18.02.2015 Views

Spatial Pattern of Socio-economic Landslide Vulnerability and its ...

Spatial Pattern of Socio-economic Landslide Vulnerability and its ...

Spatial Pattern of Socio-economic Landslide Vulnerability and its ...

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

<strong>Spatial</strong> <strong>Pattern</strong> <strong>of</strong> <strong>Socio</strong>-<strong>economic</strong> <strong>L<strong>and</strong>slide</strong> <strong>Vulnerability</strong> <strong>and</strong><br />

<strong>its</strong> <strong>Spatial</strong> Prediction by means <strong>of</strong> GIS-Fuzzy Logic<br />

in Kayangan Catchment Indonesia<br />

Guruh SAMODRA 1,2 *, Guangqi CHEN 1 , Junun SARTOHADI 2 , Kiyonobu KASAMA 1 ,<br />

Danang Sri HADMOKO 1<br />

1 Civil <strong>and</strong> Structural Engineering Department, Kyushu University, Japan<br />

2 Environmental Geography Department, Universitas Gadjah Mada, Indonesia<br />

The objective <strong>of</strong> this research is to identify the spatial pattern <strong>and</strong> to explore the spatial distribution <strong>of</strong> socio<strong>economic</strong><br />

l<strong>and</strong>slide vulnerability in Kayangan Catchment. Household survey was employed in this research in<br />

order to construct several indicators <strong>of</strong> socio-<strong>economic</strong> l<strong>and</strong>slide vulnerability. It was conducted by transect<br />

walk <strong>and</strong> structured (closed question) interview. The information <strong>of</strong> <strong>economic</strong> <strong>and</strong> social vulnerability was<br />

taken from 112 respondents through multistage r<strong>and</strong>om sampling based on the settlement obtained from l<strong>and</strong>use<br />

map from Bakosurtanal (Survey <strong>and</strong> Mapping Agency <strong>of</strong> Indonesia). Photograph analysis was also used to<br />

accompany the vulnerability analysis <strong>of</strong> household survey. Mean centre, st<strong>and</strong>ard distance, directional<br />

distribution <strong>and</strong> average nearest neighbor analysis were employed in order to figure out the spatial pattern <strong>of</strong><br />

socio-<strong>economic</strong> l<strong>and</strong>slide vulnerability. Therefore the spatial pattern <strong>of</strong> vulnerability accompanied by<br />

knowledge driven expert judgment was employed to build fuzzy membership analysis. Fuzzy logic technique<br />

was applied in order to predict spatial distribution <strong>of</strong> overall settlement polygon in Kayangan Catchment.<br />

Nearest neighbor analysis <strong>of</strong> low, medium <strong>and</strong> high vulnerability level showed z-scores -4.52, -5.26 <strong>and</strong> -1.86<br />

respectively. It implied that the spatial distribution <strong>of</strong> each level <strong>of</strong> vulnerability was clustered. There were<br />

interlinked between natural feature, social behavior, <strong>economic</strong> level <strong>and</strong> vulnerability level. The fuzzy operator<br />

Or, Sum <strong>and</strong> high gamma value were successfully applied to predict spatial distribution <strong>of</strong> socio-<strong>economic</strong><br />

vulnerability level. The result coincided with the <strong>economic</strong>, social <strong>and</strong> ecological dimensions <strong>of</strong> livelihood in<br />

Kayangan Catchment.<br />

Key words: <strong>Socio</strong>-<strong>economic</strong>, <strong>L<strong>and</strong>slide</strong>, <strong>Vulnerability</strong>, <strong>Spatial</strong>, Prediction<br />

1. INTRODUCTION<br />

The recent increasing number <strong>of</strong> disaster dem<strong>and</strong>s a response for administrator <strong>and</strong> policy maker to<br />

minimize loss <strong>of</strong> life, damage property <strong>and</strong> people suffering. Several mitigation strategies <strong>and</strong> measures<br />

focusing on hazard have been addressed to overcome those objectives. However, it has not been fully<br />

effective due to the uniqueness <strong>of</strong> socio-<strong>economic</strong> community <strong>and</strong> environmental response to cope with<br />

disaster. It causes mitigation strategies <strong>and</strong> measures are inadequate to address the disaster as a threat.<br />

Recognizing broader spectrum <strong>of</strong> hazard <strong>and</strong> more appropriate framework analysis are necessary to<br />

identify the unique process <strong>of</strong> socio-<strong>economic</strong> <strong>and</strong> environmental response to natural hazard. <strong>Vulnerability</strong><br />

concept is widely proposed to overcome the drawbacks <strong>of</strong> limited <strong>and</strong> centered hazard analysis. As stated in<br />

the Hyogo Framework 2007 that vulnerability analysis are dem<strong>and</strong>ed for risk analysis dealing with natural<br />

disasters <strong>and</strong> the societal consequences <strong>of</strong> disaster (UNISDR, 2007). Nowadays, vulnerability becomes<br />

central concept in disaster research <strong>and</strong> mitigation strategies (Menoni, et al., 2012).<br />

Recent development <strong>and</strong> debate <strong>of</strong> vulnerability analysis related to terminology (Costa <strong>and</strong> Cropp, 2012),<br />

concept (Wolf, 2012), methodological, robustness <strong>and</strong> legitimacy (Kaynia, et al., 2008; Fekete, 2011) are<br />

documented well. In addition, spatial data analysis with GIS techniques <strong>and</strong> procedure has become major<br />

