REVIEW OF INTERNATIONAL GEOGRAPHICAL EDUCATION
ISSN: 2146-0353 ● © RIGEO ● 11(3), SUMMER, 2021 www.rigeo.org Research Article The Development of the Flood Inundation Area Model in the Way Sekampung Sub- Watershed in Lampung
Aprizal1 Sofia W. Alisjahbana2 Doctoral Program in Civil Engineering, Civil Engineering, Bakrie University, Jakarta Tarumanagara University [email protected] [email protected]
Any Nurhasanah3 Civil Engineering, Bandar Lampung University [email protected]
Abstract Flood disasters often occur in most areas in Indonesia. Floods are the most common disasters that cause losses and casualties. Flood disaster mitigation needs to be done to overcome and reduce the impact of losses. Mitigation of flood disasters will be helpful if you can predict in advance the potential for inundation caused by flooding, namely by modeling flood inundation. So far, research on modeling the area of flood inundation in Indonesia can be categorized as relatively rare. It is due to the many obstacles that surround it. The problem faced is the unavailability of complete data caused by limited funds and the complexity of most of the watersheds in Indonesia. The study aims to develop a flood inundation area model based on land use conditions and rain variables. It is done by maximizing minimal data and integrating empirical and 2D hydrodynamic modeling formulated with statistical models. The method used is to collect secondary data in the form of rain and land use data and primary data by conducting field surveys of flood inundations that have occurred. Then, it is followed by hydrological modeling to obtain flood hydrographs by integrating HEC-RAS, HEC-GEO RAS, and Arc GIS to generate variables to be modeled statistically using the Multiple Linear Regression approach. The modeling uses samples in some Way Sekampung sub-watersheds in Lampung Province, each with two rain time series data. The independent and dependent variables are the variables to be modeled where the independent variables are Watershed Area (X1), % Forest Area (X2), % Residental Area (X3), % Agricultural / Rice Field Area (X4), % Mixed Plantation Area (X5), % Other Area (X6), River Slope (X7), River Length (X8), Rainfall (X9), Flood Peak Time (X10), Flood Discharge (X11) and one dependent variable, namely Flood Inundation Area (Y)Four independent variables do not affect statistical modeling, namely X4, X6, X7, and X9. The model obtained is Y = 150,442 -0,242 X1 -0,412X2-0,515X3- 0,241X5+ 2,050 X8 -0,704 X10+ 0,020 X11 4. MAPE test results (Mean Absolute Percentage Error) The model equation shows a value of 4.672% of the data. Thus, the model is in the Class 1 category that is Very Accurate.
Keywords Flood inundation, Land Use, Modeling, HEC RAS, ARC GIS
To cite this article: Aprizal; Alisj ahbana, S,W, ; Nurhasanah , A. (2021) The Development of the Flood Inundation Area Model in the Way Sekampung Sub-Watershed in Lampung. Review of International Geographical Education (RIGEO), 11(3), 1246-1256. Doi: 10.48047/rigeo.11.3.115
Submitted: 06-03-2021 ● Revised: 13-04-2021 ● Accepted: 20-03-2021 Aprizal, ; Ali sjahbana, S,W, ; Nurhasanah, A. (2021) The Development of the Flood Inundation Area …
Background
As development growth increases, land use also increases, leading to changes in land cover in a watershed (Bedient, Huber, & Vieux, 2008). As a result, erosion and landslides are increasingly uncontrollable in the upstream watershed area, and there is a narrowing of the river channel in the middle watershed area. Meanwhile, the existing rivers and canals (drainage systems) can no longer accommodate rainwater runoff in the downstream watershed area, resulting in flooding. Floods that are very detrimental almost every year also occur in the Sekampung watershed. The term modeling is known to understand natural phenomena, including floods. Modeling is the miniaturization of a complex biological phenomenon into a simple one or vice versa. Ravindran, Phillips, and Solberg (1976) in operations research. Phillips states that what is meant by a model is a simple representation of something real. Modeling is an option in understanding the phenomenon related to flooding. According to J. Teng et al., research on flood phenomena through flood modeling can be grouped into two groups, namely: 1) empirical, conceptual approach and 2) hydrodynamic approach consisting of 1D, 2D, and 3D modeling. (Teng et al., 2017)
3 Dimention Spatial Flood Inundation Modeling Method
2 Dimention Measurement
1 Dimention
Empirical Hydrodynamic Method method
Figure 1. Classification Diagram of Flood Inundation Area Modeling
Each approach has advantages and disadvantages. The conceptual, empirical approach is supported by measurements, surveys, and remote sensing—statistical models developed from these data-based methods (Schumann et al., 2009; Smith, 1997). Hydrodynamic model approaches such as one-dimensional (1D) (Brunner, 2016; DHI, 2003), two-dimensional (2D) (e.g., DHI, 2012; Moulinec et al., 2011), and three-dimensional (3D) (e.g., Prakash et al., 2014; Vacondio et al., 2011) simulates the movement of water by solving equations supported by computer technology.(Teng et al., 2017). Currently, floods often occur in developing countries such as Indonesia, which has many rivers. Flood discharge data and flood history are complicated to obtain so that the empirical and hydrodynamic modeling of the inundation area is experiencing considerable obstacles. This study tries a new development to model the flood inundation area, which can overcome the problem by generating flood data from rain data by preparing DEM and SHP maps using HEC-RAS 5.0.
