The great majority of the historical and recorded earthquakes used in the probabilistic analysis of the DBE were tectonic earthquakes. There have been some earthquakes in the region associated with volcanic activity and eruptions, for example on Dominica and Montserrat. Earthquakes of this type can do a great deal of damage, but their effects are generally confined to a local area, thought usually to be less than 2 kilometers in radius. Robson (1964) does attribute a long series of small earthquakes felt in St. Lucia in 1906 to this type of earthquake (Shepherd).

For the pseudo-static analysis of the stability of the slope at Black Mallet, a more empirical approach is adopted, as recommended by Seed (1979). In a pseudo-static method of analysis the effects of an earthquake on a potential slide mass are represented by an equivalent static horizontal force determined as the product of a seismic coefficient 'g,' and the weight of the potential sliding mass. The numerical value of 'g' depends on the intensity of the earthquake. Freeman (1932) proposes the follow values for 'g':

Professor Seed (1979,1983) has stated that virtually any virtually any earthen structure can withstand moderate earthquake shaking with peak accelerations of about 0.2g with no detrimental effects.

Table 4.1 contains a record of seismic events in St. Lucia front 1997 to 1999 as reported by the office of Disaster Preparedness. During the period of June to November 1999, five earthquakes of magnitudes ranging from 5.0 to 2.5 have been reported. Also attached is an insert of additional seismic data on St. Lucia from the Seismic Research Unit of the University of the West Indies, St. Augustine, Trinidad.

The field investigation is the central and decisive part of a study of landslides and landslide prone areas. The investigation serves two essential purposes, namely:

Unstable areas prone to sliding usually exhibit symptoms of past movement and incipient failure; most of these can be identified in a field investigation before design. Such investigations can show how to prevent or at least minimise future movements mid they can suggest alternative areas that are less likely to slide. Ideally, field investigations should commence long before construction is anticipated and sometimes continue long after the area has been changed by the anticipated construction.

The subsurface exploration program at Black Mallet commenced on October 18^th, and was completed on November 30^th. 1999. Five piezometer boreholes (SP99-BM-1 to SP99-BM-5) were drilled to depths varying from 6.0 m to 8.25 m below existing ground surface at selected locations within the movement zone, utilizing hollow stem augers attached to a track-mounted, DHM-163R drill rig with a motorised cathead. A 75 mm diameter tri-cone drill bit was used to penetrate rock fragments and boulders encountered during the subsurface exploration program.

Standard Penetration Tests (SPT) were conducted at regular intervals of depth in all the boreholes in accordance with ASTM D1586. To perform this test, a spoon sampler is lowered down the borehole until it rests on the layer of soil to be tested. It is then driven into the soil for a length of 450 mm by means of a 65 kg weight hammer free falling 760 mm for each blow. The number of blows required to drive the last 300 mm is recorded and this figure is designated the 'N'-Value of the soil, The first 150 mm of driving is ignored because of possible loose soil in the bottom of the borehole from the boring operations. After the sampler has been removed from the borehole it is opened and its contents examined. The recovered disturbed soil samples were visually classified in the field, placed in plastic bags and scaled to retain their natural moisture content.

Three undisturbed shelby tube samples were recovered at selected depths in the piezometer (SP) boreholes. Hollow stem augers and a tri-cone drill bit were used to advance the boreholes to the required sampling dept. A 75 min diameter hollow, metal "shelby" tube is then fitted to drill rods attached on ends and lowered through the hollow stem auger and pushed into the soil with the drilling head to recover the undisturbed sample. A pocket penetrometer was utilised in the field to determine the unconfined compressive strength of the subsoils.

On completion of drilling, a perforated, 38 mm diameter plastic PVC standpipe piezometer was installed in the five borcholes to monitor groundwater levels and to conduct hydraulic conductivity tests. In boreholes SP99-BM-1 and SP99-BM-2, the tip of the piezometers were located at mid-height of the sand layer. SP99-BM-3 to SP99-BM-5 piezometers were installed with their tips located 1.5 in above the base of the colluvial material.

The piezometer installations were conducted as follows:

The borehole was backfilled with auger cuttings up to the base of the soil layer to be tested. Bentonite pellets were soaked in diesel fuel to retard their reaction with water and were lowered down the borehole to form a 0.5 m thick seat at the base of the layer.

