Inundation Hazard Zone Created by Large Lahar Flow at the Baekdu Volcano Simulated using LAHARZ

Abstract

The Baekdu volcano (2,750 m a.s.l.) is located on the border between Yanggando Province in North Korea and Jilin Province in China. Its eruption in 946 A.D. was among the largest and most violent eruptions in the past 5,000 years, with a volcanic explosivity index (VEI) of 7. In this study, we processed and analyzed lahar-inundation hazard zone data, applying a geographic information system program with menu-driven software (LAHARZ)to a shuttle radar topography mission 30 m digital elevation model. LAHARZ can simulate inundation hazard zones created by large lahar flows that originate on volcano flanks using simple input parameters. The LAHARZ is useful both for mapping hazard zones and estimating the extent of damage due to active volcanic eruption. These results can be used to establish evacuation plans for nearby residents without field survey data. We applied two different simulation methods in LAHARZ to examine six water systems near Baekdu volcano, selecting weighting factors by varying the ratio of height and distance. There was a slight difference between uniform and non-uniform ratio changes in the lahar-inundation hazard zone maps, particularly as slopes changed on the east and west sides of the Baekdu volcano. This result can be used to improve monitoring of volcanic eruption hazard zones and prevent disasters due to large lahar flows.

1. Introduction

The Baekdu volcano has a peak of 2,750 m and is located on the Changbai Mountain Range, which forms the border between South Korea and China (Yun and Choi, 1996). Baekdu volcano’s high altitude leads to primary damage by lava flows following eruption, as well as secondary damage due to flooding and lahars from the caldera lake at the top of the mountain. Lahars flow throughout volcanic eruption, and are fed by snow, ice, or lakes at the top of the volcano; speeds can reach 65 km/h over a distance of 80 km (Newhall et al.,1997b). A lahar constantly changes its speed and volume, depending on the region through which it flows, and can increase if the lahar flows into a river or lake instead of a valley. As the lahar moves away from the volcano, it slows down at lower slopes and its volume decreases; however, on steep slopes, its speed can exceed 200 km/h (U.S.G.S. Volcano Hazards Program-Lahars). Lahars exhibit fluid flow, but retain some solid properties, resulting in enormous loss of life and property on impact(Schilling, 1998). For example, on June 15, 1991, the volcanic eruption of the Pinatubo volcano in the Philippines, which had a volcanic explosivity index (VEI) of 6, caused damage to low-lying villages due to large amounts of volcanic ash and about 3 km3 of lahar. According to a 1996 survey, sediments from the Pinatubo volcano eruption were at temperatures near 500°C in 1991, causing explosions upon contact with rivers or groundwaer (Newhall et al., 1997a). Since 1991, the volcanic potential of the Pinatubo volcano lahar has led to the development of the first lahar monitoring system. Early detection by this system could prevent hundreds of casualties(Newhall et al., 1997b). Past investigations of lahar flood areas have generally been conducted by examining the sediments in the area and estimating the locations and ages of lahar sediments. However, most volcanoes are difficult to access, and there is a lack of lahar data and sediment observations for some volcanoes (Schilling, 1998). In the case of the Baekdu volcano, it is difficult to predict the extent of damage based on past events due to the lack of quantitative data collected during or following eruptions. The Baekdu volcano has experienced one massive eruption, also called the Millennium eruption, and several smaller eruptions have been recorded historically (Decker and Decker, 1991).

Previous research on the Baekdu volcano has included a study of the applicability of the geographic information system software LAHARZ (Jung et al.,2013), a simulation analysis of volcanic flow (Kim etal., 2014), and a simulation of ash eruption (Kim,2011). However, in the previous study, there was no direct calculation of the hazard area caused by Lahar in the eruption of Baekdu volcano. Therefore further study is still required. We therefore aimed to estimate potential hazard zones due to lahar damage using the LAHARZ simulation and VEI. We performed tw siulations: in the first, the lahar volume was predicted;and in the second, we measured the slope of each region and determined the weighted lahar ratio.

2. Study Area

The Baekdu volcano is located at 41°59′34″N,128°04′39″E (Fig. 1) and its highest peak is 2,750 ma.s.l. The outer diameter of the outer ring surrounding the caldera (Cheon-Ji or Heaven Lake) is 4.4 km south–north by 3.7 km east–west; the lake has an area of 9.82 km2, an altitude of 2,189 m, a maximum depth of 374 m, and a volume of about 2 billion tons (Suh etal., 2013).

