Employing the Flash Flood Potential Index (FFPI) with Physical Environmental Factors in Baling, Kedah through GIS Analysis

Zulhisham, N.A.¹ and Md Sadek, E. S. S.*²

¹Department of Survey and Mapping Malaysia, Selangor, Malaysia

E-mail: atiqahzulhisham@gmail.com

²College of Built Environment, Universiti Teknologi MARA, Shah Alam Selangor, Malaysia

E-mail: eran@uitm.edu.my

*Corresponding Author

Abstract

On July 4, 2022, Sungai Kupang, Baling, Kedah experienced a devastating flood that caused 3 fatalities, destroyed or damaged 17 houses, affected 3,546 residents, and resulted in losses estimated at RM25.91 million. The flood was triggered by heavy rainfall in the highland area, which caused multiple landslides to occur simultaneously. The landslides led to a debris flow phenomenon in four main river branches, ultimately resulting in a tragic debris and mud flood in the lowlands and downstream villages. The aim of this study is to analyze the location of flash flood occurrences in Baling and to estimate the likelihood of flash floods based on the identified land physical factors. This study also identifies the critical area of Baling basin that have high potential for a flash flood and evaluates the effectiveness and applicability of the FFPI model compared with historical flood events and remote sensing imagery which have occurred in the few watersheds area. The FFPI model, which was created for the first time in 2003, is used in this study to analyze the flash flood that occurred in Baling by considering slope, land cover, soil data, and vegetation. The FFPI technique is applied in five scenarios to determine the flash flood potential, and the value used is also based on the references. A value of 1 on the index denotes a minimal probability of flash floods, while a value of 10 indicates the highest probability. Based on the findings, the study area had a high possibility of having a flash flood at an index value of 7. The danger level of a severe flash flood is present throughout the research region in all scenarios when the value is more than 50%. The outcome is then utilized to do comparisons using historical information on flash floods and their hotspots area, as well as utilizing satellite imagery to determine the true scale of the flood. This is also important to reduce the impact of floods occurrence in the same place as well managing risk and to plan for disaster-mitigation operations.

Keywords: Baling, FFPI, Flood, Flash Flood Potential Index, GIS, Kedah

1. Introduction

Climate change refers to the long-term alteration of a location's temperature, resulting in unpredictable changes in weather patterns. This phenomenon is responsible for the increasing frequency and intensity of typhoons, floods, rainstorms, and winter storms [1]. The burning of fossil fuels such as coal, oil, and gas has contributed to climate change by heating, moistening, and energizing the Earth's atmosphere. The impact of climate change extends beyond nature and affects human society, particularly in economic, health, and social aspects. The planet's water cycle is also affected by climate change, with wet areas becoming wetter and dry areas becoming drier. Intense downpours are becoming more common, leading to an increased risk of flooding. Extreme rainfall is also on the rise, further contributing to flooding. This is because warmer air is capable of holding more moisture, leading to more intense storms. According to Doswell III [2], flooding is a weather-related calamity that affects people worldwide. Floods occur when water spills over onto dry land, typically due to heavy rainfall. However, flooding can also be caused by factors other than the weather. As a result, a comprehensive understanding of floods must consider the underlying mechanisms connecting meteorological phenomena to floods. Regardless of the event that triggers a flood, the root cause of flooding is the atmospheric processes that produce precipitation, which may have occurred a long time ago.

2. Materials and Methods

This section outlines the scope of the study, as well as the data collection methods and GIS analytical techniques used. The study employed FFPI methods to assess the probability of flash floods in a particular location based on its land physical characteristics.

2.1 Study Area and Data Collection

Baling District is a major town in the northern state of Kedah with an area of 1,529 km². The river basin is situated in the north-west part in Peninsular Malaysia and the catchment area is around 4210 m³. The study area has atmosphere shifting from the rain-storm season where it covers the aggregate region of 11,252 km ² and the zones incorporate three states which are Kedah, Perlis, and Penang. The place was chosen due to frequent flash floods occur by continuous rainfall coupled with forestry activity. Due to the proximity of the catchment areas to Sg Kupang and Sg Ketil, there is a higher probability of flooding in the area. The data collected are mostly from proprietary data and open-source data. The data are in vector and raster format. Table 1 shows the data collected for the classification of flash flood potential, while Table 2 shows the data collected for the verification of flash flood from https://search.asf.alaska.edu/.