*Email: guruh.samodra@gmail.com<br />

1<br />

- 520 -


tool <strong>and</strong> trending topic in order to figure out place-based vulnerability analysis. However, there are several<br />

major considerations should be underlined in the development <strong>of</strong> spatial vulnerability analysis. Scale<br />

analysis <strong>and</strong> the robustness <strong>of</strong> method are important aspect to employ GIS technique in vulnerability<br />

assessment.<br />

The scale analysis <strong>of</strong> vulnerability assessment is usually driven by the availability <strong>of</strong> data <strong>and</strong> usually<br />

applied in site specific location or over to the large area. It was usually based on the analysis <strong>of</strong> secondary<br />

traditional vulnerability data based on administrative unit i.e. age, gender, literacy, <strong>and</strong> population. The main<br />

drawback <strong>of</strong> this procedure is the unmatched <strong>of</strong> administrative data <strong>and</strong> the boundary <strong>of</strong> study area. For<br />

instance, the problem arises when dem<strong>and</strong>ed on the analysis <strong>of</strong> catchment area in which the boundaries <strong>of</strong><br />

catchment is oversize or undersize with the administrative unit. Due to lack <strong>of</strong> data on the household<br />

characteristic, the vulnerability per settlement area is also difficult to be estimated by administrative based<br />

data. The choice in the scale analysis <strong>of</strong> observation <strong>and</strong> <strong>its</strong> spatial unit is necessary in spatial vulnerability<br />

assessment (Fekete, 2010).<br />

To overcome the limitation <strong>of</strong> administrative data, the socio-<strong>economic</strong> vulnerability data should be<br />

collected by direct survey (primary data). Those are only available obtained through census or <strong>and</strong><br />

community based methods. However, data collected using house-to-house surveys <strong>and</strong> community based<br />

methods have limited efficiency <strong>and</strong> transferability (Ebert et al., 2009). Medium scale vulnerability<br />

assessment based on catchment analysis focused on household level is challenging in order to breakdown the<br />

drawbacks <strong>of</strong> traditional spatial vulnerability assessment. It will be very valuable to explain the spatial<br />

interdependencies <strong>of</strong> l<strong>and</strong>slide phenomenon in the framework <strong>of</strong> environmental system. Sampling technique<br />

was proposed in order to overcome timely <strong>and</strong> costly census procedure. However, the limitation <strong>of</strong> the<br />

sampling technique is related to the generalization <strong>of</strong> spatial unit attribute. Limited sample will only describe<br />

a small portion <strong>of</strong> whole spatial unit attribute in the entire the study area. Generalization <strong>of</strong> the sample data<br />

should fulfill the requirement <strong>of</strong> vulnerability information in each spatial unit in the whole study area. It<br />

requires an underst<strong>and</strong>ing <strong>of</strong> relationship between vulnerability information <strong>and</strong> the environment factors<br />

affecting them. <strong>Spatial</strong> data analysis focusing on detecting pattern is a capable tool to underst<strong>and</strong> processes<br />

which are responsible for observed patterns (Fischer, 2002).<br />

<strong>Spatial</strong> pattern analysis focuses on the describing space as an important role to derive socio-<strong>economic</strong><br />

process in the space-time context. It is powerful analysis to recognize phenomenon related to the spatial<br />

interaction among observed field. Household sample is derived to draw inferences about underlying process<br />

<strong>of</strong> socio-<strong>economic</strong> vulnerability <strong>and</strong> <strong>its</strong> environment based on the settlement block spatial unit. Average<br />

nearest neighbor analysis was employed in order to figure out the spatial pattern <strong>of</strong> socio-<strong>economic</strong> l<strong>and</strong>slide<br />

vulnerability. Geographic distribution <strong>of</strong> mean centre analysis, st<strong>and</strong>ard distance <strong>and</strong> directional distribution<br />

were also employed in order to identify spatial tendency <strong>and</strong> spatial distribution <strong>of</strong> socio-<strong>economic</strong><br />

vulnerability. It was intended to explore the interdependencies between socio-<strong>economic</strong> l<strong>and</strong>slide<br />

vulnerability with <strong>its</strong> environmental factor. Then, the spatial pattern analysis was applied as an input <strong>of</strong> fuzzy<br />

membership technique to predict socio-<strong>economic</strong> l<strong>and</strong>slide vulnerability in the whole settlement block <strong>of</strong><br />

Kayangan Catchment. Therefore, the objective <strong>of</strong> this paper is to propose medium scale spatial vulnerability<br />

analysis based on the socio-<strong>economic</strong> indicators through fuzzy logic technique.<br />

2. STUDY AREA<br />

The research area was conducted in Kayangan Catchment Kulon Progo, Yogakarta Indonesia (Figure 1).<br />

The area <strong>of</strong> Kayangan Catchment extends 4 sub-districts i.e. Girimulyo, Nanggulan, Samigaluh <strong>and</strong><br />

Kaligesing. It is located in the middle Java Isl<strong>and</strong> <strong>and</strong> comprises 35 km 2 . The average annual rainfall in<br />

