Abstract
Parts of the Central Cordillera of Luzon are among the most challenging locations on Earth in which to build and maintain transport infrastructure, a situation perhaps most epitomized by the Halsema Highway. Since its original construction in the 1920s, the highway has undergone phased improvement and has become a socio-economic lifeline to the rural communities it serves. Unfortunately, in 1990, continuing road improvement works had to be abandoned owing to the outcome of a 7.8 Ms earthquake and the effects of subsequent typhoon damage. Earthquake reinstatement works were designed in 1998 and constructed between 2001 and 2006. The engineering geological challenge this posed was compounded by the effects of severe typhoon rains during the reconstruction period, requiring continual assessment of changing slope and drainage conditions. Since construction, the road has been affected by several new typhoons including, most notably, Typhoons Pepeng and Ompong in 2009 and 2018 respectively. Field inspections in 2010 and 2018, combined with the interpretation of satellite imagery available in Google Earth, have allowed the performance of the works implemented between 2001 and 2006 to be assessed and the outcome has been largely favourable. Very little of the recorded damage has occurred in the locations of earthquake reinstatement. Instead, several new areas of slope failure and subsequent blockage and damage to the road have developed. Many of these areas can be explained with respect to their underlying engineering geology and geomorphology. However, there are just as many that owe their origin to the pattern of rainfall and runoff arising during the passage of individual and successive typhoons, modified significantly by drainage management practices in the road corridor, where engineering serviceability and land use practices sometimes have conflicting objectives.
‘If the engineer detailed to examine the foundations reports solid rock when his drill has merely struck an insignificant boulder, no amount of excellence in construction methods will ever make the building safe that rests on that foundation … Data painstakingly gathered, intelligently correlated and safely filed, whether it be hydrographic, topographic, geological or what, cannot but go a long way in helping the country to achieve economic independence; and the district engineer who sees this clearly and acts promptly and perseveringly on his vision – he is the engineer whom the generations to come will be proud to honor’
This quotation from West (1918, p. 2), Highway Engineer in the Bureau of Public Works (BPW), Government of the Philippine Islands, was a fitting reminder to his colleagues of the need to ensure thorough investigation and recording of ground conditions prior to embarking on construction works in the challenging terrain of the Philippines. The warning could not have been more apt in the context of road construction in the Central Cordillera of Luzon, and with respect to the prospect of a road between Baguio City and Bontoc in particular. The precursor to this road was the 1–2 m wide Mountain Trail described by Shoemaker & Williams (1918, p. 8) as ‘the most important and difficult trail to maintain’. Soon after, and under his own initiative using labour provided by local tribes, E. Halsema, the Mayor of Baguio City and BPW District Engineer of Benguet Province (Halsema 1991), upgraded the Mountain Trail to a motorable road that became known as the Halsema Highway (Fig. 1) and was opened to traffic in 1930 (Fig. 2).
Halsema Highway in the Central Cordillera of Luzon, Philippines. Geology modified from Geomatrix-Certeza (1996) and BCEOM (1999).
The newly opened Halsema Highway (DPWH file photograph, taken c. 1930 in the Mt Data area). Reproduced with permission of the Department of Public Works and Highways, Republic of the Philippines.
The Halsema Highway is now a 180 km long, 6 m wide mountain road that follows the spine of the Central Cordillera and links Baguio City, at 1510 m above sea level (a.s.l.) in the south, with Mt Data, Bontoc and Banaue further north (Fig. 1). It is a lifeline for the rural communities of the Central Cordillera that originally assisted in its construction, and that now rely on it to transport agricultural produce to the wholesale markets of La Trinidad, a northern suburb of Baguio City. The road has undergone various upgrades and modifications since its original construction and its maintenance has had to contend with the annual onslaught of typhoon rain and the effects of changing land use (most notably agricultural and roadside commercial development) within its corridor. In 2017 the annual average daily traffic (AADT) using the road was in excess of 2000, with approximately one-third comprising commercial vehicles, buses and trucks (Catajan 2019), illustrating how important the road has become.
The Central Cordillera is one of the most active plate margins in the world (Hart et al. 2002; Hearn & Hart 2019) and in July 1990 (Fig. 3) the highway was severely damaged by a 7.8 Ms magnitude earthquake. Earthquake damage was compounded by the effects of two major typhoons that occurred in August and November of the same year. At the time, the road was in the process of being upgraded, the contracts for which had to be abandoned because of the devastating effects of the earthquake. With construction works incomplete and the road unprotected, the rainfalls that followed caused significant additional damage. The challenges encountered during the design of subsequent road reinstatement works in 1997 and 1998 were described by Hart et al. (2002). Construction of these works along the 84 km section of road between La Trinidad and Mt Data, the most highly trafficked section, commenced in 2001 and was substantially completed by 2006 (Fig. 3). It was inevitable, in such a geologically complex and geomorphologically dynamic landscape, that designs prepared in 1998 would require confirmation and updating during the construction period. Accordingly, provision was made for engineering geological assessment to form a key component of works supervision.
Timeline of road improvement and earthquake reinstatement, typhoons and tropical storms, engineering geological investigations and damage inventories referred to in the text. For brevity, only 3 day rainfalls in excess of 500 mm are shown. It should be noted that typhoon rainfalls with 3 day totals considerably less than 500 mm have caused damage to the road but are not represented in this timeline. Typhoons Pedring (Nesat) and Quiel, for example, registered less than 300 mm in Baguio City during a 5 day period in late September–early October 2011 but caused the Halsema Highway to be blocked in over 50 locations (OCD 2013). It should be noted also that the ADB/IBRD-assisted Earthquake Project included provision for the construction of six Bailey bridges across failed saddles.
