High Resolution Flood Hazard Mapping Using Two-Dimensional Hydrodynamic Model ANUGA: Case Study of Jakarta, Indonesia Hengki E. Putra, Dennish A. Putro, Tri W. Hadi, Edi Riawan, I D. G. Junnaedhi, Aditia Rojali, Fariza D. Prasetyo, Yudhistira S. Pribadi, Dita F. Andarini, Mila Khaerunisa, Raditya H. Prakoswa
Abstract— Catastrophe risk management can only be done if we are able to calculate the exposed risks. Jakarta is an important city economically, socially, and politically and in the same time exposed to severe floods. On the other hand, flood risk calculation is still very limited in the area. This study has calculated the risk of flooding for Jakarta using 2-Dimensional Model ANUGA. 2-Dimensional model ANUGA and 1-Dimensional Model HEC-RAS are used to calculate the risk of flooding from 13 major rivers in Jakarta. ANUGA can simulate physical and dynamical processes between the streamflow against river geometry and land cover to produce a 1 meter resolution inundation map. The value of streamflow as an input for the model obtained from hydrological analysis on rainfall data using hydrologic model HEC-HMS. The probabilistic streamflow derived from probabilistic rainfall using statistical distribution Log-Pearson III, Normal and Gumbel, through compatibility test using Chi Square and Smirnov-Kolmogorov. Flood event on 2007 is used as a comparison to evaluate the accuracy of model output. Property damage estimations were calculated based on flood depth for 1, 5, 10, 25, 50, and 100 years return period against housing value data from the BPSStatistics Indonesia, Centre for Research and Development of Housing and Settlements, Ministry of Public Work Indonesia. The vulnerability factor was derived from flood insurance claim. Jakarta's flood loss estimation for return period of 1, 5, 10, 25, 50 and 100 years, respectively are Rp 1.30 t; Rp 16.18 t; Rp 16.85 t; Rp 21.21 t; Rp 24.32 t; and Rp 24.67 t of the total value of building Rp 434.43 t.
Keywords—2D Hydrodynamic Model, ANUGA, Flood, Flood Modeling.
I. INTRODUCTION
J
AKARTA is an important city economically, socially, and politically in Indonesia. The Central Bureau of Statistics of Indonesia's data in 2015 shows that although Jakarta is only 0.035% of the area of Indonesia, it has the largest economic power than other provinces, which is 16% of the Gross Domestic Product of Indonesia [1]. In terms of general Hengki E. Putra is with the PT Reasuransi MAIPARK Indonesia, Jakarta, 12960 Indonesia (phone: (62-21) 2938-0088; fax: (62-21) 2938-0089; e-mail:
[email protected]). Tri W. Hadi is with Weather and Climate Prediction Laboratory, Institut Teknologi Bandung, Bandung, Indonesia (e-mail:
[email protected]) Aditia Rojali is with PT Inteligensi Risiko, Jakarta, 12190 Indonesia (phone: (62-21) 2955-7217; fax: (62-21) 2955-7218; e-mail:
[email protected]).
insurance, according to the Indonesian earthquake exposure data, Jakarta’s exposure is about 23% of the total exposure in Indonesia, with a value 24% of the total aggregate exposure value. These numbers will continue to grow in line with economic growth in Jakarta in 2015 amounted 5.11% [1]. Jakarta has numerous histories of major flood events. In the last decade, the floods hit Jakarta in 2007 and 2013. The National Development Planning Agency (BAPPENAS) reported financial loss due to floods in 2007 is about IDR 5.18 T (USD 518.4 M), approximately a quarter of the Jakarta Regional Budget (APBD) at that year. These floods also inundated of 20,000 ha (about 1/3 of total area of Jakarta) [2]. Potential losses due to flood events will remain and getting bigger in the future. Insurance is one of the main components in natural disaster risk management. With its role as a tool of risk transfer, insurance will increase community resilience for disaster. In Indonesia, practically, flood insurance is an addendum in general insurance policies, but the rate is only an additional loading, it is not risk reflective. This practice will become major problems in the future. Mistakes in managing natural disaster risk can lead to insolvency of insurance companies. Flood, as natural disaster, should be viewed as a dynamic probabilistic risk. Insurance companies may not rely only on historical data analysis. This study aimed to quantify the flood risk in Jakarta and will be used as the basis for flood insurance scheme. Flood risk is calculated based on Probabilistic Flood Hazard Maps (PFHM). PFHM obtained from hydrodynamic models that simulate flood events up to a certain return period. The hydrodynamic model ANUGA was used to generate inundation map. ANUGA model was developed by the Australian National University and Geoscience Australia. ANUGA was selected because it is an Open Source Software (FOSS), reliable to simulate the process of wetting and drying in the landscape and has the advantage to use rainfall as direct input or as a combination to hydrographic. Moreover, from a technical point for simple cases, ANUGA has the same ability to the other models such as MIKE, Sobek, and Infoworks [3].