Kayangan Catchment is 2478 mm. The highest rainfall intensity usually occurs from February to March with<br />

average monthly rainfall 426 mm. <strong>L<strong>and</strong>slide</strong> usually occurs in the month <strong>of</strong> November to April during wet<br />

season. L<strong>and</strong> use in the study area can be classified into bushes, rain fed paddy field, irrigated paddy field,<br />

people forest, settlement, <strong>and</strong> dry cultivated l<strong>and</strong>.<br />

Kulon Progo area has been traditionally dominated by agricultural sector. It is also the second lowest<br />

<strong>economic</strong> growth <strong>and</strong> the welfare level has only 8.7% <strong>of</strong> total GDP among 5 regencies in Yogyakarta<br />

Province. Around 78% household working in agricultural sector <strong>and</strong> mostly (88%) are living in rural areas.<br />

With the limitation <strong>of</strong> <strong>economic</strong> development <strong>and</strong> infrastructure, 40.31 % people are living below national<br />

poverty line (BPS, 2008). Due to inadequate access <strong>and</strong> resource to <strong>economic</strong> development, it may increase<br />

socio-<strong>economic</strong> vulnerability level when disaster occurs.<br />

2<br />

- 521 -


Figure 1 Location <strong>of</strong> Study Area, Kayangan Catchment, Indonesia.<br />

3. METHODOLOGY<br />

3.1. <strong>Vulnerability</strong> Indicators<br />

<strong>Vulnerability</strong> indicators play an important role to vulnerability analysis. It should be in line with the<br />

<br />

characteristics <strong>of</strong> person or group in terms <strong>of</strong> their capacity to anticipate, cope with, resist <strong>and</strong> recover from<br />

<br />

with the three pillars <strong>of</strong> sustainable development which is pointed out to reduce vulnerability in the context<br />

<strong>of</strong> social, <strong>economic</strong> <strong>and</strong> environmental spheres (Birkmann, 2006). It explains that the vulnerability is<br />

interlinked with the individual, household or community ability to recover from disaster by strengthening the<br />

exposure, coping capacity <strong>and</strong> resilience. The complex nature <strong>of</strong> people, household, social structure <strong>and</strong><br />

culture are amongst the important features that determine how well selected indicators represent socio<strong>economic</strong><br />

l<strong>and</strong>slide vulnerability. Therefore several indicators were employed to describe exposure, coping<br />

capacity <strong>and</strong> resilience in Kayangan Catchment.<br />

Exposure is derived from building material, construction type, building age, infrastructures, <strong>and</strong> building<br />

location. Coping capacity described as capability to cope with disaster is derived from preparedness,<br />

knowledge <strong>of</strong> l<strong>and</strong>slide, age <strong>and</strong> gender. Resilience which is generally described as an ability <strong>of</strong> household to<br />

recover from disaster is derived from income, saving <strong>and</strong> building architecture representing the <strong>economic</strong><br />

welfare <strong>of</strong> a household. The indicators <strong>of</strong> <strong>economic</strong> welfare were intended to explore the <strong>economic</strong> ability <strong>of</strong><br />

household to refinance the impact <strong>of</strong> natural disaster. Several closed questions were employed to gain indepth<br />

information <strong>of</strong> preparedness <strong>and</strong> knowledge <strong>of</strong> l<strong>and</strong>slide. For instance as preparedness <strong>of</strong> evacuation<br />

route, emergency needs, important phone number, important document, participation <strong>of</strong> l<strong>and</strong>slide<br />

education/socialization, etc.<br />

The sampling technique <strong>and</strong> data mining <strong>of</strong> socio-<strong>economic</strong> l<strong>and</strong>slide vulnerability were conducted by<br />

household survey using observation, photograph analysis <strong>and</strong> interview. Household survey was conducted<br />

by transect walk <strong>and</strong> structured (closed question) interview by using questionnaire. Photograph analysis <strong>of</strong><br />

building was also used to accompany the analysis <strong>of</strong> household survey to portray the building architecture.<br />

Samples <strong>of</strong> 112 respondents were successfully aided to derive socio-<strong>economic</strong> vulnerability. Weighting<br />

3<br />

- 522 -


multiple indicators <strong>of</strong> exposure, coping capacity <strong>and</strong> resilience was applied to derive final consideration <strong>of</strong><br />

l<strong>and</strong>slide vulnerability. Finally, it was classified into low, medium <strong>and</strong> high level <strong>of</strong> vulnerability (Table 1).<br />

Table 1 <strong>L<strong>and</strong>slide</strong> <strong>Vulnerability</strong> Indicators<br />

No. <strong>Vulnerability</strong> Exposure Indicators Coping Capacity Indicators Resilience Indicators<br />

Classification<br />

1. Low 1. Building Material:<br />

traditional brick with<br />

reinforced concrete column<br />

1. Age: dominated by 15-<br />

64 years old<br />

2. Gender: dominated by<br />

1. Income:<br />

>1.500.000 IDR<br />

2. Saving: yes<br />

2. Construction type:<br />

engineered-semi engineered<br />

male<br />

3. Preparedness: well<br />

3. Building<br />

material:<br />

3. Building age: (10 m- 50m)<br />

timber<br />

3. High 1. Building Material: bamboo<br />

<strong>and</strong> timber<br />

2. Construction type: non<br />

engineered<br />

3. Building age: >30 years<br />

4. Infrastructure (distance to<br />

road): low accessibility (>50<br />

m)<br />

1. Age: dominated by 64<br />

2. Gender: dominated by<br />

female<br />

3. Preparedness: poorly<br />

prepared<br />

4. Knowledge <strong>of</strong> l<strong>and</strong>slide:<br />

no understood<br />

1. Income: <<br />

750.000<br />

2. Saving: no<br />

3. Building<br />

material: bamboo<br />

<strong>and</strong> timber<br />

3.2. <strong>Spatial</strong> <strong>Pattern</strong> Analysis<br />

<strong>Spatial</strong> pattern analysis is employed to describe exploratory point sampling data <strong>of</strong> socio-<strong>economic</strong><br />