This proved to be a prudent undertaking because Typhoons Feria (referred to internationally as Utor), Gloria (Chataan) and Igme (Mindulle) caused significant additional damage to the road, its earthworks, drainage systems and retaining structures in July 2001, July 2002 and June 2004 respectively (Fig. 4). Typhoon Feria was probably the most destructive of these typhoons, depositing 1086 mm of rainfall in Baguio City in 24 h (Table 1). During the first 3 years of the construction period alone, in excess of 200 typhoon-triggered landslides took place, necessitating additional reinstatement works that amounted to almost 15% of the original construction contract price.
Slope failure and erosion below the Halsema highway during Typhoon Gloria in 2002, precipitating a fatal accident. Photograph date: July 2002.
Highest 1 day and 3 day rainfalls recorded in Baguio City since 1995 (storm names are local)
Since 2006, the highway has been affected by several typhoons, the most significant of which were Typhoons Julian (Kammuri) in August 2008, Pepeng (Parma) in October 2009 and Ompong (Mangkhut) in September 2018, depositing respectively 970, 685 and 535 mm of 24 h rainfall in Baguio City. Although the 24 h rainfall associated with Typhoon Ompong was only the sixth highest recorded since 1995 (Table 1), 392 mm of this daily total fell in just 6 h (PAGASA 2018). The destructive capability of such intense rainfall on engineering infrastructure is considerable: according to the Department of Public Works and Highways (DPWH, formerly Bureau of Public Works, quoted in SunStar, Anonymous 2018), the equivalent of US$33 million of damage was caused to roads and bridges in the Central Cordillera by the rainfall and floods associated with this typhoon.
This paper describes the geological and geohazard setting of the Halsema Highway and the engineering geological contributions made to the design and construction of earthquake damage reinstatement works between 2001 and 2006. It also presents the findings of field inspections carried out in May 2010 and November 2018, supplemented by a review of satellite imagery available on Google Earth, to assess the performance of the constructed works and identify the ground conditions and geohazards that continue to pose difficulties to the road as a whole. The description provided by Hart et al. (2002) of the earthquake damage reinstatement design prior to its construction included an examination of ground conditions in two specific case study areas, and the performance of the design and the constructed works in these areas is also examined as part of this review. Two additional case studies are presented where significant road and slope damage took place during the construction period itself as a result of typhoon rainfall.
Figure 3 summarizes the sequence and magnitude of the major typhoons and tropical storms that have affected the Halsema Highway since 1990. It also provides a timeline of road improvement and rehabilitation programmes known to the authors, and the various investigations undertaken to facilitate these programmes. The timing of the road damage inventories described in this paper in the context of Typhoons Pepeng and Ompong is also shown.
Geological and geohazard setting
Geology and climate
The geology, geomorphology and climate of this part of the Central Cordillera have been described by Hart et al. (2002) and Queano et al. (2007). The mountains rise to c. 3000 m a.s.l. They are underlain by Late Cretaceous to Eocene metavolcanic rocks (basalt and andesite lavas) that were intruded by diorite and granodiorite during the Eocene and Oligocene, and are overlain by epiclastic and sedimentary rocks deposited during the late Oligocene and Early Miocene. Pliocene and Pleistocene tuffs, volcaniclastic rocks and relatively massive, andesitic lava flows form the youngest members of the geological sequence. The broad distribution of these rocks along the Halsema Highway between La Trinidad and Mt Data is shown in Figure 1 and summarized in Table 2. Fracturing, shearing, alteration and brecciation have combined to create complex and weakened rock mass conditions (Fig. 5a) with fault gouge frequently exposed in landslide scarps and stream channels (Fig. 5b). Tropical weathering has weakened rock masses further and created up to 40 m deep saprolites and residual soils in some areas.
(a) Fault gouge enveloping rafts of destructured rock. (b) Thick fault gouge (typically composed of soft clay) susceptible to gully erosion and sliding.
Summary of alignment geology, geomorphology, geohazards and construction
The average annual rainfall in Baguio City is 3500 mm and, although this is not unusually high by world standards, it masks significant seasonal variation. The wettest months are June to September, each receiving in excess of 400 mm on average, and August has an average rainfall of more than 600 mm. High-intensity, short-duration typhoon rainfall (Table 1) can account for much of this monthly and even annual rainfall with, for example, the 24 h rainfall recorded on 4 July 2001, during the passage of Typhoon Feria, accounting for over 25% of the total rainfall recorded that year. Average monthly temperatures range between 17 and 21 °C.
Erosion and landslide activity
Adverse geology, steep terrain (predominantly 35–45°), deep weathering profiles and extreme rainfall combine to create a landscape that is highly prone to erosion and landslides (Figs 6 and 7 and Table 2). Deforestation may also have played an important role, particularly in relation to shallow slope failure (e.g. Forbes & Broadhead 2013). Furthermore, in the wider context of drainage basin hydrology, deforestation may have led to increased runoff rates and an intensification of incision and erosion, resulting in a greater susceptibility of valley side slopes to both shallow and deeper-seated instability.
Typical gullying and shallow debris slides triggered by stream incision and toe erosion.