2.2 Data Preparation for FFPI

The FFPI methodology utilized four primary types of data or parameters, including slope, soil, land use/land cover (LULC), and vegetation index of NDVI. These parameters were prepared and classified according to the FFPI requirements as shown in Table 3 below. Slope plays a crucial role in runoff, and Figure 1(a) illustrates the percent slope of the study area, while Figure 1(b) shows the FFPI- reclassified slope. The types of soils in the study area are displayed in Figure 2(a), where Acrisols are low base saturation acidic soils, Histosols are soils with a high amount of organic matter, Andosols were formed from volcanic parent material, and Cambisols are formed from alluvial, colluvial, and aeolian deposits. The majority of Ferralsols have good physical characteristics, and they are less likely to erode than most other heavily worn red tropical soils. Figure 3(a) shows the LULC types for the study area, which are mostly mount forest and forestland. Figure 3(b) displays the LULC reclassification used in FFPI. NDVI uses red and near-infrared wavelengths to enhance vegetation features and canopy structure, and Figure 4(a) shows the vegetation index of the study area. Figure 4(b) displays the reclassified NDVI factor.

Table 1: Data collected and used for FFPI

Table 2: Sentinel-1 data collected for flash flood verification

Figure 1: Slope factor of FFPI: (a) Slope map, (b) Reclassified slope map

Figure 2: Soil factor of FFPI: (a) Soil map, (b) Reclassified soil map

Figure 3: Landcover factor of FFPI: (a) Landcover map, (b) Reclassified landcover map

Figure 4: Vegetation of FFPI: (a) Vegetation map, (b) Reclassified vegetation map.

Figure 5: Methodology flowchart of using FFPI

Table 3: The FFPI values were allocated to each dataset based on their level of vulnerability to flash flooding [3] [4] [5] [6] [7] [8] [9] [10] and [11]

Figure 7: Comparison of outcomes: (a) Historical data hotspot area, (b) Baling Historical flood, (c) Flood extent from Sentinel-1 GRD

On the other hand, Figure 7(c) shows the flood map area results from the Sentinel-1 GRD data SAR image. The flood took place in the green area that is depicted with satellite image overlaid, and when compared to the area that is depicted within the historical flood point in Figure 7(b), it can be determined that the flood took place in the area that has a high risk of flash flooding in accordance with its potential. According to the radar image that was obtained, the region that was possibly extremely vulnerable to flash flooding appears to have been affected by the abovementioned tragedy. This finding is consistent with the findings of the prior comparisons, and it appears that the flood disaster impacted the surrounding Baling’s Mountain area, which was particularly vulnerable to flash flooding.

3.3 Discussion on Method used for Study Area

Based on the results obtained, the flash flood occurred in the area that was identified to have a high potential for flash flooding by the FFPI, as explained earlier. This finding is consistent with the results of previous comparisons. To further improve the FFPI, a more comprehensive and enhanced version called FFPI Weight-Of-Evidence (FFPI_WofE) was developed in 2022 [12]. To calculate the WofE percentage value, input response variables and additional factors are required to enhance the FFPI. The Torrential Phenomena Inventory variable is also necessary in the calculation of the WofE value. By conducting a statistically based integrated study of a total of 15 elements, which best highlights the territorial development of the examined process, it will provide a territorial distinctiveness from the perspective of the FFPI. However, since the Torrential Inventory variable was not available, the percentage value could not be calculated.

During the validation process, the identification of locations affected by previous flash floods in areas corresponding to high and very high vulnerability classes in terms of FFPI highlights the effectiveness of the proposed model. This serves as a validation of the methodological approach and supports the recommendation that the results obtained should be utilized and applied in practice.

4. Conclusion

The Flash Flood Potential Index (FFPI) is a method that helps scholars and decision-makers identify areas that are at risk of flash floods. This study employs five different scenarios based on global reviews conducted in the past. The Jordanian expert who assessed the study area's characteristics (including slope, land use/cover, soil texture/type, and vegetation cover/density) changed the final scenario [11]. The weight of slope and land use were found to be the most significant variables in the equations used in this study. With the severity and frequency of slope flash floods increasing, it is becoming more important to conduct hazard and vulnerability studies that are closely linked. Past studies of areas that have experienced extreme events were often carried out locally by agencies in charge of the integrated management of catchment areas after the fact and did not consider how such events could have been estimated to reduce the potential for harm. The study in Baling explores diverse research areas through the utilization of five distinct equations, giving the FFPI value of 7 the most for scenario 4. The findings demonstrate that the parameters of slope and land use weight, as applied in scenario 4, were the most significant factors in this particular investigation. Through comprehensive analysis, all risks associated with various scenarios have been successfully identified. The outcomes of this study reveal that medium-scale risks prevail within the study area.