2 dimentional Empirical Hydrodinamic method modeling
Slice 2. Modeling Method using Statistics
Figure 2. Model Development Novelty Diagram
1247 © RIGEO ● Review of International Geographical Education 11(3), SUMMER, 2021
Study Area
Lampung Province is one of the provinces in Indonesia, which is located on the island of Sumatra. Lampung has three river areas, namely Mesuji-Tulang Bawang, Seputih-Sekampung, and Semangka. The study area is focused on the Way Sekampung sub-watershed, as shown in Figure 3 below.
Bandar Lampung – Kalianda Watershed
Figure 3. Way Sekampung Lampung Sub-watershed Study Area
Source: (Syafri, 2013)
Topography Map
DEM map shows the elevation of an area, which helps know the boundaries of a watershed, river elevation, and the earth's shape to help model a watershed.
Figure 4. DEM Study Area Sub Watershed Way Sekampung Lampung
Shapefile (SHP)
Shapefile is a data format for storing vector-based non-topological spatial data. Shapefiles are used to store digital map data in geographic information systems. This data format is developed by ESRI (Aronoff, 1989) This data format is capable of storing spatial data, such as fields (islands, provinces), lines (roads, rivers), points (cities, buildings), and information about the three spatial data (the name of a town, type of a street, etc.).
1248 Aprizal, ; Ali sjahbana, S,W, ; Nurhasanah, A. (2021) The Development of the Flood Inundation Area …
Figure 5. SHP for Way Sub-watershed Study Area, Sekampung Lampung
Problem Identification
Based on the above background, the identification of the problem in this study is: 1. What are the variables of a watershed that affect changes in flood inundation areas? 2. What is the correct model for the flood inundation area in the Way Sekampung sub-watershed?
