possible sources of error.
·    Chapter 6 includes the discussion and analysis of the hazards posed on the flanks of Merapi, taking into consideration certain villages and towns including the populations at risk and also looking into social vulnerability of the surrounding areas.
·    Chapter 7 summarizes the thesis, including the limitations of the final risk map and considerations for the future.

2. Background

2.1 Overview
This chapter begins with a broad descriptive background of Mount Merapi, covering its location within Indonesia, and then moving onto the volcano’s geological record, greatly considering the work of Newhall et al (2000), Berthommier et al (1990,1992) and Camus et al (2000) and the discrepancies between them. The chapter then concentrates on remote sensing technology and its relationship with geological records, volcanic activity and monitoring, and finishes on some critical engagement with the work mentioned already.

2.2 Area of Interest
The area of interest for this thesis is located around the summit of Mount Merapi which is situated in Central Java, Indonesia (Figure 3).

Figure 3: Map of Indonesia via Google Earth© with Mount Merapi pinpointed

Mount Merapi is located at 7º32’26’’S and 110º26’48’’E, the summit is 2,950m above sea level. The volcano’s most recent eruption(s) were on 26th October 2010 and 3rd November 2010, with the last major eruption (that caused a large death toll) in November 1994. Merapi has a varied chronologic and geologic record which is mainly due to its relatively persistent activity (the most active volcano in Java). Mount Merapi has also had an influence on culture, population as well as religion in Central Java throughout its history. Merapi exhumes ash and steam throughout the year much like Plate 1 shows:

Plate 1: Mount Merapi taken in October 2010, near to the 26th October eruption (BBC 2010)

2.3 Brief Geological Record
Studies on the geological record of Merapi are not as comprehensive as most volcanoes, which is most unusual, considering its relatively active volcanology. The geological record could be different depending on which published paper you look at: papers by Berthommier (1990), Berthommier et al (1992) and Camus et al (2000) suggest that the record of Mt. Merapi had four set stages of growth:
·    ‘Ancient Merapi’ (40,000 to 14,000BP)
·    ‘Middle Merapi’ (14,000 to 2,200BP)
·    ‘Recent Merapi’ (2,200 BP to 1786AD)
·    ‘Modern Merapi’ (1786AD to present)

Whereas Newhall et al (2000) suggest that Merapi is built in three stages:
·    ‘Proto-Merapi’ (before 5,000BC)
·    ‘Old Merapi’ (5,000BC to 0AD)
·    ‘New Merapi’ (0AD to present)

The main differences between the two groups of scientists appear to be through different interpretations of Mount Merapi’s growth period, and blast deposit and flank failure occurrences.
With regards to growth periods, Camus et al (2000) and Berthommier et al (1992) include the Plawangan and Turgo hills to ‘Ancient Merapi’ where as Newhall et al (2000) suggest the hills are relics of ‘Proto-Merapi’.

In regards to Blast deposits and flank failures of Mount Merapi, Camus et al (2000) and Berthommier et al (1992) suggest a blast deposit and flank failure to be part of an eruption dating between 6,600 and 2,200 years ago, whereas Newhall et al (2000) consider the blast deposit and flank failure occurred between 1,600 and 1,100 years ago. However; it is clear that an eruption of at least partial flank failure did occur at least once in the past 6,700 years.

Furthermore, while the scientists differ on the extent of blast deposits it is clear that a series of deposits record ‘Recent Merapi’ growth such as: ash and scoria, pyroclastic flow deposits and thick plinian tephra fall deposits that coat an area in excess of 800km². Additionally, pyroclastic surge deposits are possibly related to phreatomagmatic eruptions that left Gumuk ash (2,200 – 1,470BP) and Sambisari ash (600 – 470BP) as far as 30km from the summit. Camus et al (2000) and Berthommier et al (1992) argue that Sambisari ash and 8 metre thick lahar deposits extend 30km onto the Yogyakarta plain, which buried the Sambisari temple. However, Newhall et al (2000) believe that large explosive eruptions followed shortly after the ‘Old Merapi’ collapse, basing it upon the occurrence of pyroclastic flows to the south and west on the Yogyakarta plain and in the Kaliurang vicinity (around 25km north of Yogyakarta). Newhall et al (2000) assume that large explosive eruptions followed the large culture change in 928AD in Java and may have led to the decentralization of the Mataram civilization (a Hindu-Buddhist Javanese civilization between the 8th and 10th Century), but this decentralization is strongly contested by Berthommier et al (1992) who point out that Newhall et al (2000)’s assumption is based upon very little direct evidence.

Finally, the duration of the eruptive episodes versus quiet periods are also contested between the scientists at Merapi. Mount Merapi has been active over the past two centuries and so the distinctions of long lasting periods versus slow rate periods, such as dome growth or gravity driven destruction (which is the most common activity on Merapi), and also the times between no eruptions and explosive outbursts are hard to tell as one eruption covers another eruption due to the constant activity.

Although Newhall et al (2000) do pose some uncertainties within the geological record the theory behind the evolution of Merapi is coherent. Three main areas of interest within the last 10,000 years of Merapi’s geological record were identified:
·    Around 700AD to 900AD many Buddhist and Hindu temples were being constructed in Central Java. Eruptions of Merapi occurred before, during and after construction of these temples and many were buried during or soon after construction. Newhall et al (2000) suspect that the destruction of these temples led to (or most likely contributed to) a shift of power from Central Java to East Java in 928AD. The temples that were left were soon abandoned and later occupied by “caretakers” for several centuries.
·    Newhall et al (2000) speculate that the eruptions that occurred 700 to 800 years ago were triggered by a partial collapse of New Merapi and that these eruptions ended or most likely aided the end of the “caretaker” occupation at Candi Sambisari and Candi Kedulan settlements.
·    Comparing the last 10,000 years to recent activity, Newhall et al (2000) believe that the 20th Century has had relatively benign lava-dome extrusions and dome-collapse pyroclastic flows.