Acknowledgement

The authors would like to thank Encik Afrizal, from Department of Irrigation and Drainage (DID), Kedah, Puan Faizah from Department of Agriculture (DoA) Kedah and to everyone that help the authors in providing the data used in this study. Technical support from the College of Built Environment, UiTM Shah Alam is also greatly appreciated.

References

[1] National Geographic, (2022). Resource Library, Encyclopedia from Entry . Retrieved from Climate Change: https://education.nationalgeographic.org/resource/climate-change.

[2] Doswell, III, C., (2015). Hydrology, Floods and Droughts Flooding. Encyclopedia of Atmospheric Sciences, Vol. 3(2), 201-108. https://doi.org/10.1016/B978-0-12-382225-3.00151-1.

[3] Smith, G., (2003). Flash Flood Potential: Determining the Hydrologic Response of FFMP basins to Hydrologic Response of FFMP Basins To Heavy Rain by Analysing Their Physiographic Characteristics. A white paper available from the NWS Colorado Basin River Forecast Centre. Retrieved from 1-11. .http://www.cbrfc.noaa.gov/papers/ffp_wpap.pdf.

[4] Kruzdlo, R. and Ceru, J., (2010). Flash Flood Potential Index for WFO Mount Holly/Philadelphia . National Oceanic and Atmospheric Administration. Retrieved from http://bgmresearch.eas.cornell.edu/research/ERFFW/posters/kruzdlo_FlashFloodPotentialIndexforMountHollyHSA.pdf

[5] Arachige, R. and Perera, D. J., (2015). Mapping Flash Flood Potential Using GIS and the Flash Flood Potential Index (FFPI) in the Turtle River and Forest Watersheds in North Dakota . Master Dissertations, Department of Geography and Geographic Information Science, University of North Dakota. https://commons.und.edu/theses/1826/".

[6] Brewster, J., (2009). Development of the Flash Flood Potential Index. A Microsoft PowerPoint presentation. Retrieved from: https://www.powershow.com/view4/68aedd-Mjk3Y/Development_of_the_Flash_Flood_Potential_FFPI_powerpoint_ppt_presentation. 2009

[7] Ceru, J., (2012). The Flash Flood Potential Index for Pennsylvania. 1-31. Retrieved from: https://proceedings.esri.com/library/userconf/feduc12/papers/user/joeceru.pdf.

[8] Smith, G. E., (2010). Development of a Flash Flood Potential Index Using Physiographic Datasets within a Geographic Information System. Master Dissertations, Department of Geography, The University of Utah. 1-69. https://collections.lib.utah.edu/details?id=192329.

[9] Minea, G., (2013). Assessment of the Flash Flood Potential of Bâsca River Catchment (Romania) Based on Physiographic Factors. Central European Journal of Geosciences,Vol. 5(3), 344-353. https://doi.org/10.2478/s13533-012-0137-4.

[10] Zogg, J. and Deitsch, K., (2013). The Flash Flood Potential Index at WFO Des Moines, Iowa. Retrieved from: https://www.weather.gov/media/lmrfc/tech/2016_Vulnerability_Component_FFPI.pdf.

[11] Shawaqfah, M., AlMomani, F. and Khatatbeh, A., (2020). Mapping Flash Flood Potential and Risk Level Using GIS Techniques and the Flash Flood Potential Index (FFPI) in Amman Zarqa Basin of Jordan. Jordanian Journal of Engineering and Chemical Industries, Vol. 3(3), 81-90. https://doi.org/10.48103/jjeci3102020.

[12] Kocsis, I., Bilașco, Ș., Irimuș, I. A., Dohotar, V., Rusu, R. and Roșca, S., (2022). Flash Flood Vulnerability Mapping Based on FFPI Using GIS Spatial Analysis Case Study: Valea Rea Catchment Area, Romania. Sensors, Vol. 22(9), 1–21. https://doi.org/10.3390/s22093573.