Deep-seated (mostly translational) slides and flows. (images © Google Earth/DigitalGlobe/CNES/Airbus).
Deep-seated landslides are predominantly translational in mechanism (Fig. 7). Originally, some may have been seismically triggered, whereas others may be due to toe erosion and high groundwater tables in (1) often sheared and weakened rock masses, (2) volcaniclastic sedimentary sequences and (3) fine-grained soils found in deep weathering profiles (Hearn & Hart 2019). Slides formed in fine-grained weathering profiles may be quasi-rotational in mechanism and may also develop into flows as they travel downslope.
Many valley sides contain embayments that have formed through progressive erosion and slope failure (Fig. 8). Those formed by slope failure typically have well-defined arcuate head scarps. They frequently occupy the more gentle slopes close to ridge tops and may be the remnants of much older landslides that were seismically triggered. Rainfall-induced debris flows and debris avalanches are common on many hillsides, with source areas that appear to predominate on steep slopes adjacent to terraced cultivation (Fig. 9). In such circumstances causal factors may have included the effects of irrigation and ponded water in saprolitic and residual soils.
Typical valley side embayments. The one on the left was formed predominantly by mass movement, the one on the right by erosion (images © Google Earth/CNES/Airbus/Maxar Technologies).
Typical debris avalanche scars and channelized debris flows, with source areas on the steep flanks immediately below cultivated terraces.
Landslides and the Halsema Highway
Studies carried out prior to the design of the earthquake damage reinstatement works drew attention to the geological instability and geomorphological fragility of the terrain. The report on a post-earthquake reconnaissance survey undertaken by ECI (1991) as part of the ADB–IBRD-assisted Earthquake Project (Fig. 3) noted that the geological nature of the alignment frustrates maintenance and repair efforts, and structures often fail soon after their construction. Hazard assessments carried out by Geomatrix-Certeza (1996), as part of the IBRD-assisted Pre-Design Investigations, suggested that over 2000 landslide embayments affected about 75% of the alignment. Although some of these were more likely to be associated with progressive catchment erosion rather than landslides per se (e.g. Fig. 8), it is evident that many of the embayments crossed by the road have been formed by landslide activity and a number of these remain unstable (Hearn & Hart 2019). Continued ground movements in these embayments remove passive support to the in situ spurs that form their margins, causing loss of support, progressive stress release, further loss of strength and ultimate failure.
Landslide hazards that affect the Halsema Highway fall into three main categories, as follows.
Deep-seated, slow-moving landslides. These tend to be controlled by underlying geological structure and progressive valley side steepening or undercutting by stream erosion. They affect the road in at least 10 locations (Table 3), including the saddle at Pilando (Km 323; see Hearn & Hart 2019), the section of road through the village of Sinipsip (Km 317; illustrated in Fig. 10) and a 130 m length of alignment across sidelong ground close to the village of Sayangan (Km 301) that manifested itself through renewed ground movement during the construction period. These three landslide areas most probably predate the 1990 earthquake, although movements were probably reactivated and/or accelerated during this event.
Regressive valley side slope instability. Incision along drainage lines below the road causes instability on neighbouring side slopes (e.g. Fig. 11). This instability regresses upslope to the extent that it undermines the foundations of the road. Incision is normally most marked in stream channels that have formed within rocks weakened by faulting (Fig. 5a and b). Fault structures that cross ridge lines are often associated with incising streams on either side that have led to progressive lowering of the ridges themselves to form saddles (Hearn & Hart 2019), and the undermining of the road formation where it has been subsequently constructed across them.
Erosion and slope instability triggered by road runoff. Uncontrolled runoff from the road surface can be a major contributor to the initiation and acceleration of erosion and slope instability during the passage of individual typhoons. Significant damage, including road loss (Fig. 12), often occurs where cut slope failures or debris flows in side streams block roadside and cross-road drainage systems, causing runoff to flow over unprotected slopes below the road.
Subsidence to parts of the village and the Halsema Highway at Sinipsip.
Regressive slope failure on steep valley sides. Photograph date: October 2005.
Minor slide in cut slope (to the right of the excavator) blocked the roadside drain, diverting water across the road, triggering erosion, under-scour and collapse of the old retaining wall. Photograph date: November 2018.
The main high-volume, deep-seated landslides affecting the road and known to the authors
Land use change within the road corridor can have a profound effect on the performance of the road and its structures, principally through its effects on the drainage system. Roadside developments and the construction of accesses to commercial properties and agricultural production areas interfere with road drainage provision and can help concentrate runoff onto unprotected slopes.
Side drains and culverts are often blocked intentionally to keep water off land and property adjacent to the road, and this can have significant consequences on neighbouring slopes.
Earthquake damage reinstatement 2001–2006
Scope
Table 2 summarizes the outline scope of reinstatement works implemented between 2001 and 2006. The primary objective, geotechnically, was to reinstate sections of road that had failed during 1990. Damage caused by typhoons and tropical storms during the construction period also required rectification. There was also a requirement to widen those sections of road that remained narrower than the standard 6 m width. This involved consideration of earthworks and structural solutions and the suitability of the ground to both. Although proactive measures were taken wherever telltale evidence for the potential for future instability was identified, the purpose was not to ‘proof’ the road against all geohazard outcomes, as this would be impracticable and uneconomic. These issues were discussed at length in the Design Report, with the recommendation for a significant maintenance commitment in view of the continuing potential for slope instability. This pragmatic approach to risk management meant that slope improvement rather than slope stabilization was applied in two principal locations (Pilando and Sinipsip; see below) where slow-moving, deep-seated instability had given rise to road surface deformation, the extent of which did not warrant major expenditure (Hearn & Hart 2019).
