Recent Initatives for Enhancing Landslide Risk
Management in Hong Kong
Florence W.Y. KO
Geotechnical Engineering Office, Civil Engineering and Development Department, Government of HKSAR
Abstract. Hong Kong faces a unique long-term slope safety problem due to its dense urban development in a hilly terrain combined with high seasonal rainfall. Its slope engineering practice and landslide risk management have evolved in response to experience and through continuous improvement initiatives and technology advances. The application of state-of-the-art slope engineering practice and quantified landslide risk management has reduced landslide risk to an as low as reasonably practicable level that meets the needs of the public and facilitates safe and sustainable developments. This paper will give an overview of the slope safety system that serves to manage landslide risk in a holistic manner through an explicit risk-based approach and strategy and provide an update on the recent initiatives undertaken as part of the continuous efforts to enhance landslide risk management. These initiatives include identifying new candidates of vulnerable hillside catchments, developing a territory-wide rainfall-based landslide susceptibility model and assessing potential implications to the risk profile of natural terrain due to extreme rainfall events.
Keywords. slope safety, landslide, risk management
1. Introduction
In a dense urban setting like Hong Kong, even a relative small-scale landslide is liable to result in serious consequences.
Hong Kong has a population of 7 million and a land area of 1,100 km2, about 60% of which is hilly natural hillsides (with 75% of the land steeper than 15o and 30% steeper than 30o). Rapid population growth and economic expansion from the 1950s to 1970s have led to intensive urbanization of the foothill areas, giving rise to a large number of substandard man-made slopes formed without geotechnical control and exposing people to landslide risk from natural hillsides. The substandard slopes and steep natural terrain, combined with deep weathering profile and high seasonal rainfall, are highly susceptible to rain-induced landslides.
Following its establishment in 1977 as a central body to regulate slope safety and geotechnical engineering in Hong Kong, the Geotechnical Control Office (renamed Geotechnical Engineering Office (GEO) in 1991) has progressively developed an integrated slope safety system that serves to manage landslide risk in a holistic manner through an explicit
risk-based approach and strategy. This paper will give an overview of the slope safety system and provide an update on the recent initiatives undertaken as part of the continuous efforts to enhance landslide risk management, in particular for landslide risk arising from natural terrain. These initiatives include identifying new candidates of vulnerable hillside catchments, developing a territory-wide rainfall-based landslide susceptibility model and assessing potential implications to the risk profile of natural terrain due to extreme rainfall events that are projected as becoming more frequent in terms of occurrence due to climate change effects.
2. Slope Safety System
The slope safety system in Hong Kong comprises a range of initiatives that serve to manage landslide risk in a holistic manner through an explicit risk-based approach and strategy. The goals are: (a) to reduce landslide risk to the community through a policy of priority and partnership, and (b) to address public perception and tolerability of landslide risk so as to avoid unrealistic expectations. Apart from saving lives
© 2015 The authors and IOS Press.
This article is published online with Open Access by IOS Press and distributed under the terms of the Creative Commons Attribution Non-Commercial License.
through averting potential fatalities, the system also adds value to the society by improving the built environment through landscaping of slopes as well as landslide preventive and mitigation measures.
The slope safety system incorporates the application of fundamental risk management concepts at the policy administration as well as at the technical level. There are four key strategic components and a number of major initiatives and tasks under the system. Details of the slope safety system, together with its evolution in response to landslide disasters, are described by Chan and Lau (2008) and Malone (1998).
3. Update of Recent Initiatives
The challenges of managing natural terrain landslide risk in Hong Kong are discussed in Wong (2009) and Ho (2014). Recent initiatives undertaken as part of the continuous efforts to enhance risk management of natural terrain landslides are discussed in the following sections. 3.1. Identifying New Candidates of Vulnerable Hillside Catchments
The Landslip Prevention and Mitigation Programme (LPMitP) was launched as a rolling programme in 2010 with a target annual output (Wong, 2009). About 50% of the LPMitP resources are deployed to deal with natural terrain landslide hazards, which was commensurate with the projected risk distribution in 2010. Following the ‘react-to-known-hazard’ principle, vulnerable hillside catchments were selected based on their risk-based ranking order for action under LPMitP. The LPMitP marks a new chapter in Hong Kong’s landslide risk management, by incorporating systematic study and mitigation of natural terrain landslide risk as an integral part of Hong Kong’s long-term slope safety endeavour.
In 2013, the GEO identified the following new candidates of vulnerable hillsides that may warrant inclusion under LPMitP for priority action.