II. DATA AND METHODS Generally, the framework of this study shows in Fig. 1. The input data can be classified into hydro-meteorology and geographic data. Hydro-meteorology data are rainfall and discharge data; the geographic data consist of topographic, soil type, land-use, and river network data. Topographic data is used to determine the water flow direction, both in hydrologic and hydraulics model. Land-use and soil type data are used as basic parameterization to calculate the flow rate of runoff. The rainfall data was transformed into discharge in each upstream for hydraulic model simulation purpose.
Fig. 2 Rainfall observation stations in 13 catchment areas around Jakarta. The red triangle is hydrological station and the red one is meteorological station
The statistical distribution methods such as Normal distribution, Log–Normal distribution and Log–Pearson III are used to construct design rainfall based on maximum daily rainfall data. The type of distribution that has been generated from the frequency analysis should be conducted fit tests using Chi-Square (x^2 Cr) and Kolmogorov-Smirnov (KS) method. The results of the fit test from both methods show that all types of distributions for each station in the study area are acceptable (Table I). Fig. 1 Research methodology
A. Rainfall Frequency Analysis Rainfall frequency analysis is used to determine the return period of maximum daily rainfall. Rainfall data is collected from the observation stations, which consist of 10 meteorological stations and 9 hydrological stations around Jakarta, as seen in Fig. 2. It represents the catchment areas of 13 rivers in Jakarta. This rainfall data is the daily maximum rainfall with minimum 10 years range of data.
TABLE I DISTRIBUTION TYPE AND FIT TEST OF 19 RAINFALL OBSERVATION STATIONS Distribution No. Station KS 𝑥 2 𝐶𝑟 Type 1 Jakarta (BMKG) Log Pearson III 15.444 0.004 2
Cengkareng
Log Normal
22.111
0.113
3
Halim
Log Pearson III
22.111
0.146
4
Tanjung Priok
Gumbel
16.778
0.0742
5
Serang
Log Pearson III
11.889
0.074
6
Tangerang
Log Pearson III
25.222
0.103
7
Pondok Betung
Log Pearson III
39.444
0.138
8
Citeko
Log Pearson III
22.556
0.101
9
Curug
Log Pearson III
20.778
0.088
10
Dermaga
Log Pearson III
24.778
0.207
11
Pasir Jaya
Log Pearson III
0.4
0.186
12
Ranca Bungur
Log Pearson III
2.0
0.113
13
Bendungan Pasar Baru
Log Pearson III
3.6
0.145
14
Log Pearson III
2.18
0.111
Normal
4.0
0.215
16
Cengkareng Drain Perkebunan Gunung Mas Gadog
Normal
0.4
0.089
17
Cibinong
Log Pearson III
3.636
0.057
18
Fakultas Teknik UI
Gumbel
1.273
0.092
19
Cawang
Log Pearson III
1.2
0.102
15
B. Hydrologic Analysis using HEC-HMS HEC-HMS was used to derive the rainfall-runoff hydrograph for 1, 5, 10, 25, 50, and 100 years return period based on design rainfall. HEC-HMS simulates the rainfallrunoff processes of dendritic watershed systems. Hydrologic elements are arranged in a dendritic network and computations are performed in an upstream-to-downstream sequence [4]. The physical representation of a watershed is accomplished with a basin model [5]. Basin model in HEC-HMS is set up for each sub-basin using two hydrologic elements: sub-basin and junction. Sub-basin element handles the infiltration loss, base-flow computations, and rainfall-runoff transformation process. Junction element handles the observed flow data and is mainly used for comparison between the observed flow hydrographs and the simulated flow hydrographs. The topographical data used in basin model is the Digital Elevation Model (DEM) of year 2000 with 1 m spatial resolution. This data is corrected through field surveys due to the changing of Jakarta’s topography since 2000. DEM and catchment area can be seen in Fig. 3.