Methodology
The method used in conducting this research is as follows:
1. Collecting data: Digital Elevation Model (DEM) is obtained from download on Digital Elevation Model National (DEMNAS) Indonesia. Shapefile (SHP) is obtained from download on Shapefile (SHP) Indonesia Info Geospatial. Two time-series data, 1996-2007 and 2004-2015, in the four Way Sekampung sub- watersheds for calculating the Nakayashu hydrograph debit. Inundation Event Map. 2. Processing Digital Elevation Model (DEM) data for the four Way Sekampung sub- watersheds using geometry modeling in the ArcGIS 10.3 application and Global Mapper version 20.0 for use in the HEC-RAS modeling version 5.0.7. 3. Analyzing and modeling the 2D flood distribution in the four Way Sekampung sub- watersheds (without any water structures), based on the river geometry data generated by the Ras Mapper with the Nakayshu hydrograph debit parameters in 2007 and 2015. After the data formation process carried out by ArcGIS version 10.3 and Global Mapper version 20.0 is complete, the next step is to enter the data into the HEC-RAS version 5.0.7 application. It is done to correct the imperfect geometry due to the weak level of DEM accuracy by making a two-dimensional model. After the geometry data is fixed and the new DEM is formed into two dimensions, the researchers enter data for unsteady flow, which means that the water depth can change with time. In the unsteady flow stage, the input data is the debit data calculated by the Nakayashu hydrograph. After all the geometry data and flow data are entered, the authors analyze and model the flood distribution by clicking the Run Unsteady flow menu. The author then clicks compute to complete the calculation of the water level profile. The last step in modeling the distribution of floods is to display the results of the flood inundation simulation. 4. Map and calculate the area of flood inundation that occurs. The author uses the help of the RAS Mapper feature in the HEC-RAS 5.0.7 program to create a flood inundation map, with the following steps:
1249 © RIGEO ● Review of International Geographical Education 11(3), SUMMER, 2021
After the results of the hydraulics computation with the input discharge time are complete, the results can be seen in the RAS Mapper feature by activating the map layer from google satellite or others. The author then computes the inundation boundary, which is the area value of the flood inundation simulation. After that, it can be concluded. The final preparation of flood inundation maps in the four Way Sekampung sub-watersheds can be carried out. 5. Researchers validate the area of flood inundation that has occurred with the simulation results. 6. Researchers do statistical modeling using SPSS Statistics 25 software. 7. Researchers test and validate the resulting model. 8. Conclusions and suggestions
Result and Discussions
The data processing and calculation processes are sequentially presented below.
Changes in Land Use of four Sub-watersheds
Table 1. Area of land use in the four sub-watersheds of Way Sekampung
Way Galih Watershead Way Kandis Watershead Type of land use Model Land Use N Type of land use Model Land Use No 1 2 o 1 2 1 Forest 10,3 7,63 1 Forest 20,16 8,31 2 Residential area 25,72 82,42 2 Residential area 41,81 106,34 Agriculture/rice Agriculture/rice 3 field 197,91 282,17 3 field 85,56 134,55 4 Mixed plantation 168,22 33,19 4 Mixed plantation 160,64 58,83 5 Etc 8,05 4,79 5 Etc 0,77 0,91 Watershead Area 410,2 410,20 Watershead Note: in Km2 Area 308,94 308,94 Note: in Km2
Way Katibung Watershead Way Pisang Watershead Type of land use Model Land Use Model Land Use No No Forest 1 2 Settlement 1 2 1 Forest 7,61 1,95 1 Forest 50,8 10,86 2 Residential area 29,37 31,8 2 Residential area 13,44 16,18 Agriculture/rice Agriculture/rice 3 field 100,91 176,03 3 field 19,85 31,39 4 Mixed plantation 92,01 21,07 4 Mixed plantation 66,05 93,73 5 Etc 4,21 3,26 5 Etc 5,2 3,18 Watershead Watershead Area 234,11 234,11 Area 155,34 155,34 Note: in Km2 Note: in Km2
Nakayasu Synthetic Unit Hydrograph
Synthetic Unit Hydrograph is obtained from the conversion of daily rain with several calculation stages. It helps know the relationship between discharge and time in a watershed, in this case, the four sub-watersheds in the Way Sekampung watershed.
1250 Aprizal, ; Ali sjahbana, S,W, ; Nurhasanah, A. (2021) The Development of the Flood Inundation Area …
(2) Das Way Kandis (1) Das Way Galih
(3) Das Way Ketibung (4) Das Way Pisang
Figure 6. HSS Time Series 1 (1996-2007) in Way Sekampung Lampung sub-watershed
(1) Das Way Galih (2) Das Way Kandis
(3) Das Way Ketibung (4) Das Way Pisang
Figure 7. HSS Time Series 2 (2004-2015) in Way Sekampung Sub-watershed, Lampung
Inundation Area of Four Sub-watersheds
The following is a 2-D inundation area image obtained through calculations and HEC Ras in 4 Way Sekampung sub-watersheds, with a Q50 discharge for two time-series rain data. Figure 6 (1.a and 1.b) is the Way Galih watershed, Figure 6 (2.a and 2.b) is the Way Kandis watershed, Figure 6 (3.a and 3.b) is the Way Katibung watershed, and Figure 6 (4.a and 4.b) are Way Pisang watersheds.