General approach
Locations identified during the project feasibility study in 1997 where the entire road width required reinstatement are shown on the geological map in Figure 1. It can be seen that the large majority of these locations were coincident with faults. Rock masses would have been inherently weak in these areas and susceptible to erosion and mass movement (Hearn & Hart 2019).
As discussed by Hart et al. (2002), to avoid further destabilization of the slopes above the road, the preference was to reinstate the road width using retaining walls on the downslope side rather than further excavation into the hillside. Retaining walls up to 6 m in height were constructed in mortared masonry. Walls above this height were built as grid structures with mortared masonry infill panels between reinforced concrete (RC) columns and beams and high-friction rip-rap backfill with RC tie beams (Figs 13 and 14a, b). Because of concerns over bearing capacity and the hazards posed by erosion and landslide regression on the slopes below, walls were often founded several metres beneath existing ground levels. A nominal seismic acceleration of 0.1 g was accommodated in the design of retaining walls. No such provision was made in the design of slope works given the marginal stability of the majority of hillsides crossed by the alignment.
Standard cross-sections showing the use of masonry and RC grid retaining walls and shotcrete to stabilize and protect cut slopes in weathered rock and to provide sufficient width for road reinstatement.
Typical examples of road reinstatement and slope protection.
Responding to highly variable and changing ground conditions
The steepness and configuration of the topography below the road sometimes prevented wall construction, and hillside excavation was the only alternative. Excavations in unstable rock masses above the road were supported by masonry and RC grid retaining walls and protected, where necessary, with reinforced shotcrete (Figs 13 and 14c), although there was a practical limit on how far upslope this protection could extend. Failed sections of road across narrow saddles were reinstated by lowering the vertical alignment where geometrically feasible to create a wider platform in less weathered rock. Where space was insufficient and where ground stability problems remained, saddle sections were often reconstructed using RC grid walls (Fig. 15), with shotcrete side slope protection (Fig. 14d) and erosion control below culvert outlets and side drain turnouts (Fig. 16a and b). Gabion retaining walls, with concrete foundations, were used in a limited number of cases, particularly where ground conditions might give rise to differential settlement. These walls were easy and quick to construct without the concerns of mortar content and interlock that affected some of the masonry walls.
Road reinstatement across an unstable saddle, replacing a post-earthquake Bailey bridge with RC grid retaining walls.
Typical erosion control measures below culvert outlets and side drain turnouts, comprising baffles, cascades, checkdams and stilling basins (photograph dates: (a) July 2003; (b) October 2005).
The constructed works were based on the designs prepared in 1998 and finalized when ground conditions were exposed in excavations. Furthermore, the damage caused by Typhoons Feria, Gloria and Igme during the period between 2001 and 2004 (Table 1), and other high-intensity tropical storms, necessitated an almost constant reassessment of the changing requirements for road reinstatement, slope stabilization, drainage control and erosion protection (e.g. Fig. 17).
Deteriorating ground conditions between design (1998) and construction (2005). This illustration shows how slope regression on one side of a saddle caused the available platform width for road reinstatement to reduce from 15 m to less than 10 m during a 7 year period. Slope regression was arrested using shotcrete, drainage control and erosion protection (Fig. 14d).
Geotechnical supervision ended in 2005, and in 2006 the DPWH took over the responsibility for supervising works completion. In August 2005, three critical locations (Km 267 (case study 3, see below), Km 301 (Sayangan) and Km 317 (Sinipsip)) remained without construction works. Drilling and mapping investigations had been carried out between April and June of the same year to investigate slope instability that was either triggered or accelerated by typhoon rainfall. Although preliminary designs were prepared for road reinstatement and slope improvement based on these investigations, these works were never fully implemented owing to construction programme constraints.
Construction cost
The final earthquake reinstatement cost amounted to the equivalent of PhP12 M (US$0.27 M) per km, compared with an original contract price in 2001 of PhP8.5 M (US$0.19 M) per km. One of the main reasons for the price overrun was the cost of reinstating sections of road that became severely damaged during the typhoons and tropical storms that occurred between 1998 and 2006, with approximately 50% of the additional cost being spent on erosion control below culvert outlets and slope protection works.
To put these costs into a national perspective, IBRD-assisted road improvement projects in less mountainous parts of the Philippines, which were completed at approximately the same time (IBRD 2008), had outturn costs of up to PhP10 M per km. The cost of improving the 66 km Bontoc to Tinglayan mountain road (Fig. 1), completed in 2012, was PhP14 M per km (Ngalob 2011). By contrast, the cost of improving the 31 km mountain road between Abatan and Cervantes (Fig. 1), completed in 2009, amounted to PhP21 M per km (Anonymous 2011) while the cost of improving the 104 km long Baguio City to Aritao Road (Fig. 1) completed in 2008, was between PhP20 M and PhP30 M per km (Sasao 2010). Both of these more expensive roads involved the construction of several bridges and, in the latter case especially, costly riverside scour protection works. Consequently, for a mountain road in the Philippines, the earthquake damage rehabilitation cost of the Halsema Highway might be considered to have been more or less ‘par for the course’.