The design rainfall for 1, 5, 10, 25, 50, and 100 years return period can be seen in Table II. TABEL II DESIGN RAINFALL FOR 19 RAINFALL OBSERVATION STATIONS* Return Period Station 1 2 5 10 25 50 100 Jakarta 56.44 118.08 171.40 213.32 274.80 327.05 385.39 (BMKG) Cengkareng 86.14 101.14 138.05 169.46 206.33 259.79 309.77 Halim 80.12 116.88 157.62 192.44 246.49 294.99 351.55 Tanjung 49.45 116.44 160.04 187.45 222.09 247.79 273.30 Priok Serang 52.77 85.18 105.62 119.35 136.92 150.25 163.80 Tangerang 64.31 101.90 140.08 171.96 220.67 263.80 313.53 Pondok 80.44 114.42 154.42 189.31 244.30 294.33 353.13 Betung Citeko 65.05 105.47 136.67 159.63 191.36 217.02 244.50 Curug 68.47 91.40 119.07 142.95 180.12 213.44 252.20 Dermaga 89.34 120.08 173.27 227.40 324.99 424.99 555.34 Pasir Jaya 82.08 126.30 143.38 152.31 161.78 167.82 173.16 Ranca 8.16 41.90 84.04 123.66 190.10 253.62 330.64 Bungur Bend. Pasar 41.40 86.01 115.04 133.29 156.36 173.47 190.46 Baru Cengkareng 37.82 87.86 117.31 136.01 158.90 175.44 191.56 Drain Gunung 55.44 119.45 142.53 154.62 164.51 175.78 183.47 Mas Gadog 58.30 115.81 143.22 158.87 176.51 188.43 199.41 Cibinong 62.79 107.80 133.99 150.87 171.87 187.29 202.66 FTUI 81.41 115.88 135.80 148.60 164.44 176.10 187.68 Cawang 19.46 97.52 146.39 175.03 206.88 227.71 246.28 * in millimeter.
Fig. 3 (a) Digital Elevation Model (DEM) and (b) catchment area of domain
Sub-basin loss model was calculated using the SCS Curve Number method, while direct runoff calculation used synthetic unit hydrograph SCS Curve Number method [6]. It is therefore necessary to determine the hydrologic parameters in the form of Curve Number (CN) and the percentage of impervious area (PIA). The data used to generate the CN and PIA is a land-use map produced from LANDSAT TM-8 satellite imagery year 2003 analysis and Soil Type Map. CN obtained by classifying a land-use map based on SCS TR-55 [7] and the grouping of soil types based on hydrology groups [8], which was developed by the U.S. Soil Conservation Service using the infiltration characteristics of soils. It can be seen in Fig. 4.
HEC-HMS was run for two days simulation with an hour interval. It combined basin model, meteorological model, and control specifications. The output of simulation is maximum discharge that will be used as an input ANUGA model (Table III). TABLE III MAXIMUM DISCHARGE OF THE RIVER FOR RETURN PERIOD 1-100* Return Period River 1 5 10 25 50 100
Fig. 4 (a) Land-use map, (b) soil hydrology group, (c) curve number, and (d) percentage of impervious of domain area
Meteorological model in HEC-HMS is the major component that defines meteorological boundary conditions for sub-basins [5], such as precipitation and evapotranspiration. In this study, Thiessen polygon method is used to calculate areal precipitation that determines the influence of rainfall to the sub-basin, as seen in Fig. 5.