Fig. 1. Location of Baekdu Volcano, South Korea

The Baekdu volcano was formed by a basaltic magma eruption about 150-100 million years ago, followed by subdivision of the volcanic body 600,000-10,000 years ago. The top of the mountain collapsed with a large eruption of pumice about 4,000-1,000years ago, forming the caldera. The historical record of volcanic eruptions indicates large and small eruptions from the beginning of the 11th century to the beginning of the 20th century (Yun and Cui, 1996). Since the occurrence of amagnitude 7.3 earthquake inWangcheng, China in June 2002, the frequency of earthquakes that signal imminent volcanic eruption has increased around the Baekdu volcano (Wu et al., 2005). On August 23, 2003, amagnitude 2.3 earthquake nearBaekdu volcano caused a crack in the mountain slope, another precursor of volcanic eruption. A landlide occurred on Setember 8, 2004, due to a magnitude 3.7 earthquake near Baekdu volcano; in the same year, many trees were destroyed, apparently due to an eruption of toxic volcanic gas. Temperatures of Cheon-ji lake has increased, the contents of helium and hydrogen in volcanic gas have increased by more than 10 times, and the surrounding terrain has increased by more than 10cm (Yun et al., 2012). In addition, the velocity difference between P and S waves around the Baekdu volcano indicates that there is an extensive magma chamber beneath the Baekdu volcano (Ri et al., 2016).

3. Method

In this study, we used the LAHARZ program to quantify the potential extent of lahar damage. The LAHARZ program was developed by analyzing the flow of 27 volcanoes in nine countries using digital elevation models(DEMs) and empirical Equations(1)and (2). The area of the volcanic advection was determined from the volume (V) of the volcano and the ratio of volcano height (H) to lahar horizontal distance(L), Crosssection Area of lahar flow (A), area of lahar flows (B). Fig. 2 shows the areas calculated using Equations (1) and (2), varying the H/L ratio from 0.05 to 200 (Iverson et al. 1998).

Fig. 2. Factors influencing volcanic flooding zones in the LAHARZ simulation program. (A) Cross section Area of lahar flow, given a volcano height (H) to lahar flow length (L) ratio of 0.05;(B) area of lahar flows, given an H/L ratio of 200 (Iverson et al., 1998).

\(A=0.05 V^{\frac{2}{3}}\)       (1)

\(\mathrm{B}=200 \mathrm{~V}^{\frac{2}{3}}\)       (2)

BecauseLAHARZutilizes a relatively small number of input parameters, it is very effective for estimating approximate damage in cases where field survey data are insufficient or data acquisition is difficult. It is also useful for planning the evacuation of people near the volcanic area because it can quickly and easily estimate volcanic flood areas(Schilling, 1998).In this study, we simulated six water systems for two scenarios: constant volcanic flow rate and a flow rate that varies according to the slope. The LAHARZ algorithm consists of several steps Table 1), with the lahar flow determined by input variables at each step.

Table 1. LAHARZ launch stage, input data, and parameters

LAHARZ comprises a unidirectional algorithm that can calculate only one flow at a time. Therefore, we selected six flow directions for this study: two to the east, one to the north, two to the west, and one to the south. These flows were designated as East1, East2, North1, West1, West2, and South1, respectively (Fig. 3); their positions and coordinates are shown in Fig. 3. These points were selected from the outside of the PROXIMAL-HAZARD ZONEBOUNDARY, which showed irregular flow when the frst volcanic eruption occurred. The lahar discharge volume was determined according to VEI; the volume corresponded to boundary values between VEI 3 and 7 (Table 2).

Fig. 3. Location of lahar starting point.

Table 2. Eruption volume and examples for varying volcanic explosivity index (VEI) values

The first simulation assumed that the lahar flow was equal in all directions, and the volume of the specified lahar was therefore divided equally between the flows. In the second simulation, we assumed that the lahar volume would vary according to the slope of each water system. In order to calculate the slope at each point, the altitude of each point was measured and the slope was calculated using the altitude. Table 3 shows the altitude of each site and the ratio of lahar in the  each simulation.

Table 3. Elevation of lahar starting point and its input lahar ratio

The following equation (3) was used to make the ratio of lahar parameter in 2nd simulation using altitude. Altitude (A), each point (n).

\(\frac{\frac{1}{A n}}{\sum \frac{1}{A n}}\)       (3)

A total of 60 simulation results were obtained and plotted for the two simulations and six lahar volumes calculated from boundary values between VEI 3 and 7; the results were analyzed using Google Earth to determine the areas of simulated lahar dmage.

4. Results

The number of pixels in the simulation results and the resolution of the DEM used in this study are summarized in Table 4. The first LAHARZ simulation assumed tha lahar flow was uniform in all directions at the same rate. The north flow direction comprised a relatively large, flat area; and the south flow direction comprised a deep valley along the Yalu River. Thus, the northward-flowing lahar flowed through the water system and also escaped through a water system to the side. In comparison, the southward-flowing lahar was narrower and longer (Fig. 4). The results for the east and west lahar flows were similar to those for flows to the north and south. The middle region of the study area is flat terrain; however, the simulated lahar flowed along the water system, which is surrounded by mountains. Therefore, the result was a narrow and long flow toward the end of the study area. In addition, an accurate flooding area was calculated using the number of pixels acquired through the results of the LAHARZ program (Table 4). Fig. 5 shows a region that protrudes abruptly from the lahar flow. This area corresponded to missing values in the DEM, due to an error in the LAHARZ program. These errors can be resolved using higher-resolution DEMs. In this study, we used 30 m SRTM DEMs; future studies should increase the DEM resolution (Munoz et al., 2009).