vulnerability. The data comprises the spatial coordinate data represented in two dimensionally space <strong>and</strong> the<br />

categorical level <strong>of</strong> socio-<strong>economic</strong> vulnerability. Nearest neighborhood, spatial mean centre, st<strong>and</strong>ard<br />

distance <strong>and</strong> directional distribution (st<strong>and</strong>ard deviation ellipse) were employed to show the proximity <strong>of</strong><br />

particular spatial point location in relation with <strong>its</strong> environmental spheres.<br />

Nearest neighborhood technique was employed in order to explore whether the point <strong>of</strong> categorical<br />

socio-<strong>economic</strong> vulnerability looked to be clustered or dispersed visually. It measures the ratio between the<br />

observed distances <strong>of</strong> each point toward <strong>its</strong> nearest point neighbor <strong>and</strong> the expected mean distance for the<br />

point given in a r<strong>and</strong>om pattern. If the ratio is less than 1, it means that the exploratory point sampling seems<br />

to be clustered pattern. In the other h<strong>and</strong>, dispersed pattern is described by the ratio more than 1.<br />

Central feature analysis estimates the average central point location <strong>of</strong> a set <strong>of</strong> particular spatial point<br />

location. It measures the mean centre <strong>of</strong> x-coordinate <strong>and</strong> the mean centre <strong>of</strong> y-coordinate. Mean centre<br />

analysis is useful to show the concentration <strong>of</strong> each categorical socio-<strong>economic</strong> l<strong>and</strong>slide. Therefore, the<br />

tendency <strong>of</strong> the spatial distribution can also be addressed as well. St<strong>and</strong>ard distance describes the<br />

compactness or spatial deviation <strong>of</strong> the point analysis toward <strong>its</strong> central features. It seems likely to be<br />

st<strong>and</strong>ard deviation in traditional statistic. In GIS platform, it is represented by circle feature which has the<br />

radius equal to the st<strong>and</strong>ard distance value. Directional distribution is rather similar with st<strong>and</strong>ard distance<br />

circle. St<strong>and</strong>ard distance circle calculate the st<strong>and</strong>ard distance simultaneously in the x <strong>and</strong> y direction.<br />

However, st<strong>and</strong>ard distance circle cannot pose the directional bias in point distribution. In the directional<br />

distribution, the x <strong>and</strong> y direction is calculated separately. It calculates the st<strong>and</strong>ard deviation <strong>of</strong> the x <strong>and</strong> y<br />

coordinates from the mean centre. The result is axes <strong>of</strong> an ellipse or st<strong>and</strong>ard deviation <strong>of</strong> ellipse which is<br />

drawn as an ellipse feature. Thus, by those techniques, the spatial characteristic <strong>of</strong> categorical point socio<strong>economic</strong><br />

vulnerability such as central tendency, dispersion <strong>and</strong> directional trends can be posed easily.<br />

3.3. Fuzzy Logic Technique<br />

The fuzzy logic technique was employed in order to consider the environmental factor <strong>of</strong> socio<strong>economic</strong><br />

l<strong>and</strong>slide vulnerability represented in a map as members <strong>of</strong> a set. It defines the design <strong>of</strong> a<br />

membership function that expresses the degree <strong>of</strong> membership in respect to the attribute <strong>of</strong> interest. For<br />

4<br />

- 523 -


example, based on the spatial pattern analysis, the environmental factor <strong>of</strong> l<strong>and</strong>slide vulnerability represented<br />

by an area could be areas on the map defined as areas highly vulnerable to l<strong>and</strong>slide. In this paper, slope <strong>and</strong><br />

distance to road were chosen as predictor variable to extrapolate the spatial socio-<strong>economic</strong> vulnerability <strong>of</strong><br />

l<strong>and</strong>slide. As a geographic phenomenon, vulnerability can be spatially referenced, analyzed <strong>and</strong> mapped<br />

using geospatial tools. Empirical evidence through frequency distribution diagram showed that two<br />

geography features i.e. slope <strong>and</strong> distance to road were play important role for defining the socio-<strong>economic</strong><br />

vulnerability in Kayangan Catchment. The frequency distribution pattern <strong>of</strong> distance to road <strong>and</strong> slope<br />

toward the number <strong>of</strong> settlement in a given vulnerability degree was employed to determine fuzzy<br />

membership function. Thus, the final prediction <strong>of</strong> socio-<strong>economic</strong> vulnerability was calculated by fuzzy<br />

membership function <strong>and</strong> fuzzy operator.<br />

Fuzzy membership function is generated by relating map classes <strong>of</strong> environmental variable to<br />

membership value. The value <strong>of</strong> membership ranges from 0 to 1 representing the degree <strong>of</strong> certainty <strong>of</strong><br />

membership. An area <strong>of</strong> environmental factor will be given 1 if it has certainly a member <strong>of</strong> a set or will be<br />