Angke Atas
72.55
157.88
193.52
249.39
301.45
Buaran
10.58
23.56
29.42
41.33
52.27
359.04 65.24
Cipinang
33.98
76.04
90.00
108.92
123.98
140.33
Grogol
18.075
38.66
45.83
56.25
65.10
75.04
Jati Kramat
9.80
22.07
27.59
38.57
49.07
61.61
Kalibaru Kalibaru timur Krukut
8.43
16.57
18.58
21.00
22.75
24.47
24.08
47.33
53.08
60.00
65.00
69.92
11.87
25.94
29.67
34.50
38.25
42.17
Mookevart
3.90
14.00
16.70
20.10
22.70
25.40
Sunter Kanal Banjir Barat Ciliwung Kota Pesanggrahan *in m3/s
26.46
58.58
69.48
85.46
99.63
115.62
28.30
62.65
90.21
102.45
111.29
122.32
28.30
62.65
90.21
102.45
111.29
122.32
114.00
248.10
304.10
391.90
473.70
564.20
C. 1D Flood Modeling using HEC-RAS Hydraulic model used in this study is HEC-RAS. This model has been widely used for practical purposes, such as estimating flood discharge and also producing flood hazard maps using delineation method of water excess to topography in each river cross-section. HEC-RAS model simulation output really depends on the accuracy of the geometric data. Geometric data as the input for HEC-RAS is DEM data combined with river cross section data corrected through field surveys. A sample of river cross section data can be seen in Fig. Field survey points can be seen in Fig. 6.
Fig. 6 (a) Sample of river’s cross section data in Kali Krukut, (b) the river’s cross section survey point in red dot, and (c) final river’s cross section data in domain area with green line Fig. 5 Areal precipitation based on Thiessen polygon method
HEC-RAS calculates discharge based on the coefficient of friction or Manning's values. Manning’s value is produced based on land-use data that derived from LANDSAT TM-8 image analysis, as seen in Table IV. D.2D Flood Modeling using ANUGA ANUGA use finite volume method to solve shallow water wave equation. The domain area is defined with mesh model that built elements of the triangle, where each centroid of the triangle elements has a value of water depth , horizontal momentum vector in -direction and in -direction . Other physical parameters that are included in calculation is the topography of the land and the absolute depth of the water (stage level), symbolized by . The output resolution is determined based on the size of the triangular elements. The density and size of the elements in a region determine the resolution of simulation results [9]. Higher density of the elements and smaller the distance between the node of elements provide higher resolution of simulation results. In lakes and river area, the distance between its nodes of elements is constrained ≈ 4 m; 5-6 m in coastline and the riverbank or flood plain area; and no constrain distance of elements in outside of these area, as seen in Fig. 7.
III. RESULTS This study used report of Jakarta’s flood 2007 published by Jakarta Disaster Management Agency (BPBD Jakarta) to verify the output of ANUGA and HEC-RAS. BPBD reported the flood area in village (kelurahan) spatial resolution. The verification process performed quantitatively through overlaying the output of 2007 Jakarta flood event reconstruction with BPBD’s report, as seen in Fig. 9. In all domain area, match test shows that the ANUGA’s output has 99% accuracy, while the accuracy of HEC-RAS’s output is 98%.
Fig. 9 2007 Jakarta flood area from BPBD’s report (red polygon) that overlay with output of 2007 flood reconstruction from (a) ANUGA and (b) HEC-RAS
Fig. 7 (a) Sample of mesh model and density of elements in (b) lake area and in (c) coast area
Domain area of ANUGA simulation can be seen in Fig. 8, the total area is 461.58 km2.
Fig. 8 Domain of ANUGA simulation
In more detail analysis, ANUGA produced a better resolution of flood inundation area because there are discontinue flood inundation areas in the output of HEC-RAS, as can be seen in Fig. 10. Therefore, this study used ANUGA to produce Probabilistic Flood Hazard Maps (PFHM).