The investments made in road rehabilitation in many other parts of the world are often significantly higher than the reconstruction cost of the Halsema Highway. For example, Alexeeva et al. (2011) showed that the cost of rehabilitation and reconstruction of 7 m wide (two-lane) carriageways in eastern Europe and central Asia at 2009 prices was generally between US$0.2 M and US$0.5 M per km. Although some of these road works (perhaps the more expensive) will have been in hilly and mountainous areas, these terrains are unlikely to be as extreme or as unstable as that crossed by the Halsema Highway. Despite the cost overrun, the earthquake and typhoon reinstatement of the Halsema Highway should therefore be considered as low cost by international comparison.
Post-construction condition surveys
An inventory carried out in late 2006 by Nippon Koei (2007) registered 113 soil slides, 70 rock slope failures, 17 debris flows, two deep-seated landslides (possibly at Pilando, Sinipsip or Sayangan; Table 3) and 116 cases of road undermining between La Trinidad and Mt Data. Although this dataset cannot be verified, it probably reflects damage caused by Typhoon Florita in July 2006. This typhoon was accompanied by significant rainfall (Table 1; Fig. 3) and triggered almost 50 landslides in Baguio City (Dumlao 2006). It provided further evidence of the continuing susceptibility of the landscape (and the roads constructed within it) to typhoon rainfall, and underlines the fact that roadside slope management is, accordingly, always a matter of work-in-progress, requiring long-term slope maintenance and repair programmes to be devised and implemented. As far as can be ascertained, none of the 2001–2006 works were affected by this typhoon.
Typhoon Pepeng damage assessment
In October 2009, Typhoon Pepeng deposited 1868 mm of rain in Baguio City during a 7 day period, 1221 mm during a 3 day period (Table 1) and 640 mm during a 12 h period (Santos 2013). This represented the largest 3 day rainfall likely to have affected the Halsema Highway since Typhoon Feria in 2001. The 7 day rainfall was the largest in the entire rainfall record for Baguio City available to the authors, which dates back to January 1995.
Although the track of Typhoon Pepeng crossed the north of Luzon (i.e. further to the north than the Halsema Highway (GFDRR 2009)), large parts of Luzon were significantly affected, with 58 fatalities in Baguio City alone owing to landslides (Santos 2013). The rainfall distribution map reproduced in GFDRR (2009) shows the entire Halsema Highway to be affected by an 11 day rainfall of > 1000 mm. According to GFDRR (2009) this rainfall resulted in floods with 50 year return periods in some drainage basins.
Field observations
A visit was made to the road by one of the authors in May 2010 during which damage along the first 28 km from La Trinidad was recorded. Despite the fact that the visit took place 7 months after the passage of Typhoon Pepeng, the following outcomes could be identified.
The road had been blocked or partly blocked by debris fans, debris slides and rock slides in over 30 locations for a total length of c. 850 m.
The road edge had been exposed through failure of the slopes below in five locations for a total length of c. 90 m.
Partial loss of carriageway or minor settlement had occurred in four locations for a total length of c. 75 m.
Total loss of carriageway as a result of slope failure below had occurred in three locations for a total length of almost 300 m (Fig. 18).
Failure of entire carriageway over a length of c. 150 m as a result of deep-seated rock slide (photograph date: May 2010).
As described below, this first 28 km (Km 257–285) appears to have been the most heavily affected section of the entire alignment as far as Mt Data during Typhoon Pepeng.
Remote sensing observations
Google Earth contains post-Typhoon Pepeng epochs of high-resolution satellite imagery dated December 2009 and March and April 2010. This imagery covers a little over 60 km of the 84 km length of alignment between La Trinidad and Mt Data. The remaining 20 km or so is captured by January 2012 imagery, which is also of high resolution. The 2009–2010 imagery allows areas of slope failure and erosion above and below the road to be identified, as well as sections of road itself that have been affected by undermining and subsidence (Fig. 19). Although blockages to the road would have been cleared by the DPWH before any of this imagery was taken, it is still possible to speculate where they had occurred from the extent of landslide deposits immediately above the road and the evidence for debris placed onto the slopes below. Obviously, the January 2012 imagery will be less indicative of Typhoon Pepeng damage, as landslide scars and debris may have become revegetated and the required road repairs implemented during the intervening period. Nevertheless, with a little interpretation, it is possible to gain some impression of the distribution of damage, and this is shown in Figure 20a, according to damage type and length of road affected.
Example of high-resolution imagery dated December 2009 used to develop a post-Typhoon Pepeng inventory of road damage. Image © Google Earth/DigitalGlobe.
Summarized distribution of post-Typhoon Pepeng (a) and post-Typhoon Ompong (b) road damage.
Review of damage distribution
The distribution of road damage attributed to Typhoon Pepeng is concentrated in three main cluster areas: Km 264–269, Km 280–283 and Km 312–318. All three clusters are underlain by metavolcanic and dioritic rocks, or a combination of the two, and are coincident with a greater number of mapped fault structures than elsewhere along the alignment (Fig. 1). By far the greatest density of damage, including road loss and road edge exposure, was recorded between Km 264 and Km 269 where there were 12 cases of road blockage, three of road edge loss and four of road failure (Fig. 20a and Table 4). Case Study 3 (see below) is located along this section of road, where 1997–1998 drilling investigations identified closely jointed, sheared and faulted andesite with multiple failure surfaces to depths in excess of 30 m. These ground conditions may be representative of the 5 km section of road as a whole that crosses steep, sidelong ground forming the slopes of what appears to be a fault-controlled valley side (Fig. 1).
