Jakarta Land Subsidence and Inundation Vulnerability Based on SAR Data Agustan1, Hartanto Sanjaya1, Takeo Ito2 1
Center of Technology for Natural Resources Inventory (PTISDA) Agency for the Assessment and Application of Technology (BPPT) BPPT 1st Bulding, 20th Floor, Jl. M.H. Thamrin No. 8 Jakarta, Indonesia Email:
[email protected] [email protected] 2
Graduate School of Environmental Studies, Nagoya University Furo-cho, Chikusa-ku, Nagoya – Japan Email:
[email protected]
Abstract: Heavy rains in mid-January 2013 caused terrible floods in Jakarta, the capital city of Indonesia. The floods inundated about 41 square kilometers, or eight percent of Jakarta. It is less than the 2007 event where floods inundated 231 square kilometer, however, the total material losses of the 2013 event was much higher, reaching 32 trillion rupiahs in 2013 compared to 4.3 trillion rupiahs in 2007. Apart from regular flooding, Jakarta is facing a condition of high rate land subsidence. Land subsidence is a hazard resulting in negative impacts and could lead to serious problems, such as increasing risk of flooding, cracking the buildings and infrastructures, destructing local groundwater systems, and generating tension cracks on land and reactivating faults. This research describes the relation of inundation vulnerability with land subsidence in Jakarta based on Synthetic Aperture Radar (SAR) data. Land subsidence is identified by implementing interferometry (InSAR) technique of ALOSPALSAR data using GAMMA SAR software, whereas the inundated area of January event is revealed from RADARSAT data using NEST software. The principle behind using radar for flood mapping is that the strength of the radar backscatter depends on the roughness of the surface it interacts with. Water bodies generally create a very smooth and homogenous surface so the radar image over that area will be black or dark. To distinguish flooded areas from permanent water bodies, an image acquired during or after the flood is compared with another image acquired before the flood. Both images need to have been acquired from the same sensor geometry or orbit direction. By processing four years time span ALOS-PALSAR data, it is found that long-term observation gives clear information of land subsidence in Jakarta and tends to increase in term of area width and number of locations. By comparing RADARSAT data before and after the peak of heavy rain, it is found that areas affected by land subsidence were also inundated in longer time. Keywords: SAR, land subsidence, inundation, Jakarta 1. Introduction As a capital city with more then ten millions population, Jakarta faces a serious problem related to disaster i.e. flood during rainy season. Flood history in Jakarta is recorded on 1979, 1996, 1999, 2002, 2007, and 2013. During the flood event, inundated area usually occurred on the low elevated area that was estimated to cover 40% of Jakarta’s total area or approximately 780 km2. In 2007, the flood cause 4.3 trillion rupiahs with 57 casualties (Sagala et al., 2010), whereas the 2013 event cause 41 casualties with 32 trillion rupiahs as summarized in Table 1.
!
"!
Table 1. Summary of Jakarta Flood Events (Modified from Sagala et al., 2013, BNPB and OCHA, 2013) Variables Causes
1996 event The capacity of river is smaller than the incoming water’s runoff. Low stage of river capacity and major canal are caused by the high conversion of area around these rivers and canals into settlement function, sedimentation, and illegal waste disposal
2002 event Land use in urban areas which lots of buildings and settlements has led to the decreasing of land absorption ability as well as narrowing the river canal in downstream area.
2007 event Beside of poor drainage system, flood was preceded by heavy rain from afternoon on Feb 1 to the next day on Feb 2. It was worsened by the high volume of water in 13 rivers in Jakarta which originated from Bogor – Puncak – Cianjur and the tide of Jakarta’s sea water.
2013 event Flood in Jan 2013 was less intense compared to 2007 but the rainfall was widely distributed upstream and downstream. 17th Jan was considered the worst but it is predicted to be continued and the city should be put on alert level 1 till 27th Jan.
Inundation Point
90
160
70
109 (Google crisis map)
Rainfall Intensity
288.7 mm
361.7 mm
401.5 mm (geographically concentrated)
40-125 mm (tentative) (geographically widely distributed)
Evacuee (people)
30 thousand
380 thousand
398 thousand
>100 thousands (tentative)
Dead (people)
10
22
57
41
Losses
> IDR 1 Billion
IDR 1.8 Trillion
IDR 4.3 Trillion
>IDR 32 Trillion
The inundated and vulnerable area due to the increasing rainfall rate is identified in the same location from one event to another, mainly in the western region of Jakarta as illustrated in Figure 1. During the 2013 event, some locations are identified as suffering from longer inundated time. One factor that increases the inundated area is land subsidence factor (Murakami et al., 2012).
Figure 1. Inundated Area due to the 2013 Jakarta flood event (Source: BNPB Map)
!
#!