The lower 3.0 m length of the PVC plastic pipe was slotted, at 150 mm intervals with a hacksaw and a scaled end lowered to the selected elevation in the borehole by adding 6.0 m lengths of plastic pipe connected on ends by couplings and sealed with PVC adhesive. The borehole was then backfilled to within 2.0 m of ground surface with 12 mm diameter clean, crushed rock placed around the piezometer. A betitonite/cement slurry was used to top the upper 2.0 m of the borehole to form an impermeable seal to prevent the ingress of surface water into the piezometer. A cap was placed over the protruding end of the standpipe piezometer and is removed when recording groundwater levels.

The plastic casing used for this project is of 70 mm diameter and 3.0 m lengths and is made of an ABS (acrylonitrite/butadiene/styrene) compound. The casing is manufactured with aligning grooves for controlling the orientation of the probe and are joined together with rigid couplings.

Five boreholes (SI99-BM-1 to S199-BM-5) were drilled to depths ranging from 6.5m to 11.0m for slope inclinometer installation. The boreholes were drilled approximately 1.5 m distance from the piezometer locations to observe any correlation between groundwater fluctuations and subsurface movement.

The plastic slope inclinometer casing sections were layed out and couplings marked to show location of the alignment key. A bottom cap was placed at one end of the first section, sealed with ABS cement and taped to prevent soil or grout from entering the casing. The remainder of the casing was assembled by adding on sections connected with couplings and scaled with ABS cement. During the installation process, the axis of one pair of grooves was kept parallel to the anticipated direction of movement i.e. downslope. The orientation was maintained throughout the installation process. A grout slurry composed of Portland cement, bentonite and water was used to backfill the borehole and to provide stability to the casing.

The slope inclinometer casing was installed in vertical boreholes, which pass through a zone of suspected movement. The bottom of the casing was anchored in stable ground and serves as a reference. The inclinometer probe was used to survey the casing and establish its initial position.

The recovered disturbed and undisturbed soil samples were placed in plastic bags in the field and scaled to retain their natural moisture content. The disturbed samples were utilised in our laboratory to determine insitu moisture content, grain size distribution, Atterberg limits and to confirm the field classification of the subsoils. Three undisturbed shelby tube samples were packaged and transported by air to Trintoplan Consultants Ltd. in Trinidad for shear strength testing in their laboratory. The results of the shear strength tests are as follows:

The general subsoil stratigrapily at the site consists of colluvial material overlying highly weathered andesite bedrock. A sand deposit exists between the colluvium and bedrock at the location of boreholes SP99-BM-1 and SP99-BM-2 and may represent an old river deposit was not encountered in the channel which was buried by the colluvium. The sand deposit was not encountered in the borcholes located upslope to the east, west and south of the site. The tip of the standpipe piezometers installed in boreholes SP99-BM-1 and SP99-BM-2 were located in the sand material and the piezometers showed water level readings indicative of artesian pressure from a confined aquifer.

Colluvium is defined as any soil deposit which has moved downstope to its present position by a mass wasting process such as creep, slope wash, erosion or landslide. This material was encountered in all the boreholes at this site. It consists of boulders and cobbles within a sandy, clayey silt matrix. The stratum ranges in thickness from 4.0 m in boreholes SP99-BM-3 and SP99-BM-4 to 6.0 in boreholes SP99-BM-2 and S199-BM-5. Most values for the natural moisture content of tl-ils material ranged from 12 to 36

The 'N'-values from the Standard Penetration Test ranged from 2 blows to 46 blows per 0.3m indicative of a soft to very stiff material. The 'N'-value generally increases with depth. Standard Penetration Test (N-values) are included on the borehole logs.

In boreholes SP99-BM-1, SP99-BM-2 and S199-BM-1 a sand deposit was encountered in a confined aquifer under hydrostatic pressure. The sand layer ranges in thickness from 2.0 in to 4.25 in. A sand sample was recovered at a depth of 3.0 -4.0 m below the riverbed at the toe of the failed slope during the excavation for placement of the rock toe berm.

The 'N' values from the Standard Penetration Test ranged from 10 blows to 44 blows per 0.3m. The dense values were from boreholes SP99-BM-1 and S199-BM-1 located in close proximity to the Marchand River where a drainage path was available for dewatering of the sand grains. Upslope at the location of borehole SP99-BM-2 the sand is loose and is confined under hydrostatic pressure.