The last statement concerning ‘benign’ activity in the 20th Century is rather worrying, especially considering the eruptions in 1930-31, 1969, 1994 and 2010 which collectively caused around 1,700 deaths. If these, according to Newhall et al (2000) are ‘benign’, the population of Java should be very wary of the dangers that are imposed by Mount Merapi. Newhall et al (2000) suggest that eruptions could sweep through and beyond the ‘Forbidden Zone’ and even through the ‘First Danger zone’ (Figure 6), and there is no reliable method, at present, to anticipate whether or when Merapi will interrupt its relatively benign activity of the 20th Century with a larger explosive event (Newhall et al 2000). Even though, the recent eruptions of 26th October through to the 9th November 2010 killed around 200 people, the eruptions generated a large series of pyroclastic flows that typically occur on average every 8 to 15 years and Newhall et al (2000) are trying to emphasis the lack of lava dome extrusion and flank failure that has occurred in the 20th Century which can cause much greater damage.

2.4 Remote Sensing and the Geological Record
The emergence of satellite remote sensing in the last decade has provided a much more systematic and synoptic framework for scientific knowledge of the Earth, which in turn improves measurements with numerical modelling; which enhances understanding of where and when a natural hazard occurs and therefore resulting in either reducing or observing socio-economic impact caused (Tralli et al 2005). Unfortunately, due to the sporadic and indeterminable timings of volcanic hazards, measurements are usually few and far between. However, with the break-through in remote sensing, observations can be constant, aiding and observing changes, in this case volcanic alterations such as thermal imagery and gas emissions. Some measurement strategies that incorporate these factors are:

Table 1: Examples of remote sensing measurement strategies on volcanoes

All of the above remote sensing technologies can and have contributed to volcanic risk assessment, mitigation and responses within the last several years (Tralli et al 2005), such as in the recent eruption(s) of Eyjafjallajokull and the Haiti earthquake in 2010.
Even though geological records go back several thousand years and in some cases millennia; remote sensing can help outline extents and can give visual characteristics and contrasts more easily (depending on if the image is taken in the visible light spectrum, near-infrared or fake colour) by giving an aerial view, and in some scenarios pick out more of a contrast between different land areas than seen at ground level.

Remote sensing is no different for Merapi. Take for example Figure 4 of Mount Merapi; the whitish areas by the summit and down the south and south-west flanks can be identified as pyroclastic flows and lahar deposits within current or old radial river channels.

Figure 4: Black and white satellite image (Surface Radiance VNIR) received from NASA by request and edited via ArcMap 9.3 (image taken in 2003). (The white area SE to the summit is a cloud and is not to do with any explosive traits)

Satellite based observations much like that of ASTER is leading to new levels of understanding of complex Earth processes that often lead to disasters. Satellite observations are continuing to demonstrate the potential of remote sensing systems in the operations of decision making that impact upon loss of life and property, as well as providing a better foundation for aerial imagery and constant surveillance.

Unfortunately remote sensing techniques still leave a few questions and problems unanswered. Considering Camus et al (2000), Lavigne et al (2000) and Thouret et al (2000)’s research on the records of Mount Merapi. According to their reconstructions; (mapping and historical accounts) explosive episodes much larger than the eruption in 1930 – 31 sweep the flanks of Merapi at least once on average every century (Thouret et al 2000). Which is in contrast to the much more frequent and much smaller episodes of pyroclastic flows which are due to partial or full dome collapse (for example: the last major pyroclastic and major lahar flows were in 1994, with the last minor pyroclastic flows in October 2010).

2.5 Critiques
The techniques within remote sensing such as geodetic measurements have helped scientists to understand volcanoes to a greater depth; including gas travel, local topography and terrain changes. However, Camus et al (2000), Lavigne et al (2000) and Thouret et al (2000)’s research poses several questions which cannot be answered with remote sensing; why does the eruptive history culminate toward a large voluminous eruption when the debris flows, lahar flows and pyroclastic flows happen so often? Will this eruptive history continue on or will it shift to what it was like before 1700’s when Merapi cycled through it stages; ‘Ancient’, ‘Middle’, ‘Recent’ and ‘Modern’ quickly? If the latter does happen, when will it happen? And how many people will be at risk? To understand these questions, a further analysis into the eruptive history of Mount Merapi and the surrounding area is needed.

Mount Merapi is a complex volcano with multi-cited hazards, but to what extent of danger do these hazards pose? And with that, how many people are at risk from which hazards? To resolve the complexity of the hazards on Mount Merapi, this thesis has elected to analyse four separate hazards with different ‘risk values’ (covered in further detail in Methodology, Results and Analysis sections) they are:

·    A stream risk buffer zone which has a risk value of 1. This field is needed to show a zone surrounding streams that might be at risk from lahar flows. As lahar flows tend to flow down radial valleys and can overflow the banks.
·    Four slope zones which each have a risk value of 1 (totalling 4 when all are overlaid). This field is needed to show that the surrounding area may be subject to landslides and due the seismic originality of volcanoes may provoke these further. Also the surrounding area may also be laden with thick ash which can be very unstable and can aid in the production of lahars.
·    Multiple lahar and pyroclastic zones which each have a risk value of 1.5. This field is needed to show the direct threats that are imposed on the surrounding areas by previous lahar and pyroclastic flows with the worst case scenario being death.
·    Five gas zones which each have a risk value of 0.5. This field is needed to show that even though some areas may be out of direct threat from lahar and pyroclastic flows they can also be affected by gas damage.

3. Study Site

3.1 Overview
This chapter covers an overall view on the archipelago of Indonesia and the island of Java, with regards to its site and location, climate, topology, demographics, culture and religion and covers how each of these factors have been influenced by Mount Merapi. This chapter then concentrates on the aspects of: Mount Merapi’s activity (including recent activity), populations at risk and current monitoring strategies.

3.2 Introduction
Indonesia is an archipelago of around 17,508 islands (about 6,000 of them are inhabited) (Witton and Elliot 2003). The archipelago is situated between 4ºN and 10ºS Latitude and 95ºE and 124ºE Longitude and shares the borders of Papua New Guinea, East Timor and Malaysia. In 2010, Indonesia’s population was 227 million (World Bank 2010). The five largest islands by size in Indonesia are: Java, Sumatra, Kalimantan, New Guinea and Sulawesi. Java has the largest population of the islands at around 136 million living in 1,026 people per km² (and is the most populated island in the world) (Witton and Elliot 2003) which accounts for around 62% of the population of Indonesia (Indonesian Embassy 2005).