Land subsidence is generally related to geological subsidence i.e. sediment consolidation due to its own weight and tectonic movements; or related to human activities such as withdrawal of ground water and geothermal fluid, oil and gas extraction from underground reservoirs, and collapse of underground mines (Raucoules et al., 2007; Yuill et al., 2009). The amount of subsidence or uplift can be estimated from the number of concentric fringes that appear in the interferogram as illustrated in Figure 2. Land subsidence in Jakarta had been observed by utilizing geodetic observation i.e. by Global Positioning System (GPS) observation technique (Abidin et al., 2008) and by interferometric synthethic aperture radar (InSAR) technique (Abidin et al., 2011; Ng et al., 2012; Chaussard et al., 2013). All of these previous studies concluded that Jakarta suffer a sinking phenomena due to its rapid subsidence rate, approximately 260 mm/year in northern part of Jakarta. This research describes the relation of inundation vulnerability with land subsidence in Jakarta based on Synthetic Aperture Radar (SAR) data. 2. Data and Methods To understand the expansion of subsided region in Jakarta, InSAR technique is applied to identify the location in regional scale. Eight (8) ALOS-PALSAR (Advanced Land Observing Satellite with Phased-Array Synthetic-Aperture Radar, L-band with wavelength ~ 23.6 cm) data in fine beam mode, ranging from 2007 to 2011 are processed with GAMMA SAR Software (Wegmuller and Werner, 1997). These data are obtained from Japan Aerospace Exploration Agency – Earth Observation Research Center (JAXA-EORC). However, since the ALOS satellite had stopped the mission on 2011, the inundated area of the 2013 Jakarta flood event is assessed by processing RADARSAT-2 data (C-band with wavelength ~ 5 cm), obtained from Canadian Space Agency (CSA) with NEST SAR toolbox software (Engdahl et al., 2012).
Figure 2. Phase difference of L-band satellite based InSAR technique due to land subsidence The ALOS-PALSAR data were observed on January 31st 2007, August 3rd 2007, September 18th 2007, February 3rd 2008, May 5th 2008, June 20th 2008, August 8th 2009, August 11th 2009, November 11th 2010 and February 11th 2011. The first epoch is set as the master, so the time difference for each interferograms are: 184, 230, 460, 506, 920, 1380 and 1472 days respectively. The RADARSAT data were observed on December 26th 2012, February 2nd 2013, and February 5th 2013. The data is observed in fine beam mode and in SAR Georeferenced Fine (SGF) format. SGF format indicates that the data is converted to ground range, is multilooked processed, and is oriented in the direction of orbit path. As stated on Table 1 before, the Jakarta 2013 flood event is mainly caused by two days high rainfall on January 17th to 18th 2013.
!
$!
The InSAR technique produce an interferogram which is obtained by crossmultiplying, pixel by pixel, of two SAR images. The SAR image can be seen as a mosaic of small picture elements (pixels) that corresponds to a small area of the Earth’s surface that can be defined as a resolution cell. Each pixel contains a complex number that carries amplitude and phase information about the microwave field backscattered by all objects in corresponding resolution cell projected on the ground. These kinds of information are stored in complex format by adapting IQ (In-phase and Quadrature) data format. Therefore, SAR image is also known as single look complex (SLC) that is composed of a regular grid with complex values or phasors (Hanssen, 2001) and can be decomposed into amplitude A or real R and phase " or imaginer I components.
y = Ae i"
!
Interferogram = y(master) " y(slave) Interferogram = AM AS e i(# M $# S ) %I( 4 + (RM $ RS ) # = # M $ # S = tan $1' * = $ & R) , # = # curv + # topo + # orb + # defo + # atm + # noise
#=
!
..........(1)
where: y is the SLC data that represents the electric field of a plane electromagnetic wave, A ! is amplitude of the !electromagnetic pulse, and " is phase angle. The amplitude represents the ! quantity of electromagnetic field scattered back grouped in each SAR image-sampling cell or ! pixel, whereas the phase represents an ambiguous measure of distance between sensor and ! each area on the ground corresponding to an image pixel (Raucoules et al., 2007).! The first SAR image is called ! master and the second one slave. As a result, the interferogram amplitude is the amplitude of master image multiplied by that of slave image, whereas the interferometric phase is the phase difference between the images.
4+ 4 +HB/ 4+ 4+ Bsin(- $ . ) + + B// + D + # atm + # noise , ,RM sin - , ,
........(2) .........(3) ........(4) .........(5) ........(6)
where, " is the interferometric phase or phase difference derived from master (M) and slave (S) images in one point, RM and RS are the geometric distances of satellite to target both ! acquisition times, and " is the SAR system wavelength. The distance between the two satellite positions is known as the baseline B , and can be decomposed into parallel baseline ! B// , the component along the radar’s line of sight; and perpendicular baseline B", the ! is perpendicular ! component which to the line of sight. Look angle in one point " and " is ! baseline, whereas D is possible deformation. angle of satellite ! has range pixel spacing of approximately 4.6 m, and 3.2 The ALOS PALSAR image m for the azimuth pixel spacing. Further, co-registration and resampling ! of slave image to ! master image to conform the geometry were performed. Then phase!co-registration by the range spectral shift and ! the azimuth common bandwidth filtering. It was found that the coregistration of the two images was 0.1. This indicated the geometry of both images was conformed in sub-pixel accuracy. This information ensured the high level of interferometric correlation. 3. Results and Discussion The interferogram was derived after the ALOS-PALSAR data was looked down into 4 and 6 in range and azimuth direction respectively. Hence, the interferogram pixel spacing was approximately 19 m in range and 19 m in azimuth. The Earth’s curvature interferogram was removed based on the first term of Equation (6). Since the perpendicular baseline for all interferograms was small (