Fig. 10 Comparison of output ANUGA and HEC-RAS in Cipinang River, the discontinue flood area in green circle
Probabilistic Flood Hazard Maps of Jakarta as output of ANUGA can be seen in Fig. 11.
type. The loss is only material damage, excluding content damage and business interruption loss. Jakarta flood losses estimation from ANUGA can be seen in Table IV. TABLE IV JAKARTA FLOOD LOSS ESTIMATION FROM ANUGA
Model ANUGA
Return Period
Loss Estimation (million)
1 year 5 years
Rp 1,303,043.72 Rp 16,179,007.78
10 years
Rp 16,852,777.69
25 years
Rp 21,208,105.66
50 years
Rp 24,320,382.51
100 years
Rp 24,672,844.75
IV. DISCUSSION AND FUTURE DEVELOPMENT The output of fit test using Chi-Square and SmirnovKolmogorov method for rainfall distribution and daily maximum rainfall in rainfall frequency analysis shows an insufficient result. Hence, for a better rainfall distribution in Jakarta area, it is necessary to do an advance rainfall frequency analysis. Hydrologic analysis can be improved with better temporal resolution of rainfall data, since this study only use daily maximum rainfall. In hydrologic analysis, estimating areal precipitation using another method is necessary. Thiessen polygon has a limitation because it includes the rainfall data from observation stations located outside of catchment area border to calculate areal precipitation. The topography of 13 watershed areas in Jakarta is not uniform, therefore it needs another method to calculate areal precipitation that accommodates topography effect. Topographic and land-use data need to be updates due to rapid changing of these variables in Jakarta. In loss calculation, the vulnerability value need to be redefined based on other flood parameter such as flood water velocity and period of inundation. The vulnerability value also needs to be defined based on building type. ACKNOWLEDGMENT The authors would like to thank all related government agencies; Indonesian Institute of Science (LIPI), Jakarta Disaster Management Agency (BPBD Jakarta), Balai Besar Wilayah Sungai Ciliwung-Cisadane and Public Works Ministry, Indonesian Institute of Aeronautics and Space (LAPAN), Indonesian General Insurance Association (AAUI), and Australian National University (ANU).
Fig. 11 Probabilistic Flood Hazard Maps of Jakarta for (1) 1 year, (b) 5 years, (c) 10 years, (d) 25 years, (e) 50 years, and (e) 100 years
The losses are quantified based on the combination of housing data of year 2010 and the housing value data of year 2005. The total value of the buildings in Jakarta is Rp 434.43 t. The building’s vulnerabilities are constructed from loss ratio of flood insurance claim data from 2002, 2007 and 2013 flood events. This vulnerability function is only based on flood depth and use assumption that all the buildings are the same
REFERENCES [1] [2]
Badan Pusat Statistik Provinsi DKI Jakarta. (2015). Statistik Daerah Provinsi DKI Jakarta 2015. Kementerian Negara Perencanaan Pembangunan Nasional/ BAPPENAS. (2007). Laporan Perkiraan Kerusakan dan Kerugian Pasca Bencana Banjir Awal Februari 2007 di Wilayah JABODETABEK (Jakarta, Bogor, Depok, Tangerang, dan Bekasi).
[3]
[4] [5]
[6]
[7]
[8]
[9]
Environment Agency. (2013). Benchmarking the Latest Generation of 2D Hydraulic Modelling Packages. U.K. Department for Environment, Food & Rural Affairs. U.S. Army Corps of Engineers, Hydrologic Engineering Center. (2013). Hydrologic Modeling System, HEC-HMS User's Manual, Version 4.1. Cunderlik, J., & Simonovic, S. P. (2004). Calibration, Verification and Sensitivity Analysis of the HEC-HMS Hydrologic Model. Water Resources Research Report, 11. United States Department of Agriculture, Natural Resources Conservation Service. (2004). Hydrologic Soil-Cover Complexes. In National Engineering Handbook Hydrology. United States Department of Agriculture, Natural Resources Conservation Service. (1986). Urban Hydrology for Small Watersheds Technical Release 55. United States Department of Agriculture, Natural Resources Conservation Service. (2009). Hydrologic Soil Groups. In National Engineering Handbook Hydrology. Kim, B., Sanders, B. F., Schubert, J. E., & Famiglietti, J. S. (2014). Mesh Type Tradeoffs in 2D Hydrodynamic Modeling of Flooding with a Godunov-based Flow Solver. Advances in Water Resources, 68, 42-61.