3.3 Local Geography and Social Context
Java is almost of entirely volcanic origin. The island contains 38 volcanoes forming an east to west volcanic arc which have all at one point been active (20 of which have been active in the last Holocene) (Witton and Elliot 2003). The highest volcano on the island is Mount Semeru (3,676m) with the most active volcano being Mount Merapi (2,968m) (Ricklefs 1993).

The climate varies between humid and tropical climate with two distinct seasons; the rainy season and the dry season. Indonesia’s rainy season (and coincidentally the highest risk months for syn-eruptive lahars) runs from October to April with the wettest month being in January (an average of 335mm over 19 days) (Climate and Temperature 2010). The dry season of Indonesia runs from May to September with the driest month being August with an average of 50mm over 5 days.

Indonesia has a varied demographic, with a high majority of Indonesians of Malay relation, the remaining natives are Melanesian. Java, however, entails a slightly different demographic, only three ethnic groups co-exist on the island: Javanese (~70%), Sudanese (~20%), and a small group of Madurese (10%) (Witton and Elliot 2003). Unfortunately for the population of Java a lot of the larger towns are situated around or near volcanoes that have been active in the last Holocene, putting a tremendous of risk on the population of Java.

Mount Merapi is the most active volcano in Java as has changed the culture on the flanks and the surrounding area. The local ‘culture of hazard’ (Dove 2008; Donovan 2010) on Merapi is not shared by the Indonesian government; the government views the volcano as something beyond the ‘normal social order of things’ (Dove 2008) and as a consequence has become prominent in government resettlement programs. Dove (2008) also states that Merapi villagers display remarkable harmony in their opposition to resettlement. In the aftermath of the 1994 eruption 7,692 households in villages lying in the danger zone were interviewed and less than 1% expressed any interest in transmigrating. Many villagers saw the government resettlement program as just ‘another hazard’ and many preferred the hazard that they knew to the one that they did not (Dove 2008). This problem of not migrating away because of rebelling against the government has related in many issues concerning evacuation and resettlement strategies especially after the 2010 eruptions.

Within Java, religion is rather homogenous, over 90% are Muslim with small portions of Catholicism, Buddhism and Hindu (Van der Kroef 1961). In the last 1000 years, religion in Java has shifted around from Central to East Java and vice versa due to more transport links but was originally caused by Mount Merapi and the devastation it caused upon the temples on the surrounding flanks around 928AD (Newhall et al 2000).

With so many people with so many backgrounds, ethnicities, and religions, why do they all choose to live in Indonesia and especially Java, which has a vast amount of active volcanoes? What are the push and pull factors? And if there are, are there factors which the local population are unaware of, for example: the possible hazard extents of Mount Merapi?

Before these questions can be answered an overview on the activity of Mount Merapi must be given, which will in turn, highlight the questions just given, especially the last.

3.4 Mount Merapi’s Activity
A large proportion of the 175,000 deaths due to volcanic activity over the last two hundred or so years worldwide have occurred on the island of Java, Indonesia (Chester 1993). There are 129 volcanoes and mountains just on the island of Java, and the most active is the volcano; Mount Merapi. Written historical records show that Merapi has had at least thirteen major eruptions with human casualties recorded since 1006 (61 eruptions if including minor eruptions).

Merapi’s activity has a varied chronology depending on volcanic impacts;
·    Lahars occur on average every 3 – 4 years, causing short term damage, such as: land damage and minor building damage (last occurred in 2008 and 2010).
·    Brief explosive intervals occur every 8 – 15 years which generate lahars and pyroclastic flows, which have previously generated partial dome collapse and destroyed part of the pre-existing dome (last occurred in 1994 and 2010).
·    Very violent explosive episodes occur on average every 26 – 54 years which generates pyroclastic flows, surges, tephra-falls and lahar flows. Last occurred 19th December 1930 – 31 when large pyroclastic flows travelled 12km from the summit covering an area of 20km² and destroying 13 villages killing over 1,300 people (this type of eruption is heavily overdue).

Even though the last century speaks of multiple hazards, the villagers on the flanks of Merapi speak of only two hazards that threaten their lives (Dove 2008): ‘ampa-ampa(s)’ and lahars.

‘Ampa-ampa(s)’ are the most feared aspect of Mount Merapi. ‘Ampa-ampa’ is the eruption of a type of pyroclastic flow that consists of revolving clouds of super heated gases (known as ‘nuee ardente(s)’ in international literature). These super heated clouds descend the slopes at speeds of 200 to 300kmph and have internal temperatures of 200-300ºC which can instantly carbonise wood. These ‘nuee ardente(s)’ establish a much higher threat to the life on the flanks of Merapi than the slow moving more frequent lava flows (Dove 2008).

A lahar is a mud flow composed mainly of volcanic ash lubricated by water derived from the bursting of a crater lake, from snowmelt or from pro-longed torrential rain causing the volcanic ash to flow under gravitational movement (Whittow 1984). Lahars are common throughout most volcanoes in the world. At least 23 of the 61 eruptions of Merapi since the mid 1500’s have produced lahars (Lavigne et al 2000). The total area covered by these lahars cover around 286km² on the flanks of Merapi. The lahars at Merapi are commonly triggered by rainfalls that average around 40mm in 2 hours which occur in the rainy season between November and April and have average velocities of 5 to 7 m/s. Although the velocity of lahars can vary heavily depending on the terrain or obstacles it encounters, for example; the lahar may pick up debris along the way from past eruptions of fallen trees or lahars may coincide with a river valley and become a highly concentrated stream flow, which can reach up to 60kmph such as in Nevado Del Ruiz, Colombia in 1985 (Naranjo et al 1986).

There is also a possibility that lahar flows may be syn-eruptive or post-eruptive, the differences are:

·    Syn-eruptive lahars or hot lahars are generated by rainfall during or relatively soon after an eruptive episode. At least eight of the 61 reported eruptions at Merapi since the 1500’s are syn-eruptive (Lavigne et al 2000). The average frequency of syn-eruptive lahars at Merapi is one every 30 years. Normally lahars that occur on the flanks of Merapi occur in a few rivers on the flanks, for example: the Senowo River, the Blongkeng River and the Batang River. However, on the 19th December 1930 and on the 7th January 1969 lahars occurred along nine of the rivers that surround the summit with the largest damage (due to lahars) on the Western flank of the volcano.