The following weathering classification scheme proposed by Anon (1972) was found to be workable for the weathered rocks in the tropics. The terminology applied to the bedrock material description is as follows:

The Limit Equilibrium method for slope stability analysis was used in this design to determine the magnitude of the factor of safety. When a slope has failed, the factor of safety is unity and the analysis can then be used to estimate the average shearing resistance along the failure surface. The factor of safety is that factor by which the shear strength parameters may be reduced in order to bring the slope into a state of limiting equilibrium along a given slip surface.

A commercially available computerized model 'G-SLOPE' developed by Mitre Software Corporation of Edmonton, Alberta, Canada was used to perform back-analyses of the failed slope (cross-section 'A-A') and in the design of slope remedial stabilisation works. Three sections (A-A'; B-B'; and C-C') were selected to perform effective stress stability analyses on the failed slope. The analyses utillsed the simplified Janbu limit equilibrium method of slices for an irregular slip surface. A potential slip surface is selected and the potential sliding mass is divided into a number of vertical slices and the stability of each slice is considered in turn, on the assumption that the factor of safety for each slice is equal to the factor of safety of each of the others. Each slice is acted upon by its own weight, which produces shearing and normal forces on its vertical boundaries and shearing and normal forces along its base. The shearing and normal forces acting on the vertical boundaries depend on the stress-deformation characteristics of the material comprising the sliding mass.

The factor of safety 'FS' of a slope Is usually defined as the ratio of available shear strength to shear stress on the critical surface. The factor of safety in the simplified Janbu method is expressed by the following equation:

FS = f {[ c'b + (W - ub) tan O'] [ 1/cos m]}

Where,

The back-analysis performed at section A - A' on the failed slope was to determine the shear strength of the soil mass at failure. In order to conduct this exercise, the technique applied involved analysing the original slope geometry and piezometric levels before failure and working backwards by altering the shear strength parameters of the subsoils until a factor of safety of unity is achieved. A value of zero cohesion was assumed for the saturated silty sand layer and for the partially saturated colluvium. The results of the back-analys is show the shear strength of the sand layer at failure was 0 = 21.0 degrees.

Partially saturated colluvial or tropical residual soils develop excessive negative pore water pressure (soil-suction) during the dry season which helps to maintain stability. During heavy successive rainfalls negative pore water pressure is reduced dramatically as rapid infiltration occurs resulting in a reduction of shear strength of the overlying colluvial material and failure generally occurs.

A 1.0 in thick soil layer was used in the analyses to represent the equivalent surgharge applied by buildings on the slope.

The three cross-sections chosen for stability analyses were as follows:

The stability of composit slip surfaces was performed through the sand layer moving on the bedrock surface.

The following soil parameters were utillsed in the effective stress analyses:

Documented information on land degradation due to landslides in St. Lucia is contained in a few published and unpublished reports, including DeGraff, 1985; DeGraff et al, 1989 and Prior and Ho, 1972. The historical record of tropical storms and hurricanes, which have affected the island of St. Lucia from 1938 to the present, suggests that during this period there were at least thirteen major landslide-producing storm events on the island. In addition to these, landslides invariably develop on an annual basis during the rainy season or during high-rainfall periods of non-storm years. The high frequency and widespread distribution of these slope failures are evidence that landsliding is a dominant erosional process on the island.

Several types of landslides have been documented, including debris flows, debris slides, rock falls, rock slides and landslide complexes. Of the several landslide types, debris flows are the most common and are the main contributor to land degradation. These flows occur in soil or weathered rock, are typically small in size, and are initiated as shallow failures in the upper regions of tl-ie slope. The failed material, saturated with water becomes mobilised, flows downslope and carves deep erosion channels. Debris flows also erode the upper slopes of road bed, leading to the collapse of several road foundations. The significant volume of soil and other landslide debris carried by these flows to streams result in increase sedimentation of rivers and contribute to flooding downstream.

Colluvial slopes are generally of limiting or marginal stability because over geological times the formation of a colluvial slope is a dynamic process, which naturally involves slope movement with gravity and water.