Summary of road damage attributed to Typhoons Pepeng (October 2009) and Ompong (September 2018)
Without information on rainfall distribution along the alignment, it is unclear why a typhoon that tracked to the north of the Halsema Highway should have its greatest impact on the southernmost part of the alignment. Typhoon Ondoy (Ketsana), which preceded Typhoon Pepeng in late September 2009, tracked across the southern part of Luzon, causing significant damage in Manila (e.g. GFDRR 2009) but deposited only c. 80 mm of rain in Baguio City during a 4 day period. Typhoon Ondoy was, therefore, unlikely to have been responsible for the damage between Km 264 and Km 269, unless this area received higher rainfall than Baguio City. The detailed distribution of typhoon rainfall in mountainous terrain is complex and goes largely unrecorded. It often occurs as geographically discrete cells, with intense rainfall affecting one catchment while adjacent catchments receive much less.
Typhoon Ompong damage assessment
Typhoon Ompong tracked across the northerly tip of Luzon. However, according to the Ompong rainfall distribution map prepared by PAGASA (2018), the highest rainfall (generally 500–1000 mm in 4 days) was deposited over an area to the south of this track and was centred approximately on the entirety of the Halsema Highway.
Field and remote sensing observations
Figure 20b shows the distribution of damage observed during a 3 day visit by the authors to the Halsema Highway in November 2018, c. 2 months after the passage of Typhoon Ompong. This damage is summarized in Table 4 according to hazard type and impact on the road. Two months would have been sufficient time to allow the DPWH to clear all of the debris from the carriageway, but there was clear evidence, in the form of roadside debris, dumped spoil, damage to vegetation, and blocked culverts and stream channels, that provided good indications of where the road had been blocked or partially blocked by cut slope failures and landslides, including debris flows.
Review of damage distribution
Apart from a single, minor case of road edge loss near Km 269, all Typhoon Ompong damage was located between Km 290 and Km 330 where a total of almost 380 m of road had failed or had been undermined by new (first-time) failure on the slopes below (Table 4). The greatest damage to the road pavement owing to below-road slope failures occurred between Km 310 and Km 320 with a cluster recorded between Km 318 and Km 320; that is, immediately north of Sinipsip (Fig. 20b). In this area, the road is underlain by destructured, faulted and highly weathered metavolcanic rocks (principally andesite; Fig. 1). Interpretation of Google Earth imagery suggests the presence of a degraded back scarp on the cultivated slopes close to the ridge-top above the road. Erosion in the stream channels below the road may have reactivated some of this much earlier movement.
The section of road between Km 300 and Km 318 was the worst affected by slope failures that blocked the road, debris flows and scour below culvert outlets. This section of the alignment is underlain by dioritic and andesitic rocks (Fig. 1), as are large parts of the remainder of the alignment, and so underlying lithology alone cannot be the cause of this concentration of damage. Although side slopes are steep in this section, they are not the steepest along the alignment. Elevation, however, is the highest of the entire alignment (Fig. 1), and it may be the case that orographic rainfall was more severe here. Additionally, there appears to be a higher incidence of cultivated land on the slopes above the road corridor than elsewhere, and this may be an important factor, certainly in the incidence of debris flows.
A considerable proportion of the damage was judged to have been caused by the instantaneous effects of uncontrolled road runoff (e.g. Fig. 12) and erosion below culvert outlets. Also, there were several cases where the road had been blocked by debris flows, including instances where below-road retaining walls had been destroyed by the passage of these flows over the road edge. As illustrated in Figure 9, many of the source areas of these debris flows originate on steep slopes immediately below cultivated land, and it may be the case that catchment runoff patterns have changed significantly in places since the earthquake reinstatement works were implemented.
Performance of the 2001–2006 constructed works
Almost all of the damage associated with Typhoon Pepeng over the first 28 km of the road had occurred in locations that had not been problematic during the design and construction of earthquake reinstatement works. Indeed, virtually all of the previously reinstated sections of road, including retaining walls and scour protection works, appeared to have survived the effects of Typhoon Pepeng. Similarly, almost all of these works remained intact following Typhoon Ompong. This vindicates the approach adopted in 1997–1998 and 2001–2005, whereby (1) detailed engineering geological observation and mapping, supported by trial pitting and drilling investigations, were used to develop as robust a design as possible and (2) the constructed works were implemented with close engineering geological supervision, ensuring they were compatible with the ground conditions exposed and able to accommodate ground conditions that had changed since the original designs were prepared.
Case studies
Hart et al. (2002) used two case studies (Km 282 and Km 314) to illustrate the technical challenges faced by the design, and these areas are reviewed in Figures 21 and 22 in the context of their post-Typhoon Ompong condition. A third case study (Km 267) is included in Figure 23 to illustrate how the scale and pace of erosion and slope instability defied several attempts at reinstatement during the construction period. Although this area has now stabilized, a deep-seated slope failure nearby removed a different and previously stable section of road for c. 150 m during Typhoon Pepeng (Fig. 18), and is included in the case study by way of contrast. Finally, a fourth case study (Km 264) is added (Fig. 24) where road damage did not take place until the construction period itself. In this case, although the potential for road damage was recognized and accommodated in the 1998 design, the works could not be implemented in time before erosion and slope instability occurred during construction. The principal observations, conclusions and lessons learnt from these case studies are summarized in Table 5. The implications are carried through to the Discussion.