1251 © RIGEO ● Review of International Geographical Education 11(3), SUMMER, 2021
1a 1b
2a 2b
3a 3b
4a 4b
Figure 8. Flood inundation Q50th in the Way Sekampung Sub-watershed, Lampung
Determination of Statistical Variables
It shows the concept of a hydrological and hydraulics modeling plan to obtain related variables in correlation/regression modeling.
River River Slope River Length Watershed Flood Concentrati Inundation Watershed + on Time Area shape + + Flow + + Coefficient + + Flood Discharge +
Rainfall Intensity
Rainfall
Depth of Rainfall Figure 9. Concept Diagram of Relationship between Research Variables
1252 Aprizal, ; Ali sjahbana, S,W, ; Nurhasanah, A. (2021) The Development of the Flood Inundation Area …
Several independent variables:
X1 = Total watershed area (km2) X2 = Ratio of Forest Area to Watershed Area/Sub Watershed X3 = Ratio of Residential Area to Watershed/Sub-watershed Area X4 = Ratio of Agricultural Area to Watershed/Sub-Watershed Area X5 = Ratio of Mixed Plantation Area to Watershed Area/Sub Watershed X6 = Ratio of Other Areas to Area of Watershed/Sub-watershed X7 = River slope X8 = River Length (km) X9 = Rainfall (mm) X10 = Peak Flood Time (hours) X11=Flood discharge (m3/s) The dependent variable: Y1 = Inundation Area (km2)
Correlation Test
It shows the independent variable (X) has a strong correlation to the dependent variable (Y).
Classical Test
Table 2. Classical Model Test List (Priyono, 2017)
No Classic Test Type Result Conclusion 1 Normality test Results follow the trend line Ok 2 Multicollinearity Test Tolerance > 0.100 and VIF Ok value < 10.00 3 Heteroscedasticity Test There is no clear pattern in Ok the scatterplot image, and the dots spread above and below 0 on the Y-axis. 4 Autocorrelation Test (1) < (1,248)< (3) Ok 5 Partial Variable Effect Test From the Partial t-test data Ok (Multiple Linear Regression), the value of Sig. < 0.05 6 (t-test) From the Simultaneous F Ok Test data (Multiple Linear Regression), the value of Sig. < 0.05 7 Simultaneous Variable Effect Test R2 = 0.983 and Adjusted R2 = Ok (Simultaneous F Test) 0.980
Optimal Multiple Regression Equation
Based on the table of determinant coefficient values, Adjusted R Square is 0.980. It explains that the ability of the independent variable in explaining the dependent variable is 98.0%, the remaining 2.0% are other variables. MAPE Test (Mean Absolute Percentage Error)
After obtaining the statistical model equation, the researchers test the equations presented in table 4 and figure 9 below. MAPE (Mean Absolute Percentage Error) is 4,672 %.
1253 © RIGEO ● Review of International Geographical Education 11(3), SUMMER, 2021
Table 3. Variables and Constants Affecting the Model
Coefficientsa Standardize Unstandardize d Collinearity d Coeffi cients Coefficients Statistics Std. Toleranc Model B Error Beta t Sig. e VIF 1 (CONSTANT) 150.442 7.599 19.797 0.000
WATERSHED -0.242 0.006 -1.969 -38.520 0.000 0.167 6.005 ARE % FOREST 2.050 0.050 2.513 40.750 0.000 0.114 8.744 AREA % 0.020 0.001 0.563 15.187 0.000 0.317 3.156 RESIDENTIAL AREA %MIXED -0.704 0.034 -1.272 -20.695 0.000 0.115 8.683 PLANTATION AREA RIVER -0.412 0.051 -0.262 -8.126 0.000 0.420 2.383 LENGTH PEAK FLOOD -0.515 0.075 -0.295 -6.898 0.000 0.238 4.199
DISCHARGE -0.241 0.031 -0.398 -7.876 0.000 0.170 5.865
a. Dependent Variable: Flood inundation area
Y = 150,442 -0,242 X1 -0,412x2-0,515X3-0,241X5+ 2, 050 X8 -0,704 X10+ 0,020 X11
Table 4. MAPE Calculation
Error Error absolute value Time Inundation Forecasting Error absolute divided by the actual index area data model value value
t At Ft At -Ft | At -Ft| | (At -Ft)/At| 1 52.350 49.234 3.116 3.116 0.060 2 58.080 56.663 1.417 1.417 0.024 3 21.900 23.306 -1.406 1.406 0.064 4 37.420 36.190 1.230 1.230 0.033 5 32.159 31.424 0.735 0.735 0.023 6 39.068 34.429 4.639 4.639 0.119 7 45.070 44.138 0.932 0.932 0.021 8 48.040 49.508 -1.468 1.468 0.031 Total 0.374 n 8
MAPE 4.672
1254 Aprizal, ; Ali sjahbana, S,W, ; Nurhasanah, A. (2021) The Development of the Flood Inundation Area …
= 0,374 100 8 = 4,672 %
70.000
60.000
50.000
(km2)
40.000 Area
30.000
20.000 Inundation
10.000 INUNDATION AREA DATA
INUNDATION AREA MODEL 0.000 0 1 2 3 4 5 6 7 8 9 Data Number
Figure 9. Data Margin Bar Graph and Inundation Area Model Based on the MAPE value category, the model is in the Class 1 category, which is Very Accurate.