given 0 if it has not a member <strong>of</strong> a set. Values are given based on the spatial pattern <strong>and</strong> the relationship<br />

between the particular spatial data interest with the environmental variable. Different with regression or<br />

logistic regression technique which is mainly data driven approach, the fuzzy logic technique give an<br />

approach in which an expert can be freely control the weighting process. The spatial pattern analysis <strong>of</strong><br />

socio-<strong>economic</strong> vulnerability samples towards <strong>its</strong> environmental factor predictor was employed to generate<br />

the membership function <strong>of</strong> each environmental predictor. In addition, it is also generated by function<br />

represented the relationship between environmental factor <strong>and</strong> socio <strong>economic</strong> vulnerability. Fuzzy<br />

membership linear representing the function <strong>of</strong> slope <strong>and</strong> fuzzy small representing the function <strong>of</strong> distance to<br />

road were employed in this research.<br />

Fuzzy operator was employed to combine or overlay the membership function map <strong>of</strong> each<br />

environmental factor. There are five operators on fuzzy technique as follows fuzzy <strong>and</strong>, fuzzy or, fuzzy<br />

product, fuzzy sum <strong>and</strong> fuzzy gamma operator. Fuzzy <strong>and</strong> operator sets the minimum value <strong>of</strong> the input<br />

membership as a result <strong>of</strong> overlay. It is defined as:<br />

µ combination = MIN (µ A, µ B, µ C, ) (1)<br />

In the other h<strong>and</strong>, fuzzy or operator sets the maximum value <strong>of</strong> the input membership as a result <strong>of</strong><br />

overlay. It is defined as:<br />

µ combination = MAX (µ A, µ B, µ ) (2)<br />

The fuzzy product works with multiplying the input <strong>of</strong> membership. It is defined as:<br />

<br />

µ combination = <br />

(3)<br />

The fuzzy sum works with adding the input <strong>of</strong> membership. It is defined as:<br />

<br />

µ combination = <br />

(4)<br />

The fuzzy gamma is algebraic product <strong>of</strong> fuzzy sum <strong>and</strong> fuzzy product. It is defined as:<br />

µ combination = (Fuzzy sum) * (Fuzzy product) 1- (5)<br />

Where µ combination is the calculated fuzzy membership function, µ A is the membership value for map A<br />

represented by a membership value <strong>of</strong> a pixel <strong>and</strong> µ B is the membership value for map B <strong>and</strong> so on. µ i is the<br />

fuzzy membership function for the i-th map <strong>and</strong> i=1, 2, n <br />

ship will be similar to fuzzy<br />

<br />

In this paper, only were employed.<br />

Fuzzy <strong>and</strong> & fuzzy product were not employed in this research. Fuzzy <strong>and</strong> was not employed because <strong>of</strong> the<br />

type <strong>of</strong> the overlay would return the minimum value <strong>of</strong> the pixel membership. Fuzzy product works by<br />

multiplying each <strong>of</strong> the fuzzy membership value. The value <strong>of</strong> fuzzy product overlay is always less than the<br />

membership value <strong>of</strong> the input therefore the value was very small. Therefore, both <strong>of</strong> those techniques were<br />

difficult to predict the socio-<strong>economic</strong> vulnerability based on the fuzzy membership <strong>of</strong> environmental factor<br />

predictors.<br />

5<br />

- 524 -


4. RESULT AND DISCUSSION<br />

4.1. Central Tendency <strong>and</strong> <strong>Spatial</strong> <strong>Pattern</strong> <strong>of</strong> <strong>Socio</strong>-<strong>economic</strong> <strong>L<strong>and</strong>slide</strong> <strong>Vulnerability</strong><br />

Since the nature <strong>of</strong> people, social structure <strong>and</strong> culture <strong>of</strong> Kayangan Catchment were less influenced by<br />

the development <strong>of</strong> technology <strong>and</strong> modernization, the natural feature would more dominantly influence the<br />

way <strong>of</strong> life <strong>of</strong> the people in Kayangan Catchment. The area preference <strong>of</strong> building settlement was one<br />

example <strong>of</strong> social behavior mainly influenced by natural feature <strong>and</strong> condition. There were several options<br />

for a household to build a settlement, but there was limited option for low income household. Settlement<br />

type was <strong>of</strong>ten a reflection <strong>of</strong> welfare level. It was usually that low income household could not choose<br />

whether they would build their house in more gentle or steep topography. Therefore, low income household<br />

tend to build settlement based on their l<strong>and</strong> given from their parent or ancestor. It was usually located in<br />

steeper, more remote <strong>and</strong> low accessibility. For high income household, there were several options to build a<br />

settlement. They usually build a settlement in relatively gentle topography <strong>and</strong> have good accessibility. Good<br />

accessibility was the most common reason where they would build a settlement. Even, with the reason <strong>of</strong><br />

accessibility, they would build a settlement by excavating <strong>and</strong> cutting the slope into more gentle topography.<br />

Low income household living in low accessibility <strong>and</strong> hilly area had low <strong>economic</strong> ability <strong>of</strong> household<br />

to refinance the impact <strong>of</strong> natural disaster. They usually depend on the agricultural product <strong>and</strong> have low<br />

education level. Besides living in hilly area which was more prone to be l<strong>and</strong>sliding, they also had high<br />

socio-<strong>economic</strong> vulnerability level. There were interlinked between natural feature, social behavior,<br />