(a) Historical Landslide Catchments (HLC) - The original inventory of vulnerable hillside catchments (i.e. HLC) was compiled in 2007, based on the proximity of the historical landslides
in the Enhanced Natural Terrain Landslide Inventory (ENTLI) to existing developments. The ENTLI is an enhancement of the Natural Terrain Landslide Inventory that was compiled in the mid-1990s (King, 1996). A natural hillside catchment is defined as an HLC if the selection criteria in Figure 1 are met. The landslides considered under these criteria include both relict and recent channelized debris flows (CDF) and open hillslope landslides (OHL).
Figure 1. Existing HLC selection criteria.
In June 2008, an exceptionally intense rainstorm (Figure 2) hit Lantau Island of Hong Kong resulting in about 2,400 landslides on the natural hillsides (Figure 3). The June 2008 rainstorm gave new insight on the mobility of natural terrain landslides. In particular, CDF triggered by a severe rainstorm could be much more mobile than previously observed. A review of the runout distribution of recent CDF prior to the June 2008 rainstorm shows that about 85% of the recent CDF have runout distance ≤ 100 m. At this percentage of 85%, the CDF that occurred in 2008 alone have runout distance up to about 200 m (Figure 4).
Figure 2. 4-hour rainfall in the June 2008 rainstorm. 200
100 300
Figure 3. Widespread natural terrain landslides and debris flows on Lantau Island in June 2008.
Figure 4. Distribution of runout distance of recent CDF in the ENTL (up to 2009).
In light of this, it is considered appropriate to extend the buffering distance between the crown of a recent CDF and the upslope boundary of an important facility to within 200 m. The extended HLC selection criteria are shown in Figure 5.
Figure 5. Extended HLC selection criteria. (b) Hillside Pockets - A hillside pocket is defined as an area of hillside that is located within the predominantly developed area and
satisfies all of the following three criteria: (i) maximum slope angle > 20o ; (ii) elevation difference > 8 m; and (iii) plan area > 400 m2. An example of hillside pockets is shown in Figure 6. During the June 2008 rainstorm, landslides occurred on some hillside pockets close to existing developments. These indicate that the landslide risk posed by certain hillside pockets is not low.
Figure 6. An example of hillside pockets (shown in green patches in between existing developments).
Through a study, about 1,700 hillside pockets affecting important facilities such as buildings and major transportation corridors were identified. Of these 1,700 hillside pockets, about 300 hillside pockets are considered to have known hazards. These include those with relict or recent landslides, and those with known disturbance as anthrogenic disturbance is one of the key contributing factors for landslides within hillside pockets. A risk-based ranking system has been developed to prioritize the 300 hillside pockets for action under LPMitP.
(c) Sizeable Catchments with Major Drainage Lines (MDC) - MDC were delineated for all natural hillsides according to the criteria shown in Figure 7. About 380 MDC affecting multi-storey buildings or clusters of low-rise buildings were identified. Many of these MDC do not satisfy the HLC selection criteria (Figure 5), i.e. there were no landslides in close proximity. However, these MDC can pose a potential threat in respect of a low-frequency,
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 1 10 100 1,000 10,000 Cu m ul at iv e Pe rc en ta ge (% ) Runout Distance (m) CDF (Up to 2007) CDF (2008 only) CDF (Up to 2009) 200 85%
large-magnitude landslide, especially under extreme weather conditions, due to the presence of large catchment areas and long drainage lines. Further work is in progress to assess the characteristics of the MDC in terms of landslide risk. Where considered appropriate, deserving MDC would be prioritized for action.
Figure 7. Delineation criteria of a MDC. 3.2. Developing a Territory-wide Rainfall-based Landslide Susceptibility Model
Lately in 2014, a new rainfall-based landslide susceptibility model was developed for assessing the susceptibility to landslides of natural terrain in Hong Kong. The purpose of the model is to make approximate territory wide predictions of the scale of landslide impacts for anticipated precipitation events, in order to provide information necessary for emergency preparedness and planning purposes. The model is one of the few substantial attempts to introduce rainfall intensity as a predictor in a statistical manner. Rainfall is rarely considered in landslide susceptibility analyses carried out elsewhere, as usually adequate rainfall data is neither available nor reliable, which renders relating rainfall to landslide occurrence difficult, if not impossible. However, landslides are very sensitive to rainfall and if rainfall is not considered in a susceptibility analysis, any direct application of the results of the susceptibility analysis would not be very accurate and may mislead important risk-based decision making.