Conclusions and Suggestions
The conclusion of this research is:
1. The flood inundation model obtained is Y = 150,442 -0,242 X1 -0,412X2-0,515X3-0,241X5+ 2,050 X8 -0,704 X10+ 0,020 X11 Based on the table, the value of Adjusted R Square is 0,980 or 98%. In general, R2 is used as information about the suitability of a model. In regression, R2 is used to measure how well the regression line approaches the original data value created by the model. If R2 is equal to 1, then the number indicates the regression line fits the data perfectly. 2. Of the 11 independent variables, namely Watershed Area (X1), % Forest Area (X2), % Resident Area (X3), % Agricultural / Rice Field Area (X4), % Mixed Plantation Area (X5), % Other Area (X6), Slope River (X7), River Length (X8), Rainfall (X9), Flood Peak Time (X10), Flood Discharge (X11) and one dependent variable namely Flood Inundation Area (Y), four independent variables do not affect anything, namely X4, X6, X7, and X9. 3. The equation constant shows that an increase in Flood Discharge (X11) will increase the Flood Inundation Area. 4. From MAPE (Mean Absolute Percentage Error) test results, the Model Equation shows a value of 4.672% of the data. So, the model is in the Class 1 category, which is Very Accurate. Suggestions related to this research are:
1. Further research needs to be done with more complete and thorough measured data in recording in the watershed. From data on rainfall, temperature, humidity, and river debit, to geomorphological data collection of rivers and watersheds, they can be more accurate in forecasting flood inundation. 2. Modeling with more land-use variations can be considered in the future so that the results of forecasting flood inundation areas are better.
1255 © RIGEO ● Review of International Geographical Education 11(3), SUMMER, 2021
3. This research can be developed with geodatabases and other geographic information systems. It is expected to predict potential flood disasters in real-time to reduce both life and material impact in an area.
Bibliography
Aronoff, S. (1989). Geographic information systems: a management perspective. doi:https://doi.org/10.1080/10106048909354237 Bedient, P. B., Huber, W. C., & Vieux, B. E. (2008). Hydrology and Floodplain 25 Analysis: Pearson Education, Inc., New Jersey. Priyono, I. (2017). Effect of Quality Products, Services and Brand on Customer Satisfaction at McDonald's. Journal of Global Economics, 5(2), 1-4. doi:10.4172/2375-4389.1000247 Ravindran, A., Phillips, D. T., & Solberg, J. J. (1976). Operations research: principles and practice. Syafri, I. (2013). The Meaning Of Coordination In The Arrangement Of Water Resource Management Strategic Plan In Indonesia. 191-197. Teng, J., Jakeman, A. J., Vaze, J., Croke, B. F., Dutta, D., & Kim, S. (2017). Flood inundation modelling: A review of methods, recent advances and uncertainty analysis. Environmental modelling & software, 90, 201-216. doi:https://doi.org/10.1016/j.envsoft.2017.01.006
1256