<strong>economic</strong> level <strong>and</strong> vulnerability level. It would affect the spatial pattern <strong>of</strong> settlement in Kayangan<br />

Catchment.<br />

Figure 2 Sampling Distribution, Mean Center, St<strong>and</strong>ard Distance (Left)<br />

<strong>and</strong> Directional Distribution (Right) <strong>of</strong> <strong>Socio</strong>-<strong>economic</strong> <strong>Vulnerability</strong><br />

<strong>Spatial</strong> mean center (SMC) provides an analysis where a phenomenon tends to be centered. Figure 2<br />

shows sampling location <strong>of</strong> socio-<strong>economic</strong> vulnerability level <strong>and</strong> their spatial mean centers. It is interesting<br />

to note that low vulnerability SMC tends to be located in the lower stream <strong>of</strong> Kayangan Catchment which<br />

has more gentle topography <strong>and</strong> nearby the center <strong>of</strong> trade, education <strong>and</strong> local government <strong>of</strong>fice location.<br />

In the other h<strong>and</strong>, the SMCs <strong>of</strong> medium <strong>and</strong> high level <strong>of</strong> vulnerability tend to be located in the upper part <strong>of</strong><br />

Kayangan Catchment which has steeper topography. It means that socio-<strong>economic</strong> vulnerability is spatially<br />

associated with topography <strong>and</strong> socio-<strong>economic</strong> functional unit.<br />

St<strong>and</strong>ard distance (SD) was employed to measure the distribution <strong>of</strong> socio-<strong>economic</strong> vulnerability<br />

deviate from their SMC. Figure 2 (left) shows the spatial SD circle <strong>of</strong> socio-<strong>economic</strong> vulnerability. The low<br />

vulnerability level was the highest deviation from the spatial mean <strong>and</strong> the high vulnerability level was the<br />

lowest one. However, high vulnerability <strong>and</strong> medium vulnerability had no significance difference. It implied<br />

that low vulnerability level locations were highly dispersed. High <strong>and</strong> medium vulnerability levels were not<br />

so much deviated from their mean centre. High <strong>and</strong> medium vulnerability levels were mainly located on the<br />

hilly area <strong>and</strong> the boundary <strong>of</strong> deviation was not exceeding the hilly area <strong>of</strong> Kayangan Catchment.<br />

6<br />

- 525 -


Since Kayangan Catchment form has unsymmetrical width, SD has not fitted to represent the directional<br />

deviation <strong>of</strong> socio-<strong>economic</strong> vulnerability. Directional distribution or St<strong>and</strong>ard Deviational Ellipse (SDE)<br />

was employed to measure directional deviation <strong>of</strong> socio-<strong>economic</strong> vulnerability. The calculation <strong>of</strong> SDE<br />

made the trend <strong>of</strong> socio-<strong>economic</strong> vulnerability distribution more clear. It pointed out that low vulnerability<br />

level had highest deviation along the major axis <strong>and</strong> covered flat area. Medium vulnerability level SDE had<br />

longer major axis than high vulnerability level. It indicated that high vulnerability level tend to be<br />

concentrated on the hilly area. This implied that slope had significant role on socio-<strong>economic</strong> vulnerability in<br />

Kayangan Catchment.<br />

Nr. <strong>of</strong> Settlement Block<br />

35<br />

30<br />

25<br />

20<br />

15<br />

10<br />

5<br />

0<br />

0 50 100 150 200 250 300 350<br />

Distance to Road<br />

Nr. <strong>of</strong> Settlement Block<br />

45<br />

40<br />

35<br />

30<br />

25<br />

20<br />

15<br />

10<br />

5<br />

0<br />

0 10 20 30 40 50 60<br />

Slope (Degree)<br />

: Low <strong>Vulnerability</strong> : Medium <strong>Vulnerability</strong> : High <strong>Vulnerability</strong><br />

Figure 3 the <strong>Vulnerability</strong> Level based on number <strong>of</strong> settlement towards distance to road <strong>and</strong> Slope<br />

Furthermore, nearest neighbor analysis was employed to analyze the spatial pattern <strong>of</strong> socio<strong>economic</strong><br />

vulnerability. Nearest neighbor analysis <strong>of</strong> low, medium <strong>and</strong> high vulnerability level<br />

showed z-scores -4.52, -5.26 <strong>and</strong> -1.86 respectively. It implied that the spatial distribution <strong>of</strong> each<br />

level <strong>of</strong> vulnerability was clustered. Low level vulnerability household was clustered in more gentle<br />

area <strong>and</strong> high accessibility. However, low level vulnerability had higher tendency on deviation due<br />

to the high result <strong>of</strong> SD <strong>and</strong> SDE. In the other h<strong>and</strong> high level vulnerability household was clustered<br />

in the hilly <strong>and</strong> low degree <strong>of</strong> accessibility. Medium <strong>Vulnerability</strong> was also clustered in the hilly<br />

area but the accessibility is higher than low vulnerability level. Figure 3 showed the interlinked<br />

between vulnerability level <strong>and</strong> <strong>its</strong> environmental factor i.e. slope <strong>and</strong> accessibility (distance to<br />

road). Later, the trend <strong>of</strong> the interlinked between vulnerability level <strong>and</strong> <strong>its</strong> environmental was<br />

employed as an input <strong>of</strong> fuzzy membership in fuzzy logic technique.<br />