·    Post-eruptive lahars or cold lahars are usually smaller, but much more frequent than syn-eruptive lahars. The frequency of post-eruptive lahars depends on many variables, with the main variables being: channel rainfall characteristics, channel total volume and grain size distribution of pyroclastic deposits. For example; soon after the large eruption of 1930-31 33 lahars followed the first rainy season, but only 21 lahars followed the eruption in November 1994 (Lavigne et al 2000).

The high intensity and risk variance of lahars puts a large amount of risk on the surrounding villages, especially due to the fact that there is a yearly rainy season so the chance of lahars of varying sizes and potential damage increases every time the season re-occurs.

Table 2 shows the eruptions of Mount Merapi from 1672 to 1997 with estimated life loss and in some cases how they died and number of known affected villages.

Table 2: Adapted from Thouret et al (2000) showing Merapi’s activity from 1672 to 1997 including: death toll, number of villages affected and number of syn-eruptive lahars

3.4.1 Recent Activity on Mount Merapi
Activity on Mount Merapi has been rather benign recently, with only 12 eruptions in the last 12 years (USGS 2010), however Merapi has started to erupt again putting the growing population at risk. An eruption occurred from Mount Merapi on 28th October 2010. Cloud cover prevented satellite observations, so monitoring and warning systems were slow and delayed. Two pyroclastic flows occurred on 30th October and ash fell in Yogyakarta 30 km away. The Center of Volcanology and Geological Hazard Mitigation (CVGHM) noted four further pyroclastic flows the next day as well.
On 1st November 2010 Mount Merapi erupted again, after venting since the last eruption. Around seven pyroclastic flows occurred, traveling south-south east of the summit at a distance of 4 km. A gas and ash plume rose 1.5 km above the crater and drifted East and North. CVGHM recommended that evacuees from several communities within a 10km radius should continue to stay in shelters or safe areas. CVGHM reported that an ash plume rose to an altitude of 6.1 km (USGS 2010). On 2nd November the ash plume was seen via satellite imagery drifting 75 km north and air traffic was diverted and cancelled in and out of the Selo and Yogyakarta airports (the local airports) worsening evacuation procedures.

CVGHM reported a further 26 pyroclastic flows on 2nd November. Around 38 pyroclastic flows occurred during the first 12 hours of the day, 19 of which travelled 4 km south (plumes from the pyroclastic flows rose 1.2 km). The end death toll was estimated at 275 people and over 320,000 people had been displaced from the flanks (BBC 2010).

Taking the recent eruption(s) of Mount Merapi into consideration and its relative history whereby pyroclastic flows have a reoccurrence of 8 – 15 years and that 275 people died and over 320,000 people have been evacuated from the danger zone. If a larger event does occur, such as a flank failure or lava dome extrusion over 1.1 million people could be at risk. The logistics and risk that would fall upon the Indonesian government would be unbearable. Hopefully, this recent ‘scare’ will force the local population to rethink their area of settlement and move away and this is in fact the perfect time for the Indonesian Government to enforce resettlement programs and move populations away from the danger zone(s) and generate new hazard zones imposed by Mount Merapi.

3.5 Population at Risk
Around 16% of the population live around 16 active volcanoes on the island of Java, which accounts for about 7% of the total area of Indonesia (Thouret et al 2000). The region between Mount Merapi and Mount Merbabu (another volcano 10km North from Merapi’s summit) supports around 1.1 million people in 300 villages above 200 metres in elevation, making these people the most vulnerable to any eruption in the surrounding area.

The highest recorded death toll from Mount Merapi was in 1672 which killed at least 3000 people (Dove 2008). Also, the population of Java at the time was around 7 million people, as a comparison, the population in 2010 is estimated to be around 136 million this puts the population at a tremendous risk.
The total amount of deaths since the 1500’s is estimated to be around 7,000 people (Thouret et al 2000), if population distribution is the same as it was in 1672 (which is doubtful), using the same percentage calculation of population versus death count could put the potential death toll of over 53,000 people if an eruption of a similar proportion happened soon, posing a very serious risk on the local population.

The southern and western flanks (the most prone to volcanic activity of Mount Merapi) are part of the Yogyakarta plain, a fertile region of land used heavily for cropland (especially rice cultivation) which is replenished by nutrients by the activities of Mount Merapi. Yogyakarta (see figure 6 for location), the largest city in the Yogyakarta plain, is a city of at least half a million people which is ranked highly in Indonesian culture, history and economy and is only 30km away from the summit. Within Thouret et al (2000) it was calculated that the population on 387km² of Mt Merapi flanks (which include the Yogyakarta plain) was not made on the same basis 24 years ago (Table 3) which means about 440,000 people (which is around twice as much as in 1976) are at risk by pyroclastic flows, surges and lahars from Mount Merapi.

Table 3: Thouret et al (2000) – Population at risk, people density and growth around Merapi, 1976 – 1995

Mount Merapi presents many characteristics of the most dangerous volcanoes of the world (Crandell et al 1984) as it has a reliable eruption record with persistent activity. Many journalists and researchers (including the Indonesian Government) have divided up regions of Mount Merapi into risk areas (first used by Suryo and Clarke 1985) and it is used as the official hazard map for Mount Merapi. The regions are as follows: “Forbidden zone”, “First danger zone” and “Second danger zone”.

Within Thouret et al (2000) it states that the ‘First danger zone’ can be affected by or is prone to tephra-fall or lahar flows with pyroclastic and lava flows being out of reach. The ‘Second Danger Zone’ is situated along the radial valleys of the streams that drain from the summit (see figure 5). These radial valleys are prone to lahar flows and can and have travelled 30km from the summit and have affected or partially affected the larger towns such as Yogyakarta and Prambanan (Lavigne et al 2000). The ‘Forbidden zone’ on the other hand is closest to the summit and is prone to all variants of volcanic activity from lahar flows, landslides, highly concentrated gas emissions, pyroclastic flows, lava flows and lava bombs.

Figure 5: Official hazard map of Mount Merapi adapted from Suryo and Clarke (1985), which shows the First, Second and Forbidden zones as well as main roads and villages

The Indonesian Government hazard map made by Suryo and Clarke (1985) has become irrelevant in modern literature as many eruptions in the last 25 years have gone past the danger zones and also the map only takes into account the eruptions of 1930-31, 1961 and 1969. Thouret et al (2000) believes that this is not adequate enough for accurate hazard-zone mapping.