Case study 1. Km 282: road reinstatement following extensive road failure across steep valley sides and a deep-seated landslide.
Case study 2. Km 314: road reinstatement at the site of a large active landslide above the road.
Case study 3. Km 267: erosion and slope failure frustrates attempts to reinstate the road during construction.
Case study 4. Km 264: road bench stabilization on a saddle affected by severe erosion and slope failure on both flanks during construction.
Observations, conclusions and lessons learnt from the four case studies
Discussion
Engineering performance
The design of earthquake damage rehabilitation works focused on the need to reinstate the road width as the first priority. This required careful consideration of founding conditions during design and construction, and is the main reason why almost all of the retaining structures remain intact after more than 15 years of service. Emphasis on drainage control and scour protection was also a critical factor in helping to re-establish or maintain stability.
Excavation into the hillside above the road was necessary in several places to accommodate road reinstatement and road widening, and also to improve the stability of previously cut, over-steep excavations (case study 1). This often required the use of retaining walls, revetments, catch walls and shotcrete to improve the stability of excavated slopes. There was a limit, however, on how far upslope shotcrete could be taken because of access constraints, right of way issues and cost considerations.
Generally, the use of shotcrete proved successful in providing erosion protection and stabilization against shallow instability in jointed and weathered rock slopes. It continues to perform as intended and is now fairly inconspicuous in the landscape (Fig. 25). Without this protection many of these slopes would have ravelled and eroded under subsequent heavy rainfall, posing long-term problems for road maintenance, and leading to the loss of agricultural land on the hillsides above.
Shotcreted excavation above RC grid wall, 15 years after construction. Photograph date: 2018.
Important lessons for road construction, rehabilitation and maintenance elsewhere in the Central Cordillera and beyond
New road construction
The tendency for many of the road failures to have occurred in the embayments mapped from aerial photography (and now evident in Google Earth 3D imagery) highlights the importance of maximizing the interpretation of remote sensing in the selection of alignments for future road construction. Although not available to the 1997 and 1998 feasibility and design studies, or the 2001–2005 construction supervision team, Google Earth 3D now provides excellent opportunities for the mapping of these embayments and other landslide-related morphology in the Central Cordillera (Hearn & Hart 2019). Geomorphological interpretation of this imagery should be undertaken as part of the feasibility study for any new alignment, followed by ground truthing and focused field mapping to assess and compare route options and engineering alternatives.
Judicious earthworks design, which takes account of often extremely variable rock mass properties, is also critically important, as is the need to find ways to avoid deep and steep excavations in landslide-susceptible rocks and soils. This can be achieved only through careful alignment design and choice of cross-section. Inevitably, in such steep terrain, earthworks design alone will be insufficient, and many slopes and sections of road will need to be supported and protected using retaining walls and revetments. During design, wall foundation levels will need to be confirmed by ground investigation and there will be a requirement for engineering geological verification of all ground conditions during construction (case studies 1, 3 and 4).
Drainage control, and the provision of adequate scour protection below culvert outlets and other drainage turnouts, has been demonstrated along the Halsema Highway to be of paramount importance (case studies 1, 3 and 4). Close attention to drainage detail is possibly the single most important element of the design of any new road in the Central Cordillera once its alignment has been optimized.
Road rehabilitation and maintenance
Although a continuing commitment to drainage management is essential, it is not possible to take preventive action against all eventualities, given the fact that something as minor and as unpredictable as a blocked side drain can lead to significant road damage within minutes (Fig. 12). Another constraint relates to the management of the land on either side of the right of way, and sometimes within it. Agricultural practices, and the construction of accesses for commercial and residential use, can have profound effects on drainage patterns in the road corridor, often leading to unpredictable damage during the passage of intense typhoon rainfall. Although liaison between road authorities and neighbouring farming and commercial stakeholders can help reduce these impacts they cannot be eliminated entirely. Adequately resourced, emergency response will therefore continue to form an important element of road maintenance in the Central Cordillera. The Kennon Road, for example (Figs 1 and 26), originally constructed in the early 1900s, is one of only two roads that link Baguio City with the lowland plains and transport systems to the south, and was closed to through-traffic in 2018–2019 for 8 months owing to landslide hazards.
Heavily trafficked Kennon Road, now more often closed than open, as a result of landslides. Photograph date: 1998.
Specific engineering geological scenarios
Roads constructed in the Central Cordillera and the wider mountainous terrain of the Philippines encounter similar geomorphology and ground conditions to those described above for the Halsema Highway, although perhaps to less extreme degrees. Although each alignment will face its own unique set of ground conditions, each requiring careful investigation, it is possible to define the following broad engineering geomorphological scenarios:
(1) contoured alignments across steep valley sides and embayments in erosion- and landslide-susceptible rocks and transported soils;
(2) sections of road across narrow, weak and erosion-susceptible saddles (i.e. low points in ridge lines);
(3) alignments that follow lower valley sides with often technically challenging and expensive bridging requirements that are also susceptible to debris flows, flooding and river scour;
(4) alignments that use climbing corridors (hairpin stacks) to gain or lose vertical elevation on valley sides.