4.2. <strong>Spatial</strong> <strong>L<strong>and</strong>slide</strong> <strong>Vulnerability</strong> Prediction by means <strong>of</strong> Fuzzy Logic Technique<br />

<strong>Spatial</strong> pattern analysis with household survey described the distribution <strong>of</strong> socio-<strong>economic</strong><br />

vulnerability level in Kayangan Catchment quite well. However, the vulnerability level location<br />

obtained from household survey could not be described spatially on each settlement block due to<br />

the limitation <strong>of</strong> the samples. Fuzzy logic technique based on the slope <strong>and</strong> distance to road<br />

membership function was introduced to extrapolate vulnerability degree in each settlement block.<br />

Fuzzy membership function (Figure 3) <strong>and</strong> the fuzzy operator (Eqs. 1, 4 <strong>and</strong> 5) were generated<br />

to predict the socio-<strong>economic</strong> vulnerability level in Kayangan Catchment. It was computed for 6<br />

cases i.e. Fuzzy operator Or, Sum, Gamma 0.975, gamma 0.9, gamma 0.8 <strong>and</strong> gamma 0.7. The<br />

result <strong>of</strong> computation was mapped in order to interpret the spatial distribution <strong>of</strong> socio-<strong>economic</strong><br />

vulnerability in Kayangan Catchment. <strong>Socio</strong>-<strong>economic</strong> vulnerability map was represented by the<br />

possibility value between 0 <strong>and</strong> 1. The value <strong>of</strong> 0 indicated the full impossibility vulnerable <strong>and</strong> the<br />

value <strong>of</strong> 1 indicated the full possibility <strong>of</strong> being vulnerable. Thus, an area given as a value<br />

approaching 1 was indicated higher socio-<strong>economic</strong> vulnerability level. The final maps were<br />

presented in Figure 4 representing spatial distribution <strong>of</strong> socio-<strong>economic</strong> vulnerability level in<br />

Kayangan Catchment.<br />

7<br />

- 526 -


(a)<br />

(b)<br />

(c)<br />

(d)<br />

(e<br />

(e)<br />

Figure 4 <strong>Socio</strong>-<strong>economic</strong> <strong>Vulnerability</strong> Map based on Fuzzy Operator Or (a), Sum (b), Gamma 0975 (c), gamma 09 (d),<br />

gamma 08 (e), gamma 07 (f)<br />

(f)<br />

8<br />

- 527 -


There were significant different amongst cases <strong>of</strong> fuzzy operators. Fuzzy membership <strong>and</strong><br />

fuzzy operator play important role to generate final map <strong>of</strong> socio-<strong>economic</strong> l<strong>and</strong>slide vulnerability.<br />

It implied that fuzzy operator Or, Sum <strong>and</strong> high gamma value represented better result than fuzzy<br />

operator product, fuzzy operator <strong>and</strong> <strong>and</strong> fuzzy operator with low gamma value. Since the fuzzy<br />

operator product, <strong>and</strong> <strong>and</strong> low gamma value working with returning the minimum value <strong>of</strong> the pixel<br />

membership <strong>and</strong> the result was always less than the membership value <strong>of</strong> the input, the result <strong>of</strong> the<br />

map will be low vulnerable in the whole area. Thus, fuzzy operator product, fuzzy operator <strong>and</strong> <strong>and</strong><br />

fuzzy operator with low gamma value were not included in the analysis.<br />

Figure 5 (a) <strong>Socio</strong>-<strong>economic</strong> <strong>Vulnerability</strong> Map <strong>of</strong> Kayangan Catchment (b) High Accessibility <strong>and</strong> Centre <strong>of</strong><br />

Activities(c) Low <strong>Vulnerability</strong> Household (d) Bad Road Condition Showing Low Accessibility<br />

(e) High <strong>Vulnerability</strong> Household in Hilly area<br />

The result map <strong>of</strong> fuzzy operator Or, Sum <strong>and</strong> high gamma value tend to have similar trend<br />

describing socio-<strong>economic</strong> vulnerability level. High vulnerability level was located in the middle<br />

part <strong>of</strong> Kayangan Catchment in which covered by steep topography <strong>and</strong> low accessibility. The<br />

result <strong>of</strong> field observation showed that the settlement pattern <strong>of</strong> high vulnerability level tend to be<br />

clustered <strong>and</strong> was characterized by semi-permanent building constructed by non-engineered<br />

bamboo <strong>and</strong> timber.<br />

The clustered pattern would depend on the natural resource. A single clustered village was<br />

usually located close to water resource (spring water). It was due to water supply in the hilly area<br />

which fully depends on the availability <strong>of</strong> spring water. It is sometime also used to irrigate rain fed<br />

9<br />

- 528 -


paddy field. Agricultural activity was dominant activity in the hilly area. Almost 60% <strong>of</strong> the<br />

respondents in the hilly area worked in agricultural sector.<br />

The second type <strong>of</strong> settlement pattern was clustered linear village settlement around road. It was<br />

usually characterized by engineered-semi engineered traditional brick with reinforced concrete<br />

column. Most <strong>of</strong> them were working in service sector <strong>and</strong> categorized as medium to high household<br />

income. They were mostly educated people <strong>and</strong> had more capability to cope with disaster. Besides<br />