The susceptibility analysis considered landslide densities on natural hillsides for the years between 1985 and 2006, plus 2008 (i.e. a total of 23 years) within which year-based
contours of normalized maximum rolling 24-hour rainfall were available. Year-based normalized maximum rolling 24-hour rainfall at a location is equal to the maximum rolling 24-hour rainfall in a year divided by the mean annual rainfall (1977 to 2006) at the same location (Chan et al. 2012). In the analysis, landslide density was correlated with the normalized maximum rolling 24-hour rainfall, with effects of slope angle and solid geology taken into account. Six classes of rainfall intensity (I: 0.025-0.10, II: 0.10-0.15, III: 0.15-0.20, IV: 0.20-0.25, V: 0.25-0.30, and VI: 0.30-0.35), together with eight slope angle classes (<15°, 15°-20°, 20°-25°, 25°-30°, 30°-35°, 35°-40°, 40°-45° and >45°) and three solid geology classes (intrusive, volcanic and sedimentary) were considered. The analysis was undertaken on a GIS platform.
Figure 8 shows the year-based rainfall-landslide correlation for combined intrusive and volcanic areas. Sedimentary-origin landslides were not considered in deriving the correlation given the minor proportion and localized nature of sedimentary area, and the limited amount of data under high rainfall classes. Only 10% of the natural terrain area (which is about 0.1×660=66km2) is of sedimentary origin. Others belong to the intrusive (30%) and volcanic (60%) groups. The majority of sedimentary area clusters at north-eastern Hong Kong. Nevertheless, the limited data indicates that, under the same rainfall and slope angle classes, sedimentary area is in general as active as the volcanic area but is more susceptible to landslides than the intrusive area, and to err on the safe side, the year-based rainfall-landslide correlation for sedimentary area is assumed to be the same as the one for volcanic area.
Figure 8. Year-based rainfall-landslide correlation for combined intrusive and volcanic areas.
0.001 0.01 0.1 1 10 100 1000 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 Land sl ide D ens ity (no. / km 2)
Normalized Maximum Rolling 24-hour Rainfall
Slope Angle Class (o)
N 30 ~ 35 35 ~ 40 40 ~ 45 45 ~ 90 0 ~ 15 15 ~ 20 20 ~ 25 25 ~ 30
The year-based rainfall-landslide correlation considers rolling maximum 24-hour rainfall and the total number of natural terrain landslides observed in a year. There are many rainstorms in a year and each rainstorm would result in some landslides. For direct application in predicting number of natural terrain landslides that would occur in a rainstorm, the year-based rainfall-landslide correlation needs to be transformed into a storm-based correlation. This was done by applying global adjustment factors to the year-based landslide densities for each rainfall class. The global adjustment factor for each rainfall class was derived based on the hit area ratio between the storm-based rainfall and the overall natural terrain area. Figure 9 presents the storm-based correlation as derived in this manner. Based on the rainfall and landslide data, the correlation for intrusive and volcanic areas can be obtained by applying an adjustment factor of 0.5 and 1.5 respectively to the storm-based correlation in Figure 9.
Figure 9. Storm-based rainfall-landslide correlation. 3.3. Assessing Potential Implications to the Risk Profile due to Extreme Rainfall Events
As part of a holistic approach to managing landslide risk, the GEO is vigilant about any potential residual, or new, landslide risks that are becoming imminent. Particularly, the effect of more frequent and severe rainfall due to climate change on the risk profile of man-made slopes and natural terrain in Hong Kong is being studied.
Under an extreme rainfall event, large number of landslides could occur and many may
develop into large-volume failures with great mobility. Buildings, infrastructures and municipal systems may be damaged and severed. Consequences of large-scale landslides may also be escalated as they develop into other possible devastating secondary events. The GEO is developing and reviewing natural terrain landslide scenarios associated with different extreme rainfall events, which include assessments of different extreme event scenarios in terms of potential landslide consequence and risk. In these analyses, the hazard model may be built on a relationship between rainfall intensity and natural terrain landslide density for different terrain gradients as derived from the June 2008 rainstorm (Figure 10). The consequence model may follow the one described in Wong et al. (2006). As a result, it can be predicted that a 24-hour rolling rainfall event of a maximum intensity up to 60% of the mean annual rainfall of Hong Kong (i.e. about 0.6 × 2,000 mm) that hits the populated urban area would probably bring about 50% increase in the overall risk of natural terrain landslides (in terms of annual potential loss of life). This rainfall event corresponds to a return period of an order of tens of thousands years and is equivalent to 70% of the updated 24-hour Probable Maximum Precipitation estimate that has recently be established (AECOM and Lin, 2014). It is expected that for a given storm of this extreme intensity, the likely number of fatalities would be over tens of thousands. The 50% risk increase is in good agreement with Wong et al. (2006), which estimated that a 24-hour rolling rainfall event of an intensity equivalent to about 40% of the mean annual rainfall may result in 30% increase in the overall risk of natural terrain landslides.
The study of the effect of extreme rainfall on the risk profile of man-made slopes may also be undertaken perhaps at a later stage noting that quite a large number of sizeable high consequence man-made slopes have been upgraded to the required standard of safety using sufficiently robust design measures. Even if any of them fails, the failure volume would be well within a manageable range and disturbance to the public would be largely confined.