3.6 Monitoring
As Mount Merapi is the most active volcano in Java, there have been several monitoring strategies put in place. Seismic monitoring on Mount Merapi began as early as 1924 and there is also a network of 8 seismographs surrounding the volcano to accurately pinpoint earthquakes and tremors. Since monitoring began, scientists have discovered that no earthquakes occur about 1.5km below the summit which is thought to be the location of the magma reservoir which feeds the eruptions of Mount Merapi.

The most active section of monitoring for Merapi is that of lahars as they occur every 3 to 4 years and have such dynamic extents depending on: the effect of past eruptions (if ash is still on the surface), the rain output of the rainy reason, which radial valley(s) the lahar(s) flow down and also the dormancy time. Many researchers have studied this due to the high frequencies for example Itoh et al (2000), Lavigne et al (2000a), Lavigne et al (2000b) and Thouret et al (2000) and in turn have highlighted and aided the risk perception of lahar flows in the larger towns such as Yogyakarta.

4. Methodology

This thesis will now use old historic data with remote sensing techniques using ArcMap 9.3© and Google Earth to provide a better understanding (via a hazard risk map) of Mount Merapi from a remote sensing perspective. All data was corrected to UTM zone 49S, WGS84.

4.1 Datasets

ASTER – Global DEM
The elevation data used for this thesis was downloaded from NASA and was generated using data from the ASTER sensor. The DEM provided covered Mount Merapi and the surrounding area (including Mount Merbabu) and was given at a 30m resolution. This image provided height data, but lacked explicit detail even at 100 standard deviations.

Near Surface Infrared images (taken in 2003) was provided by NASA using ASTER sensor EO-1. ASTER sensor provides 15m resolution imagery. The images provided an overall view of Mount Merapi and the surrounding area (including Mount Merbabu). The Near Infrared image provided a more detailed look into the terrain (much like a satellite image but in gray scale) compared to the DEM but contains no height data.

Historical Data
Several maps were scanned from existing sources (e.g. Thouret et al 2000, Voight et al 2000, Camus et al 2000 and Donovan 2010). These were subsequently registered to the ASTER data baseline using ArcMap 9.3© and georeferenced, the features of interest were then digitised. In total 36 shape files were generated from historic data – 32 lahar and pyroclastic flows and 4 ash emissions.

4.2 Risk Map Layers

Figure 6: Flow chart logically depicting the structure of the thesis methodology and steps taken.

All remote sensing images are in the Results section.

4.2.1 Stream Extraction and Buffering

The first phase in this process will be using an extracted stream network. The stream network (once extracted) will have a buffer zone of 100 metres. This range has been selected as lahars flow down the bank overflowing the valleys (such as the eruptions in 1930-31, 1969, 2004 and 2010). Lahar overflow is usually related to the gradient of the radial valley, whereby the lower gradients are more vulnerable to the higher gradients (e.g. Plate 2).

Plate 2: Part of the Kaliadem Lava Tour situated near Yogyakarta showing the silt, debris and ash build up in the lower parts of the valley due to lahar overflow, http://www.tourjogja.com/berita-184-kaliadem-becoming-a-lava-tour-area.html

Extracting the stream network was calculated using ArcMap Spatial Analyst Tools. The first step was to fill the DEM to make sure there were no holes within the data (i.e. natural sinkholes). Flow direction was then calculated which provides the direction of flow from each DEM cell to it’s steepest down slope neighbour. Flow accumulation was then calculated and provides the number of cells that flow into the current cell. The ‘con’ tool was then used on the DEM with the expression of ‘Value > 250’ which removes the upper values. Stream Link was then calculated from this which links parts of the identified stream network that are missing. Stream order was then subsequently used which calculates and identifies the order of the stream segments, with respect to the direction of water flow. Finalising the process Stream to Feature was used which extracts the raster streams identified in the previous steps to create a vector shape file. The extracted streams were then subsequently buffered by using Analysis Tools, Proximity, and then Buffer. Figure 7 in the Results section depicts the end result.

4.2.2 Slope Risk Areas

The next stage is to create the risk associated with sloped areas as these areas are vulnerable to ground movement and possible landslide risk due to the ash and tephra in the local vicinity making the ground unstable, especially due to the seismic origin of volcanoes. The key parameter was the slope and this was calculated using ArcMap Slope tools on the DEM provided. Figure 8 shows the end result.

4.2.3 Lahar and Pyroclastic Prone Areas

As lahar flows and pyroclastic flows are the most frequent hazards on the flanks of Mount Merapi the risk associated with these areas have become ever more distinguishable as they affect land use, population health, population livelihood and cause building damage.

The lahar flows and pyroclastic flows were grouped together into one shape file because some sources did not distinguish if an eruption is a lahar flow or a pyroclastic flow and only a date was provided. Also some of the sources have varied terminologies for the different flows as some are of Dutch origin which date back to 1800’s such as some images from within Voight et al (2000).

This thesis will now use historic lahar flow and pyroclastic flow data found within papers by  Camus et al (2000), Donovan (2010) Thouret et al (2000) and Voight et al (2000) and create different shape files via ArcMap 9.3© via compiling all eruptions that are available via a birds eye view layout of lahar flows and pyroclastic flows. The risk areas on the map were then generated by drawing around the eruption data from the sources via ArcMap 9.3©’s Editing tool. Figure 9 shows the end result.

4.2.4 Gas emission areas

The damage caused by gas emissions varies from: death by suffocating, death by long term lung damage, ash fall damage of roofs and crop failure. Gas emissions is usually expressed via concentration of particles in a certain area (kg/m3); however it is hard to find and record old gas emissions and their concentrations as wind and other natural forces distribute or degrade the remains.

Ash emissions are less dense than ground flows and so the distance of a gas emission to a ground flow is a lot further. Also gas emissions can affect local weather and climate and in extreme scenarios can affect global climate such as Mount Pinatubo’s eruption in 1991 which caused a drop in global temperatures by 0.5ºC (Pitari 2002).

The data this thesis concentrates on for gas emissions is taken from Voight et al (2000). Creation of the data is similar to 4.2.3 via ArcMap 9.3©’s Editing tool. Figure 10 shows the end result.

4.3 Risk Calculation

The next stage in risk assessment is calculating how much risk is present in the areas using the different hazard data posed by Mount Merapi as the datasets have different risks attached to them.