<strong>of</strong> livestock as a saving system, they also saved their money in the bank. This social behavior was<br />

totally different with the people living in the hilly <strong>and</strong> remote area that had livestock only for their<br />

saving. This type <strong>of</strong> settlement pattern was concentrated in the western part <strong>and</strong> south-eastern part<br />

<strong>of</strong> Kayangan Catchment which were dominated by low topography. This place can be considered as<br />

a central <strong>economic</strong>, education, <strong>and</strong> local government activity. Those were also classified as low to<br />

medium socio-<strong>economic</strong> vulnerability level. Both spatial settlement patterns were illustrated in<br />

Figure 5.<br />

5. CONCLUSION<br />

<strong>Spatial</strong> pattern analysis had play important role to analyze socio-<strong>economic</strong> vulnerability in the rural area.<br />

Classified as a rural area, Kayangan Catchment is less experienced with agricultural modernization,<br />

industrialization <strong>and</strong> deindustrialization, <strong>and</strong> commodification <strong>of</strong> rural l<strong>and</strong>scapes. Thus, it reflects the<br />

societal character <strong>of</strong> rural behavior including social structure <strong>and</strong> culture. Since the nature <strong>of</strong> people, social<br />

structure <strong>and</strong> culture <strong>of</strong> Kayangan Catchment were less influenced by the development <strong>of</strong> technology <strong>and</strong><br />

modernization, the natural feature would more dominantly influence the socio-<strong>economic</strong> vulnerability in<br />

Kayangan Catchment. Determining environmental factors influencing socio-<strong>economic</strong> vulnerability was very<br />

helpful to predict spatial distribution <strong>of</strong> socio-<strong>economic</strong> vulnerability through fuzzy logic technique. Fuzzy<br />

membership function is generated by relating map classes <strong>of</strong> environmental variable to membership value.<br />

The fuzzy operator Or, Sum <strong>and</strong> high gamma value were successfully applied to predict spatial distribution<br />

<strong>of</strong> socio-<strong>economic</strong> vulnerability level. The result coincided with the <strong>economic</strong>, social <strong>and</strong> ecological<br />

dimensions <strong>of</strong> livelihood in Kayangan Catchment.<br />

REFERENCES<br />

(UNISDR) United Nations International Strategy for Disaster Reduction Secretariat. 2007. Hyogo Framework for Action 2005-2015:<br />

Building the Resilience <strong>of</strong> Nations <strong>and</strong> Communities to Disasters. United Nations International Strategy for Disaster<br />

Reduction Secretariat, Geneva.<br />

Birkmann, J. 2006. Measuring <strong>Vulnerability</strong> to Natural Hazards: towards Disaster Resilient Sicieties. United Nation University<br />

Press, Tokyo.<br />

BPS (Badan Pusat Statistik). 2008. Kulon Progo Regency in Figures 2008. BPS. Yogyakarta<br />

Costa, L <strong>and</strong> Kropp J, P. 2012. Linking components <strong>of</strong> <strong>Vulnerability</strong> in Theoretic Frameworks <strong>and</strong> Case Studies. Sustain Sci DOI<br />

10.1007/s11625-012-0158-4<br />

Ebert, A., Kerle, N., & Stein, A. 2009. Urban Social <strong>Vulnerability</strong> Assessment with Physical Proxies <strong>and</strong> <strong>Spatial</strong> Metrics derived<br />

from Air- <strong>and</strong> Spaceborne Imagery <strong>and</strong> GIS Data. Nat Hazards 48:275294.<br />

Fekete, A. 2011. <strong>Spatial</strong> Disaster <strong>Vulnerability</strong> <strong>and</strong> Risk Assessment: Challenges in Their Quality <strong>and</strong> Acceptance. Nat Hazards<br />

61:1161-1178<br />

Fekete, A., Damm, M., Birkmann, J. 2010. Scales as Challenge for <strong>Vulnerability</strong> Assessment. Nat Hazards 55 (3):729-747<br />

Fischer, M. M. 2002. <strong>Spatial</strong> Analysis in Geography. International Encyclopedia <strong>of</strong> the Social & Behavioral Sciences 14752-14758<br />

Kaynia, A. M., Papathoma-Kohle, M., Neuhauser, B., Ratzinger, K., Wenzel, H., Medina-Cetina, Z. 2008. Probabilistic Assessment<br />

<strong>of</strong> <strong>Vulnerability</strong> to <strong>L<strong>and</strong>slide</strong>: Application to the Village <strong>of</strong> Lichtenstein, Baden-Wurttemberg, Germany. Engineering<br />

Geology 101:33-48.<br />

Kuhlicke, C., Scolobig, A., Tapsell, S., Steinfuhrer, A., De Marchi, B. 2011. Contextualizing Social <strong>Vulnerability</strong>: Finding from case<br />

Studies Across Europe. Nat Hazards 58:789-810<br />

Menoni, S., Molinari, D., Parker, D., Ballio, F., Tapsell. 2012. Assessing Multifaceted <strong>Vulnerability</strong> <strong>and</strong> Resilience in order to<br />

Design risk-mitigation Strategies. Nat Hazards DOI 10.1007/s11069-010-9666-7<br />

Wolf, S. 2012. <strong>Vulnerability</strong> <strong>and</strong> Risk: Comparing Assessment Approaches. Nat Hazards 61: 1099-1113.<br />

10<br />

- 529 -

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!