The Halsema Highway encounters multiple cases of scenarios 1 and 2 and, owing to its largely ridge top and mid to upper valley side location, none of 3 and 4. Tables 6 and 7 contain observations and recommendations for future practice from the experience gained on the Halsema Highway in the context of scenarios 1 and 2 respectively. Observations and engineering guidance with respect to scenarios 3 and 4 from other parts of mountainous Asia have been given by, for example, Fookes et al. (1985), TRL (1997) and Hearn (2011, 2017).
Engineering geology of steep, sidelong ground in the Central Cordillera
Engineering geology of saddle crossings in the Central Cordillera
Conclusions
Mountain ranges in South and South East Asia, below c. 3000 m a.s.l., are probably among the most geomorphologically dynamic terrains in the world, on account of their geological instability, extreme weather systems and intensive land use. The Central Cordillera of the Philippines is a profound illustration, and must rank among the most challenging places to build and operate roads and other infrastructure.
By Philippines standards the cost of the Halsema Highway road rehabilitation works was ‘par for the course’, and engineering on the precipice did not involve the use of high-cost engineering solutions such as those applied in other parts of the world, including several neighbouring countries in Asia. Instead, it focused on the use of engineering structures and construction techniques already established in the country. The exception, at the time, was the use of shotcrete slope protection, which had had limited prior application and which, on the whole, has performed well. Bio-engineering measures were also designed and implemented, although these had limited impact on the performance of the works, which were focused more on the stability of jointed and weathered rock masses than on the protection of soil slopes.
Generally, the earthquake rehabilitation works have performed well since their construction more than 15 years ago, despite the passage of three record-breaking typhoons and numerous tropical storms. Close attention to engineering geological and geomorphological detail through field mapping and targeted ground investigation during both design and construction were responsible for achieving this outcome. The heavy rainfalls that occurred during the construction period itself caused significant additional damage to the road and adjacent slopes, requiring continued engineering geological assessment and geotechnical input during works supervision. Changes to land use, and especially drainage management, within the road corridor during construction also led to significant erosion and slope instability that required engineering geological resolution.
Despite the overall success of the Halsema Highway earthquake rehabilitation, there were three main locations where the anticipated depth of ground movement was such that slope improvement rather than slope stabilization was considered to be the most pragmatic option. During the feasibility and design studies it was decided that the apparent low rates of ground movement at two of these locations (Pilando and Sinipsip) did not justify major expenditure in stabilization, especially given the level of geotechnical uncertainty that prevailed. The third location (Sayangan) did not manifest itself until the construction period and, although slope improvement designs were prepared following detailed mapping and ground investigation, there was insufficient time to implement them during the geotechnical supervision period. By 2018, ground movements at this location appeared to have accelerated, so much so that this has become one of the most problematic sections along the entire alignment.
The most important lessons that can be taken from the experience gained on the Halsema Highway and applied to other mountain roads in the Central Cordillera and the wider region are summarized below.
(1) Engineering geological and geomorphological investigation during the design and construction of new roads and during the rehabilitation and improvement of existing roads is vital to engineering performance. Sufficient programme time is required to gather surface and subsurface monitoring data to inform engineering geological interpretation for engineering decision-making.
(2) Deep-seated and slow-moving landslides are frequent in the landscape and should be identified and avoided wherever possible during route selection for new roads and local realignment on existing roads. Where avoidance is not feasible, low-cost engineering interventions may be the most effective approach to slope management, depending upon observed rates of movement, traffic volumes and wider risk implications.
(3) Investment in the robust design and construction of retaining wall foundations and scour protection works is critical to the performance of the road.
(4) Drainage control, and the prevention of erosion, can lead to a progressive improvement in the stability of existing problem areas, and significantly reduce the likelihood of new ones developing, even in the absence of other engineering interventions.
(5) Although a continuing commitment to drainage management is essential, the provision of adequately resourced emergency response systems, particularly during and after the passage of typhoons and other high-intensity storms, must form a critical element of road maintenance.
Returning finally to the opening quotation from West (1918), it is hopefully evident to the reader that every effort was made to ensure adequate foundations for the reinstatement works implemented. The fundamental problem with ground conditions along the Halsema Highway is the fact that ruling out ‘an insignificant boulder’ is not nearly enough; deep weathering profiles and often extremely poor rock quality pose significant challenges and, consistent with the recommendations of West (1918), require the designer to rely on observation and dogged determination to get the design right for each set of circumstances.
Acknowledgements
The authors were involved in the IBRD-assisted design and construction of the Halsema Highway earthquake damage rehabilitation works while they were employed at Scott Wilson, now AECOM, as part of a consortium led by BCEOM, now Egis International, on behalf of the Department of Public Works and Highways, Government of the Republic of the Philippines. The geotechnical design team was led by C. Chant. Daewoo E&C was the main contractor. The authors were assisted in the production of drawings and photographic annotations by K. Finlay of KJ Creative. Unless otherwise indicated, all rainfall data quoted are taken from the Government of the Philippines, Freedom of Information website https://www.foi.gov.ph. R. P. Martin provided useful comments on drafts of this paper.
Funding
This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.
Author contributions
GJH: conceptualization (equal), data curation (equal), investigation (equal), methodology (equal), writing – review & editing (lead) and JRH: conceptualization (equal), data curation (equal), investigation (equal), methodology (equal), writing – review & editing (supporting)
Scientific editing by Cherith Moses, Colin Serridge
- © 2020 The Author(s). Published by The Geological Society of London. All rights reserved
References
Engineering on the precipice: the performance of mountain road rehabilitation in the Central Cordillera of the Philippines