4.3.1 Assignment of Risk Values

Table 4: Hazards apparent around Mount Merapi with associated risk values and reasoning

4.3.2 Generation of the Risk map

Each parameter was rasterised to a resolution of 30m (the same as the ASTER DEM). These layers were then subsequently added together where any ‘NULL’ values were ignored. Figure 13 depicts rasterised historic data and Figure 12 depicts the final risk map with all the files calculated together.

The successive product was then exported for visualisation onto Google Earth (Figure 13a and Figure 13b).

5. Results

5.1 Risk Layers
The following sections show the stages in risk map generation: Buffered Streams (5.1.1), Slope Areas (5.1.2), Lahar and Pyroclastic Prone Areas (5.1.3) and Gas Emissions (5.1.4).

5.1.1 Buffered Streams

Figure 7: DEM image with stream network extracted and a 100m stream buffer created via ArcMap 9.3© with DEM colour schemed in black and white using hill shade effect

Figure 7 shows the risk value associated with the extracted stream network around Mount Merapi. The streams (blue) have a 100 meter buffer (red) to clarify the threat around radial valleys. All streams are buffered as lahars have been known to travel down all of the radial valleys of Mount Merapi at least once in the last 200 years.

5.1.2 Slope Areas


Figure 8: DEM image with stream network extracted and slope factor data extracted from the ‘Filled’ DEM created via ArcMap 9.3©

Figure 8 shows the risk associated with slope areas surrounding Mount Merapi. Figure 10 shows the slope at ‘con’ >10 to show the full range of the parameter, as opposed to the less apparent and smaller dataset of ‘con’ >40.

5.1.3 Lahars and Pyroclastic Flows

Figure 9: Surface Radiance near Infrared image and under-laid DEM with new shape files created from scans from papers of Camus et al (2000), Donovan (2010), Thouret et al (2000) and Voight et al (2000) and put to hollow colour schemes

Figure 9 shows the risks associated with the lahar and pyroclastic flows throughout history gathered from Camus et al (2000),  Donovan (2010), Thouret et al (2000) and Voight et al (2000).

5.1.4 Gas Emission Areas

Figure 10: Surface Radiance Near Infrared Image and Google Earth image to extend the visible spectrum with gas emission shape files created from Voight et al (2000) and put to hollow colour schemes

Figure 10 shows the risks associated with historic ash emissions and the range that ash can travel from the summit while still posing a threat.

5.1.5. Rasterised Historic Data

Figure 11: Surface Radiance near Infrared image and under-laid DEM with shape files mentioned in Figure 9 and 10 converted to rasters and put to varied colour schemes.

Figure 11 shows the shape files shown in Figure 9 and 10 converted to rasters at a resolution of 30m to match the DEM beneath it.

5.2 Final Risk Map

Figure 12: Calculated risk map of Mount Merapi using 41 shape files (32 lahar and pyroclastic flows, 4 ash emission files, 4 slope files and 1 buffered stream file) with different risks associated with each group. Colour scheme set to stretched.

This final risk map shows the risks associated with the surrounding area around Mount Merapi, with every parameter: Buffered Streams, Slope Areas, Lahar and Pyroclastic Prone Areas and Gas Emissions included and overlaid. The brightest central colour shows the highest risk of 29.5.

The following images (Figures 13a and 13b) are the same risk map but overlaid onto Google Earth and showing the towns/ villages in potential danger.

Figure 13a: Final Risk Map of Mt. Merapi on Google Earth© with summit pinpointed. Colour schemed to 32 classes.


Figure 13b: Final Risk Map of Mt. Merapi on Google Earth set an oblique angle to show relief (Z value = 1). Colour scheme set as 32 classes

5.3 Sources of Error

Although a quantitative analysis of error or accuracy was not possible as no field data or independent source was available; when compared to Google Earth and the recent 2010 eruption a close correspondence was observed. The October and November 2010 eruptions occurred in a southerly direction which is similar route to some early eruptions in Mount Merapi’s history and occurred in a risk value area between 25 and 15.

6. Analysis and Discussion

6.1 Hazards of Mount Merapi
Volcanoes are dynamic systems that can produce varied explosive tendencies such as lava flows, lava bombs, nuee ardentes, lahar flows, pyroclastic flows, ash emissions and landslides (Figure 2). With such explosive traits volcanoes tend to produce high risk sites; if using Blaikie et al’s (1994) equation of risk where by ‘Risk = Threat x Vulnerability x Cost’, the three most important factors that influence risk are: Threat, vulnerability and Cost. But how does this equation fit into the concept of Mount Merapi?

As figures 2, 13a and 13b and Table 2 show, the idea of ‘threat’ is very imminent on the flanks of Mount Merapi. The threats are summarised below using Dove (2008), Thouret et al (2000), and Voight et al (2000):

·    Lahar flows occur on average every 3 to 4 years – these vary in damage and volume and can bring ruin to anything of close proximity to the summit or along radial valleys.
·    Brief explosive intervals (mainly pyroclastic flows of varying intensities) occur on average every 8 to 15 years – these can flow up to 200 to 300kmph and can have internal temperatures of 200 to 300ºC, instantly carbonising wood and usually causing fulminant shock at the moment of death.
·    Very violent explosions occur every 26 – 54 years which can produce a variety of impacts:
o    Lahar flows
o    Pyroclastic flows
o    Lava flows: A mixture of molten rock and ash that flow short distances from the summit. Generally slow moving (depending on viscosity of the lava which is heavily dependant on basaltic content). Also if ephemeral conditions are water-ridden, ash-ridden or rubble-ridden the speed of the lava flow can vary tremendously. Temperatures of lava flows on Mount Merapi can also reach up to 1200ºC (due to it being basaltic lava tendencies). Movement is generally slow so damage from lava flows is generally concerned with building or land damage.
o    Lava bombs or volcanic bombs: these are rocks that are blown out of the volcano that are 2.5 inches or larger in diameter. Lava bombs tend to be  more associated with deaths than building damage (with more deaths closer to the summit as most bombs fall in close proximity as it takes a large amount of energy to propel the lava bomb a further distance.
o    Gas emission: Can occur on any interval of Mount Merapi (3 to 4 years, 8 to 15 years and 26 to 54 years), in fact Mount Merapi exhumes gas nearly everyday of the year. However, the reason for mentioning it in the last section is because large quantities of gas and ash can become very hazardous to human health and can damage the local wildlife and climate (or perhaps global climate such as Mt. Pinatubo in 1991).  To emphasise the impact of gas and ash, the recent eruptions in October and November 2010 produced a gas plume that rose to 6.1km into the atmosphere. This plume closed airports in Selo and Yogyakarta, grounding many evacuation procedures and kept many people in the danger zone. Ash damage tends to be associated with long term impacts in human health, especially that of lung damage. Short term damage of ash emissions is usually building damage especially that of roof collapse due to the sudden build up of weight.