Table 4. Simulation result for Scenario 1 flooding area obtained using the LAHARZ sofware

Fig. 4. Result of Simulation 1. (A) Volcanic explosivity index (VEI) = 3; (B) VEI = 4; (C) VEI = 5; (D) VEI = 6; (E) VEI = 7; (F) merged results from (A–E).

strong>Fig. 5. Example of ragged edges in a LAHARZ simulation.

The assumption that the lahar flow was equal in all directions did not sufficiently represent natural phenomena. Therefore, in the second simulation, we obtained the slopes of each lahar starting point on the volcano, and distributed the lahar volume proportionally according to the slope. The results of this simulation are shown in Fig. 6 and the flooded area is described in Table 5.

Table 5. The result for Simulation 2 flooding area obtained using LAHARZ

Fig. 6. Result of Simulation 2. (A) VEI = 3; (B) VEI = 4; (C) VEI = 5; (D) VEI = 6; (E) VEI = 7; (F) merged results from (A–E).

A comparison of the first simulation, using equal lahar flow in all directions; and the second simulation, in which lahar volume varies according to slope, showed that the lahar-flooded area varied due to terrain differences in the second simulation. In order to more easily compare the differences between the simulation 1 and simulation 2, we plotted graph (Fig. 7). The largest increase in affected area occurred in West2, where the flooded area was about 2.83% larger in the second simulation than in the first(Fig. 8). The greatest decrease occurred in East2 in which the flooded area was 2.44% smaller than in the first simulation (Fig. 9). Given the overall change in the total flooded area, these changes were relatively small. Changing the H/L ratio resulted in a travel distance increase or decrease of about 5%. To obtain further information about the influence of terrain on lahar flow, we extracted and analyzed these data using the Google Earth (Fig. 10).

Fig. 7. Flooded area of lahar according to VEI at each point.

Fig. 8. Decreases in simulated lahar flow when lahar volume was (A) equal across all starting points and (B) varied according to slope.

Fig. 9. Increases in simulated lahar flow when lahar volume was (A) equal across all starting points and (B) varied according to slope.

Fig. 10. Map of LAHARZ simulation2 result in this study analyzed using the Google Earth (A), At VEI 5, lahar flow affecting the Changbai Chinese autonomous Prefecture in China, Hyesan City in North Korea (B).

At VEI values of 3 and 4, there was no direct lahar damage to residential areas. At VEI 5, however, the lahar began to flow into residential areas, affecting the Changbai Chinese autonomous Prefecture in China, Hyesan City in North Korea, and Changxing Countyin Pudong County to the west. At VEI 6, the lahar was predicted to pass through numerous villages around the Yalu River to the south of the Baekdu volcano, including Kim Dae-dong andKim Jeong-suk in North Korea. When VEI reached 7, all of the villages within a 100 km radius of the Baekdu volcano were included within the scope of the lahar.

5. Conclusion

In this study, we estimated the flooded rea and estimated damage area due to lahar flow using six simulations of six water systems around the Baekdu volcano and 30 m SRTM DEM data. We applied the LAHARZ program, which has not been previously applied to calculate lahar damage zones around the Baekdu volcano. We further analyzed our simulation results using the commercial software Google Earth to quickly identify potential lahar damage areas. At each VEI increment, the total lahar volume increased by a factor of 10. However, the flooded area did not increase in proportional to the lahar volume. When VEI was increased from 3 to 4, the flooded area became about 4.64 times larger. However, when VEI was increased from 6 to 7, the flooded area became about 2.87 times larger. This result was due to differences in terrain, such that when lahar volume was large, flow extended over adjacent water systems as well as the water system designated as the primary flow direction in the simulation. Nevertheless, the topography in the East2 and North1 directions resulted in a 1.8-fold variation in flooding area, because the northward flow passes over flat terrain with no bends in the water system, leading to widespread lahar flow that affected a larger area.

Our results showed that the LAHARZ progrm contains errors that cannot be solved within the software. Irregularities in the simulation results may be addressed using high-resolution images; however, no study has been conducted on the effects of resolution on LAHARZ simulation errors. Because the unidirectional algorithm adopted by the LAHARZ program is designed to accommodate only one water system per simulation, it carries the disadvantage that flow extending to tributaries is not considered. To solve this problem, it is necessary to repeat the simulation for each tributary. In this study, we considered only one variable, the mountain slope; however, additional variables should be considered in future studies The results of the current study will be useful for effective preparation for imminent eruptions of the Baekdu volcano.