The current slope safety system has been accustomed to dealing with the more frequent rainfall scenarios that normally have a 24-hour
0.0001 0.001 0.01 0.1 1 10 100 1000 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 Lands lide D ens it y ( no. / k m 2)
Normalized Maximum Rolling 24-hour Rainfall
Slope Angle Class (o) 0 ~ 15 15 ~ 20 20 ~ 25 25 ~ 30 30 ~ 35 35 ~ 40 40 ~ 45 45 ~ 90
rolling rainfall ranging between 10% and 30% of the mean annual rainfall. Up to now, these rainfall scenarios contribute to about 80% of the overall risk of natural terrain landslides (Wong et al. 2006). The 50% increase in the overall risk due to extreme rainfall events would modify the risk profile in such a way that the risks of normal and extreme rainfall scenarios would broadly be in the ratio of 2:1. As landslide risks from the typical hillside catchments are gradually mitigated and extreme rainfall events become more frequent, the landslide risk of extreme rainfall scenarios would become comparable to that caused by normal rainfall scenarios. It is indisputable that extreme rainfall is an increasing concern and expanded efforts should be put in place to gear up Hong Kong’s slope safety management and preparedness if the consequences, which would likely be of unprecedented scale and coverage, are to be mitigated.
Figure 10. Landslide hazard model.
4. Conclusions
There are considerable uncertainties in the assessment of hillside behavior when subjected to intense rainfall and forecast of extreme rainfall. They remain some of utmost key challenges to managing natural terrain landslide risk. Continued effort would be put in understanding natural terrain landslides such that these uncertainties would be minimized, for example research work is being undertaken to improve design of risk mitigation measures (e.g. Choi et al. 2014). More detailed assessments of landslide risk arising from extreme rainfall events are also
in the pipeline. It is clear that the conventional way of dealing with landslides through engineering measures may not be sufficient, nor cost-effective, for those landslides arising from extreme rainfall events, both in terms of failure volume and consequence. New strategies to enhance Hong Kong’s resilience against the possible extreme rainfall are needed. This will involve enhanced emergency preparedness, response and recovery, covering key aspects such as alert and warning systems, public education and engagement, emergency services of rescue and evacuation, works repairs and recovery.
References
AECOM Asia Co. Ltd., Lin, B.Z. (2014). 24-Hour PMP Updating Study, Geotechnical Engineering Office, Hong Kong. 97 (vol 1) + 200 (vol 2). (unpublished).
Chan, C.H.W., Ting, S.M., Wong, A.C.W. (2012). Development of Natural Terrain Landslip Alert Criteria, Special Project Report No. SPR 1/2012, Geotechnical Engineering Office, Hong Kong, 68 p.
Chan, R.K.S., Lau, T.M.F. (2008). Slope safety system and landslide risk management in Hong Kong, Proceedings of the First World Landslide Forum, Tokyo, 137-140. Choi, C.E., Ng, C.W.W., Song, D., Kwan, J.S.H., Shiu,
H.Y.K., Ho, K.K.S., Koo, R.C.H. (2014). Flume investigation of landslide debris–resisting baffles, Canadian Geotechnical Journal, 2014, 51(5): 540-553, 10.
Ho, K.K.S. (2014). Managing the uncertainties of natural terrain landslides and extreme rainfall in Hong Kong, Proceedings of the Second World Landslide Forum - Landslide Science and Practice (Volume 6 - Risk Assessment, Management and Mitigation), Rome, Italy, 285-302.
King, J.P. (1999). Natural Terrain Landslide Study - The Natural Terrain Landslide Inventory, GEO Report No. 74, Geotechnical Engineering Office, Hong Kong, 127. Malone, A.W. (1998). Risk management and slope safety in
Hong Kong, Proceedings of the Hong Kong Institution of Engineers Geotechnical Division Annual Seminar on Slope Engineering in Hong Kong, 1, 3-17.
Wong, H.N. (2009). Rising to the Challenges of Natural Terrain Landslides, Proceedings of the HKIE Geotechnical Division Annual Seminar on Natural Hillsides: Study and Risk Management Measures, Hong Kong Institution of Engineers, 15-53.
Wong, H.N., Ko, F.W.Y., Hui, T.H.H. (2006). Assessment of Landslide Risk of Natural Hillsides in Hong Kong, GEO Report No. 191, Geotechnical Engineering Office, Hong Kong, 117. 0.1 1 10 100 1000 10000 100000 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 0.55 0.6 0.65 La nds lid e D ens ity (no ./ km 2)
Normalized Maximum Rolling 24-hour Rainfall Legend
Landslide Density at
Terrain Gradient > 25°
15° < Terrain Gradient ≤ 25°