6.2 Analysing populations at risk
Using the information above and the maps (figures 13a and 13b especially), the sheer amount of risk imposed on such a region is exceptionally large. At current standings Figures 13a and 13b show labels of town or villages or cities that are within the danger zone, but their population statistics are lacking. So far this thesis has mentioned that over 1.1 million may be at risk (Thouret et al 2000) but how is this population distributed and to what degree of risk is imposed on the populations in these regions? For example: 20,000 people maybe in a risk value of 10, where as 100,000 maybe in a risk value of 5: even though the latter is a lower risk value the higher population will have a higher ‘literal’ risk due to a much higher population and also much more difficult logistics in evacuation procedures and more potential building and land damage. Also with long term effects in mind large economic centres might be affected, which will generate a longer extended period of re-growth on the affected area. This impact will not only place the damaged economic centre into poverty but the surrounding villages and towns that rely on that economic centre for jobs, services and products or food into poverty as well.

The estimated village regions that surround Mount Merapi’s summit are shown in figure 14:

Figure 14: Final risk map classified to 32 classes, overlaid by 77 estimated village areas taken from Donovan (2010)

As population statistics are very difficult to find due to a rapidly increasing population and a high migration within and throughout Indonesia. This thesis will pinpoint and highlight the effects on a certain few large towns that are located near Mount Merapi’s summit and using population data of those areas highlight the risk that will or could affect those populations. Locations of these areas are pinpointed in figure 15:

Figure 15: Google Earth© image with risk map (classified to 30 classes) and known village areas overlaid with towns/villages pinpointed with Mount Merapi summit pinpointed as well

The population statistics were collected from: Tageo.com – Worldwide Index – Indonesia City & Town Population (2004) which has precise x,y co-ordinate population data and FallingRain.com – World: Indonesia (1996) which has an accuracy of a 7km radius.

Klakah is a small town located 3.57km north-north-west of Mount Merapi’s summit. It is in line with most gas-emissions and has an estimated population of 72,850 (1996) (accuracy to 7km radius). Klakah is located at a risk value site of 10
Selo is a slightly larger town with a small airport; it is located 6.81km north-north-east of Mount Merapi’s summit. Selo’s population is estimated at around 76,273 (1996) (accuracy within a 7km radius). Selo is located at a risk value site of 5.5

Kemiren is a large town situated 7.53km south-west of Mt. Merapi’s summit. Kemiren is on the outskirts of the most frequent flow extents on the south-west flanks, but still within the large gas emission extents. Kemiren has an estimated population of 103,077 (1996) (accuracy of 7km radius). Kemiren is located at a risk value of 10.5

Muntilan is a much larger town than the others and is located 17.75km south-west-west of Mt. Merapi’s summit. The estimated population that resides here is 49,600 (2004). Muntilan is located just outside the ‘First Danger Zone’ provided in 1985. According to the risk map Muntilan has a risk value of 2, or perhaps 3 depending on which areas of the conurbation are within the buffer zone of the streams.

Ngaglik is a slightly smaller town to that of Muntilan with an estimated population of 39,200 (2004) and is located 23.01km south-south-west of Mt. Merapi’s summit. The risk value is very similar to that of Muntilan; risk value of 2 but 3 if the conurbation is within the buffer zone of the rivers.

Salatiga is the largest town in the region on the risk map; it has an estimated population of 121,000 (2004). It is located 23.01km north-north-east of Mount Merapi’s summit. Salatiga is situated at a risk value of 1 (although gas emissions could modify this factor if the prevailing wind shifts)

Out of the towns and villages given, the most at risk is the town Kemiren. Situated at a risk value site of 10.5 which is potentially putting around 103,077 people at a very high risk of gas emissions, pyroclastic flows and lahar flows; 7 eruptions have occurred in that area in the last 200 years. This is putting a tremendous amount of risk on that region. The evacuation procedure from Kemiren is exacerbated as well as many eruptions have passed this point on the flanks making it even more difficult as many of the roads and access points will be blocked (if not destroyed) if an eruption occurred in that particular region again.

The highest population in the region is that of Salatiga of 121,000 although a much lower risk value (1 as opposed to 10.5) the greater population will generate a larger danger as evacuation procedures will become more crowded and generate problems or queues in the process.
However, not only do these town and cities have a significant amount of risk upon them the village areas (see figure 14 and 15) show that there are many more communities with populations that are unknown. Although considering FallingRain.com’s statistics with the inaccuracy of 7km radius of the towns given, it can be assumed that there are lot more people that live on the flanks than Google Earth shows. Considering Thouret et al (2000)’s thesis on the population with the estimation of around 300 villages within a 200m elevation, it can be safely assumed that a large population of around 400,000 people are unaccounted for. This “unknown known” (Romsfeld 2002) has been known to cause natural disasters rather than natural hazards without the proper evacuation procedures.

6.3 Social Vulnerability of the village areas
Another factor that can influence and impact on the populations on the flanks of Mount Merapi is that of social vulnerability. Utami (2008) has conducted studies on the village areas (shown in figure 14 and 15) which depict the social vulnerability index (SVI) of the areas according to poverty, accessibility and gender; Figure 16 shows these areas at a classified scale:

Figure 16: social vulnerability index of village areas surrounding Mount Merapi (taken from Utami 2008)

Although population statistics were not provided in the thesis by Utami (2008) due to villages and towns having such a migrating and unconnected population Figure 16 does provide another factor of potential risk on population within the unknown areas. Figure 17 shows SVI overlaid on the final risk map:

Figure 17: Overlaid SVI map taken from Utami (2008) with final risk map set to 67% transparency and 32 classes

Figure 17 clearly shows that a large proportion of high social vulnerability is found within the south-west high risk zones which vary between 7.5 to 28.5 such as the village regions: Ngablak (A), Ngargosoko (B) and Tlogolele (C) which have a SVI of 0.5 to 1.5. Located just south-west of these village areas are: Tegalrandu (D), Srumbung (E) and Polengan (F) which have a SVI of >1.5 but are on risk factor variance of 5 to 10. This is a problem similar to that of population risk; Ngablak has a lower SVI than that of Tegalrandu but because of the risk value that is in Tegalrandu (4.5 to 8 as opposed to the larger range of 0.5 to 14 in Ngablak) it boosts the ‘literal’ risk of the area. As mentioned before, social vulnerability highlights how much money an area has and the preparedness of the region, which in turn affects how badly the region is or can be affected by a natural hazard in this case; the multiple hazard impacts of Mount Merapi. With the added curse of a booming population and the staggeringly low SVI, this generates and highlights the impacts of the need for evacuation procedures and education in the village areas surrounding Mount Merapi.

Furthermore, the recent eruptions of 2010 of Mount Merapi occurred within the large village area directly South of the summit (Hargo Binangun) which has the highest SVI (<-1.5) and therefore the least affected by hazards,  yet it still generated an estimated death toll of around 275 people. Perhaps this should be seen as a warning for the Indonesian Government to rethink their strategies and education and/or move the population that still reside on the flanks of Mount Merapi.

7. Conclusion

The previous figures highlight how hazardous the flanks of Mount Merapi can be; from known hazards of pyroclastic flows, lahar flows and ash emissions (figure 13a and 13b), estimated town and village populations (figure 14 and 15) and social vulnerability (figure 16 and 17). Although the risk map already highlights major risks there are however several drawbacks that could potentially increase the already extreme risk values that are found on the flanks of Mount Merapi.

7.1 Limitations with the risk map
There are so many factors that can influence and build up a risk map and in doing so mistakes can be made and accuracies can be changed. This risk map is no different; the following reasons explain why this certain risk map is flawed:
·    Only one journal was used for gas emissions, this is due to many journals that concentrate on Mount Merapi incorporating only the ground based volcanic activity, possibly due to gas emissions being hard to track over a long period of time especially over a period of 200 years or so as ash is washed away from the surface when the monsoon rainfall occurs every year. This suggests that the region surrounding Mount Merapi could be at more risk than originally thought. Also prevailing wind direction alters throughout the year, if and when an eruption occurs the ash emission danger area can change, which in turn puts more areas at risk.
·    Eruption data was collected only from the last 200 years. As Newhall et al (2000), Berthommier et al (1992) and Camus et al (2000) suggest Mount Merapi has been around for at least 7,000 years which means many eruptions have been missed out possibly due to extents and deposits that are hard to distinguish. Also many original eruption data results were written in Dutch and sketched, making it hard to decipher which could further increase the risk values. On top of this, Newhall et al (2000) say that Mount Merapi’s activity is benign in the 20th Century if the data is only over the last 200 years and the last 100 years are benign this potentially generates many more hidden results than originally thought, which in turns puts the region around the volcano into even more risk.
·    Also Google Earth imagery on Mount Merapi is somewhat lacking (within journals and the satellite images available freely) especially considering the vast accumulation of research on the volcano as well as the recent eruptions. With a better image, the risk map could be better geo-referenced and therefore give a better accuracy for exact risk values for exact points. Also a better image would provide a clearer image especially for calculating and analysing programs such as hazard zonation and resettlement programs for the Indonesian Government.
·    Population statistics of the regions are from the years 1996 and 2004, bearing in the mind the rapid increase in population in developing countries this could potentially greatly increase the risk imposed on the regions as well as making evacuation procedures more complex as the population has most likely rapidly increased since those dates.

As you can see from the above statements the hazard map of Mount Merapi can be seen in a rather speculative light. However; the risk map does provide a good foundation for hazards from the last 200 years of eruptions of Mount Merapi, especially that of the new conurbations on the flanks of the volcano. But considering the past hazard map of the three zones: ‘Forbidden zone’, ‘First danger zone’ and ‘Second danger zone’ used by Suryo and Clarke (1985), Voight et al (2000), Thouret et al (2000), Dove (2008) and Donovan (2010) it is a vast improvement as many eruptions had passed the extents of the ‘Forbidden’, ‘First’ and ‘Second’ danger zones and highlight further risks.

7.2 Considerations for the future
Considering the recent eruptions in October and November 2010, did Indonesia have evacuation procedures in place, bearing in mind the hazard map has not changed since 1985? In short, the answer is no. Although, to defend the Indonesian government the village area affected did have one of the better SVI on the flanks of the volcano but for it to accumulate a death toll of around 275 people makes it seem that there are more issues involved than just social vulnerability in the equation for evacuation procedures. It took the Indonesian government to evacuate around 320,000 people 5 to 7 days. This response time is extremely slow considering the speeds of the hazards that flowed down the volcano around this time (pyroclastic flows reaching possible speeds of 200 – 400kmph and lahar flows varying in tremendous speeds depending which radial valley the lahar flowed down).

This goes to show that serious lessons needs to be learnt from disasters like this happening again (around Mount Merapi especially). If an eruption occurs and a highly populated area or highly social vulnerable area is affected large death tolls will occur if measures are not put in place.

This thesis demonstrates that the hazard risk map generated provides a good, confident understanding on what has happened in the past on the flanks and the surrounding area of Mount Merapi and what might happen in the future, given the ever growing population around Mount Merapi, and the implications of this on evacuation procedures, livelihood and potential death toll.

8.    References

I would like to thank the continuing guidance of the Institute of Geography and Earth Sciences department of Aberystwyth University, especially that of Dr. Pete Bunting for his knowledge on GIS software, Dr. Carina Fearnley for her help with analysis on volcanoes and the hazards they pose and Dr. John Grattan for the inspiration to pursue this topic and Dr Kate Donovan from the University of Portsmouth for providing me with her Thesis on a very similar topic and for regular help and communication. I would also like to thank NASA for the models and data of Indonesia which was given free of charge. Without them I could not have achieved many of the goals within this thesis.
Unless clearly stated otherwise, the data collection, analysis and interpretation presented in this dissertation result from my own work alone.

This article is courtesy and copyright of the author David Harris