Factors which contributed to the landsliding at Black Mallet/Maynard Hill were:

Karl Terzaghl's (1942) concept of slope stability are specifically expressed in the following words: "Theoretically, a factor of safety (FS) of 1.0 would mean a slide but in reality a slope may remain stable in spite of FS being smaller than 1.0 and it may fall at a factor of safety greater than 1.0, depending on the character of the slope forming materials." This statement clearly indicates that in Terzaghi's opinion a slope may be stable or unstable even if the computed factor of safety is greater than 1.0.

The initial failure of the slope at Black Mallet/Maynard Hill has since extended uphill by a progressive mechanism. Residents have reported the appearance of tension cracks around and within their houses. A well defined tension crack can be observed approximately 48 in upslope from the initial scarp.

When the toe of a large slope has failed, the removal of this toe support induces failure of the remaining slope. This can happen even when in entire slope has a factor of safety with respect to residual strength in excess of unity as long as some portion, usually the toe, has a factor of safety of less than 1.0. The initial failure occurs and reduces the stability of the slope that remains; then failure regresses until a region where the factor of safety is in excess of 1.0.

Generally, stabilisation involves some or all of the following:

The water collected must then be discharged in such a way that it does not affect the stability of adjacent areas. Efforts should be made to identify the source of recharge to the sand layer at the toe of the slide at Black Malley and every attempt should be made to divert surface run-off from Parker Hill eastward to Ravine Touterrelle. Once installed, drains must be maintained in proper working condition.

The existing road drains at Black Mallet and Maynard Hill are poorly constructed and should be replaced with properly designed open drains capable of containing run-off flows. The volume of water to be collected depends on rainfall intensity, and duration and catchment characteristics such as slope geometry, area and surface coverings. The volume should be determined in accordance with local engineering practice.

During the site investigation work at the site, the accelerated slope movement of 65.0 mm a ay warranted the immediate design and construction of a stabilising rock berm at the toe of the slope to arrest slope movement. Mr. Thomas Walcott, a consulting Engineer assisted the author by designing the toe berm, using soil parameters obtained from the field classification of the subsoils and the results of the instrumentation montioring.

The stabilising berm was constructed by excavating that portion of the Marchand river at the toe of the slope to bedrock at a depth of 1.5 in below the river bed and backfilling the excavation with boulders up to 2.0 m in diameter. The berm was 5.0 in high with a 6.5 m wide base.

Lowering of the groundwater table in the confined sand aquifer by well pumping is an effective remedial measure for slope stabilisation at this site. A pumping test is required initially to determine the hydraulic characteristics of the aquifer and to provide information about the yield and drawdown of the well. In a well test, water is pumped during a certain time and at a certain rate. The effect of this pumping on the water table is measured in the pumped well and in some piezometers in the vicinity. The hydraulic characteristics of the aquifer is then found by substituting the drawdowns measured in the piezometers, their distance from the pumped well and the well discharge in an appropriate formula.

The well diameter must be selected so as to satisfy two requirements:

Piezometers are required in the well design to measure the effect of the pumping on the water table in the vicinity. A minimum of two piezometers are recommended for this site. The piezometers shall be located at specified locations in the field at a selected distance from the well based on the capacity of the pump used.

Piles installed through an unstable soil mass and into immobile bedrock beneath the failure surface can be used to arrest the movement of the failed soil mass. The use of I passive piles offers a technique that permits the installation of the stabilising system without potentially reducing the stability of the slope (as is required for wall construction).

Rowe and Poulos (1979) concluded that:

The recommended piles for this project shall be 450 mm diameter, bored cast - in - place reinforced concrete piles at a 1.5 m staggered spacing. Pile lengths shall be from 10.0 in to 12.0 in and they shall be installed with at least one third of their length into the weathered bedrock.

A set of three rows of staggered piles shall be installed along Maynard Hill and another installed on Black Mallet road and shall extend uphill to incorporate areas, which show evidence of instability.

Concrete for piling shall be high strength 5 100 psi mix and cubes shall be taken for laboratory testing. The contractor shall provide facilities for testing the integrity of each pile after installation.

The removal of the natural vegetation on slopes has been identified by the author as the single most important misuse of land contributing to the erosion problem in St. Lucia; both surface wash and a greater risk of massive earth movements (landslides, debris flows etc.) being involved. The worsening of the erosion and land slippage problems can thus be closely correlated with the expansion of farming and residences on marginally stable slopes at the expense of forest cover.

Forest cover provides the following advantages to slope stabilisation: