Natural slope problems related to roads in Java, Indonesia W.HEATH, Transport & Road Research Laboratory, Crowthorne, UK B.S.SAROSO, Transport & Road Research Laboratory, Crowthorne, UK ABSTRACT In relation to the high number of landslides which cause serious problems to Indonesia's highway network study was carried out in order to determine common patterns of failure and to establish methods of identifying potential high risk areas of slope instability in Java. It consisted of a regional appraisal and a detailed study of soil and groundwater conditions at a few selected sites in Central and Western Java. The shallow soil slopes on which such failu~-es often occur generally consist of colluvium resting on a relatively impermeable shale. The high weathered soils have a high clay content which, from mnine~ralog' ~es ts on jaivp c -:covcied froii test sites, proved to have an expansive 2:1 lattice structure and clays were generally montmorillonites and vermiculites. Other soil tests showed these soils to have a high activity value of up to 2.48. Four modes of failure and associated groundwater conditions, consisting of mass creep, lateral slides, mud flows and a rotational slumo of the road section, were identified as being characteristic of most failure sites. The report describes the instrumentation .;.esults and observations used in determining these features of slope failure. 1. INTRODUCTION During 1981 a joint study commenced between the Indonesian Institute of Road Engineering and the Transport and Road Research Laboratories Overseas Unit into the problems of slope instability in Indonesia. The general aim of the study was to seek overall patterns of failure on soil slopes, which were connected with highway problems, and to determine the mechanisms of such failure in order to improve the methods of dealing with landslides. In addition it was anticipated that such a study would provide a means of identifying areas of potential risk and that this would assist in the preparation of hazard maps for highway alignment and maintainence purposes. The study concentrated on landslide problems in the most developed area of Indonesia, that of Java, and besides examining widespread occurrences of slope failure a number of specific sites were selected for detailed examination. AS is often common with large landslide studies there were difficulties initially in determining which were the most appropriate investigation and testing procedures for use on the sites. In this respect the limited background information relating to slope stability problems in Indonesia proved to be a severe handicap. Consequently the number of slides being investigated grew and a restricted range of investigation procedures appeared to provide advantages in terms of the resources available. These included; mapping, sub-surface soil sampling and testing, the use of instrumentation to determine the rate an'd depth of failure and an investigacion of groundwater conditions including the detailed study of pore water pressures at one test site. Subsequently a range of groundwater and soil information about different slopes was collected and related to more general observations of slope failure throughout Java. In this respect the empirical records dealing with large scale slope movements, in those other parts of the world where similar slope problems exist, were of value. 2. BACKGROUND TO THE STUDY The first phase of the study consisted of the selection of four representative sites for investigation, and included the preliminary mapping of these sites and limited sub-surface exploration. Reference to this stage has been made by Saroso B.S. et al (1983). It was planned that the second stage include a more detailed investigation of the sites including the determination of the depth and rate of failure and the groundwater conditions on the slopes.. It was also proposed that the shear strength of slope materials should be determined to some degree of precision. Prior to the start of the second stage in 1983 fifteen piezometers were installed at one site in four groups along the slope. The selection of piezometers and their distribution on the slope were determined as part of a joint study involving Bristol University. During the second stage it was recognised 259 that instrumentation results and observations were more relevant from sites that were failing rather than sites with a past history of failure. This meant the rejection of one previously selected site. Additional sites, where continuous slope movement caused problems to roads, were selected including one at the Puncak, km 88, on the main Jakarta to Bandung road. Table I show~s the slides which were instrumented and observed in detail during the second stage of the investigation. Table I. Slopes included in a detailed study SITE ROOD LOCATION MORPHOLOGY OF SLOPE ABOVE ROAD UNSTABLE PART OF SLOPE Scale 1~~~~~0. ~Mad slide (RD a 2Nm) Lacation 3 Locaton 1 tnpip L enradp` Ra~in gauge " Cacks Ope standpipe .V Inlieometer -i Lecetian 4 A I ight n~o~e..t Op0 tAdpp Emtns~Oe crackiSg Open orandoipe and manement -Read~~~~~~~~~~~ I) (Km61) BANDUNG-CIREBON STEEP SLOPES OF COLLUVIUM AND SHALE 2) (Kn64) 01 0 VOLCANIC BRECCIA 3) (Km21) BANDUNG-JAITARTA STEEP ANTICLINE OF COLLUVIUM AND SHALE 4) (Kn24) 1 REEF LIMESTONE 5) (Km24) WELLERI-SUKAREJO STEEP SLOPES OF VOLCANIC COLLUVIUM AND SHALE BRECCIA AND SEDIMENTS 6) (Km88) JAKARTA-BANDONG VOLCANIC BRECCIA AND ASH OLD SLIDE DEBRIS 7) (Krn54) BANDUNG-CIAMIS ANDESITE BRECCIA VOLCANIC COLLUVIUM 8) (Km46) BANJAR-WANGON VOLCANIC BRECCIA VOLCANIC COLLUVIUM N) KRAPYAIT SEMARANG VOLCANIC ASH/SHALE MAINLY SLIDE DEBRIS SEMARANG TOLL RD. OLD SLIDE MOSS Much of the instrumentation and investigation took place at Site 1. Tomo, km 61, and Site 4. Citatah, km 24, two sites close to the Indonesian road research laboratories in Bandung. At Tomo highly porous deposits, composed of volcanic breccia which is generally old lahar, fceo gioutsdwater into colluvium which overlies folded beds of shale belonging to the Subang formation. Citatah is different only in that it is a section of a reef limestone anticline which feeds groundwater into a similar sequence of colluvium and shale. At Tomo the road failed at four locations, within the sector from km 61.7 to 67, during the course of the study. Failures at the instrumented part of the site, km 61, consisted of a 100 m long translational slide, mud slides and continuous periods of slow earch deformation and creep. Within the old lahar deposits above the road failure only occurs when the lower slope becomes oversteepened. No such failures occurred during the period of the study. Figure 1 provides information about this site with details of the main failures and locations of instrumentation. 2.1 Geology and geomorphology: The island of Java forms part of an active volcanic mountain chain, the Sunda Arc, where diastrophism has resulted in uplifted and folded sequences Of Tertiary and more recent sediments to produce a physically rugged and complex landscape. This is modified by an extensive alluvial coastal plain in the north and a series of old alluvial lake basins in the central region. The main geotectonic activity has been in the southern region of Java where the terrain consists of extremely complex morphostructural units within an uplifted sedimentary and frequently active volcanic mountain system. Fig. 1: Details of the slide at site 1; Tomo The geomorphological features of many slopes in Java are similar in that they consist of folded marls or weathered volcanic deposits overlain by sequences of colluvium derived from volcanic breccia and ash. Massive deposits of these porous breccia and ash materials occur on the steeper sections of slopes to form huge collection areas of groundwater. The contribution of extensive groundwater retention zones, reservoirs, to slope failures has been described by Denness B (1973) in relation to failures in Colombia. In Java the lower areas of such slopes are invariably covered with extensive bodies of weak colluvium, an extremely variable material, conristling of bould~ers.. cobbles and gravel in a sand/silt and clay matrix. Such conditions account for a considerable proportion of the total road network area which includes 3,200 km of main and 9,000 km of secondary roads servicing a population of 90 million. Figure 2 shows the percentage of the road network for West Java within areas of landslide risk. Fig: 2. Road-sector map of slope hazards in West Java 2.2 Characteristics of slope failure: Within the lower slope regions failure generally follows a predictable pattern and is related to both high rainfall and the excess 260 groundwater released from highly porous deposits on the upper slopes. In terms of published references, Brand E.W. (1984), has outlined the extent of information for the whole of South-east Asia. The only published reference dealing extensively with slope failures in Indnnesia is that by Wesley L.D (1977) and relates to problems in homogeneous soils on steep volcanic residual-soil slopes. Within the typical soils of Indonesia, derived from pyroclastic materials and marl sediments, Mohir E.C.J (1944) has commented on the high shrinkage, to approximately half the wet volume of these materials, and relates this to montmorillonite clays. Such clays, within the phreatic zone of groundwater, tend to form in alkali soil conditions which possibly indicates the influences the high pH calcium carbonate marls have on the weathering of volcanic materials. Rouse W.C et.al (1986) have published results of a soil study in Dominica, West Indies, in environmental conditions similar to those in Indonesia. In particular the reported distribution of amorphous clay minerals at the higher elevations and 2:1 lattice clays, including smectite, on the lower footslopes, where high levels of groundwater accelerate the weathering process, can also be identified in Indonesia but in much less distinct patterns. 2 .3 Rainfall related to slope failure. The climate of much of Indonesia is quite seasonal being influenced by the equatorial tropical convergence and characterised by a prolonged wet season from September to May. Rainfall records have been collected over a period of sixty years, Berlage Jr.H.P (1949), and indicate that average yearly rainfall varies between 1,500 and 5,000 millimetre. Table II shows the general conditions of rainfall in Java. Table II. Features of rainfall in Java (Berlage) GENERAL ANNUAL INFIL- EVAPO- PERIOD OF RAINFALL TRATION TRANSPIRATION WET=SEASON WEST JAVA (NORTH) 1,708mm 600mm 8OOmm U MONTHS HOST JAVA (CENTRAL) 1,066mm 1.360mm 710mm U MONTHS WEST JAVA (SOUTH) 3,454mm 2,820mm 410mm 10 MONTHS EAST JAVA (NORTH) 1,740.m 830mm 1,340mm 7 MONTHS In Java the principal influence on the amount of rainfall is the ground elevation and the period of the tropical convergence. Within the lower slope area it is between 1500 mm and 3,000 mm per annum. From the records of Berlage, peak rainfall generally occurs in January and then again between the months of March and May. This corresponds to the periods when the majority of landslides are reported. In terms of the effects of rainfall infiltration, and longer term groundwater conditions, such failures appear to have two distinct components. General observations, Newspaper reports and inclinometer results support the view that most slides occur during or immediately after a period of exceptionally heavy rain. Both Brand E.W (1984), in terms of Hong Kong slides, and van Bemmelen R.W. :(1949), in terms of lahar slides in Indonesia, have reported upon precipitation threshold levels of 70mm within an hour or less to initiate serious landsliding. Levels of rainfall exceeding these values are not uncommon in Java. In contrast vast quantities of groundwater are retained within the highly porous upper slope deposits of breccia and ash throughout the wet season. This is dissipated onto the lower slope deposits maintaining a saturated condition and promoting creep failure. It is this, the least spectacular and difficult to observe aspect of slope failure, which is the fundamental factor of all slope instability affecting roads in Java. 3. GROUNDWATER AND INSTRUMENTATION A difficult aspect of the study was the measure- ment of groundwater pore pressure and permeabilities within the restricted horizon of unstable colluvium. At the main instrumented slope, site one, groundwater levels measured from open standpipes at five positions on the slope were collected for a period of two years, see Figure 3. These indicated a high build up in the level of groundwater at the start of each wet season with the exception ~of an area of slope above the road. The most notable variation in groundwater level occurs at the mid-slope positions where an increase of more than 5 metres in the phreatic level was recorded. Similar patterns of groundwater build-up were also observed at sites three and five where open standpipe levels were also monitored. uppe, slope H Mid slope2 Depth (m)l 23 2 Low~er slope 14 basin 2 1 TOMO mase, table levels 2~ ~ 3 1082 1984 .9 200 I i i I 01 111 11. .a i .I L I ~ k I ~ ,~ L 4[ILE~I ~ k A L p 1983 1984 F'ig. 3: Standpipe groundwater levels 3.1 Groundwater flow and pore pressure: Groundwater flow conditions and the build up of pore pressure in the colluvium are relevant to slope failure investigations particularly in the design of drainage as a means of controlling movement. In this respect the model developed by Whipkey J.F and Kirkby m.J (1978), of the build up of a saturation front appears particularly relevant to slopes in Java. it presupposes that subsurface flow conditions generally require an impeding layer or a progressive decrease in permeability with depth before any appreciable flow occurs. A horizon of weathered shale meets such conditions on the majority of unstable slopes in Java. Figure 4 shows what are essentially the conditions during a period of steady long- duration rainfall. AS the precipitation continues a zone of saturation within the impeding layer gradually increases and 261 consequently reduces the diffusing gradient or rate of percolation into the layer. This causes a saturated layer to back-up within the upper more permeable colluvial zone and progressively extend upslope. However it does not extend uniformly because of differences in hydraulic conductivity. 1st hoin Steady rainf.ll 2.d hoio1 Arrows show relative Proportion of sub-surface flow Fig 4: Slope saturation model. (Whipkey et.al) The characteristic variations in patterns of instability on a given slope in Java may perhaps be explained in terms of the spatial distribu- tion of this pattern of saturation across a slope, perpendicular to the forward movement of the saturation advance. This is likely to vary considerably, over periods of years, and may be influenced significantly by disturbances to the slopes groundwater such as deforestation and land development. Such models, in terms of Java's slope failure problems, appear to be extremely relevant but are particualrly complex to develop and prove. 3.2 Piezometers: The piezometers used to determine ground- water pore pressure were the standard coarse ceramic filter (250mm x 50mm) Casagrande type, Casagrande A. (1949), fitted to 19 millimetre PVC open standpipes. Installation was in the prescribed manner with bentonite end seals and a 200 millimetre filter of coarse sand. It provided a hydrostatic piezometric response, to reach an 90% equalised hydraulic pressure, of approximately 10 hours. This was a serious disadvantage resulting in the loss of data relating to the transient groundwater response to storm events. 3.2.1 Recording piezometric levels: In general the instrumentation of unstable slopes is made difficult by the inaccessibility of the sites and therefore methods of recording the data from instruments are necessary. in Java a recently developed acoustic method was used to measure and record the level of piezometric water in the standpipes. For depths up to l0in it was claimed to measure water levels to an accuracy of 5 mm under constant conditions. However assumptions made in the design of this technique, and referred to by Anderson M.G and Kneale P.E (1987), regarding the type of standpipe and influence of temperature led to practical difficulties and serious errors. Fortunately additional pore pressure data was obtained by taking manual readings at three day intervals and this proved to be more reliable. 3.3 Dye tracer: Investigations were made using the dye fluoresceine LT, a yellow dye with the colour index; Acid yellow 73, constitution reference 45350 and a chemical base of the sodium salt of hydro-o-carboxy phenyl fluorine. The purpose was to determine permeabilities and groundwater flow paths across the slides. 4. SOIL PROPERTIES 4.1 Clay mineralogy: X-ray diffraction examination was used to deter- mine the clay mineralogy of soils on a number of failing slopes in Java. For each test three samples were prepared, one randomly orientated, and the second and third with the minerals orientated and glycolated and orientated. These allow the principal clay phases and the presence of swelling sheet silicates to be identified. The tests indicated relatively low levels of amorphous minerals and little Halloysite as being a characteristic of these lower slope materials. Table 3 shows the mineralogy of samples which have been examined. It also includes details of the shale at two sites. Table III. Mineralogy of soil samples from colluvium within the slope failure zone. COLLOVIAM SITE: SOIL TYPE MAXIMUM ANOINT SECONDARY ASSOCIATED CLAY MINERAL CLAY QUARTZ % CATION lC2 BRECCIA/SHALE SMECTITE 35% KAOLINITE 15% 30% Ca 3&4 L/STONE/SHALE VERMICULITE 40% MUSCOVITE 5% 20%-40% Na 6 VOLCANIC ASH GIBUSITE 20% ATTAPULGITE 10% 4 0% M9 SMECTITTE 5% 7 ASH/SHALE SMECOITE 60%, KAOLINITE 10% 1% 1TO4 SHALE MUSCOVITE 40% MIXED 10% 50% 7 SHALE KAULINITE 54% SMECTITE 34% 6% Similar clay mineralogies have been reported in East Java, Subardja and Buurman P (1980). Gibbsite was related to mid slope weathering profiles and the smectites to basin positions. The landslide slopes on which smectite clays are prevalent can be readily identified by the high rate of shrinkage and cracking of the soil that occurs during the dry season. Vermiculite was also found on shallow lower slope positions, in East Java, and defined as a pre-smectite stage of mineralogy. Unlike smectite clays which are widely associated with low values of shear strength and consequently slope failures, the vermiculites have received considerably less attention. However the clay structure, whilst not as active, is similar to montmorillonite in having a weak expanding 2:1 lattice and a high cation exchange capacity. Gibbsite was found to be associated with old volcanic ash soils at one landslide site. The presence of gibbsite in such soils has been attributed to the severe tropical weathering of igneous basic rocks with the rapid removal of silicates, Harrison J.B3 (1934). Collapse, by saturation, and a weakening of cementation has been reported by Brink A.B.A and Kantley B.A (1961) and Vargus M (1973) for lateritic soils and Foss I (1973) for andosols. However there is still no direct test evidence of soil collapse to support field observations in Java. 262 i 4.2 Soil tests: Soil samples were collected from a range of depths, up to 20 metres, including those positively correlated with~slickenslides or planes of failure determined from inclinometers. Table 4 shows a range of soil test results. Table IV. Soil test data from slopes SOILTYPE DEPTH CLAY % MINERALOGY FEATURE OF MAIN CLAY w% LL PI lkN/in An~gle **JAVA: (Main Test Site. 'Tome' Site 1) Colluvium 2m Gravelly 22 Smectoid 30% 45 81 46 57 8 3m Clayey 47 " 354 40 83 49 84 4 5m W/shale 43 30 81 48 45 14 Shale llm 20 Muscovite 40% 28 110 75 58 23 **JAVA: (Other Test Site. 'Citatah' Site 3) Colluviuet 2m Gravelly 48 Vermic- 30% 30 50 28 22 14 5m Gravelly 56 alite 40% 18 46 27 Site 4 Colluvium Im 31 Vermic 71 40 24 10 4.5m 31 ulite 70 27 10.5 13 8.Sm 32 51 29 28 4. lOfin 25 "49'27 33 4.5 13.5m 23 "44 27 29 7.5 It is noticeable from these results that the smectites are not always associated with the highest values for either the liquid limit or plasticity Index. However in terms of activity, Figure 5 such soils have a high average value of 2.48, which is within the range of 1-7 for a calcium-cation associated montmorillonite. In this respect the soil test and x-ray diffiacticmi re.~ults support the conclusion of a soil with a high smectite clay content. The less active clays including the vermiculite samples have an average activity of 0 .87 . 9Oso 70 60s0 40 30 20TO JA VA SLOPE Expansive clay;.ctivity 2.48 Londor, cay aeivity 0.95 p Java slopes; activity 0.87 (Ve,micuhites) TO 20 30 40 50 60 70 Clay f-taoni Fig 5: Activity value of soils Laboratory soil grading tests were considered to be unrepresentative in terms of the clay- size proportion of the colluvium from failing slopes. From tests performed on a much greater sample than that shown in Table 4 an average clay-size content of 26% was determined. This had a standard deviation of 12% implying that few samples had clay contents greater than 40%. Other evidence including x-ray mineralogy examination, the high activity values of the soil and its physical properties all indicate a considerably higher amount of clay. In this respect the fines in the soils appear to form clumped particles, larger than 0.002 mm, which resist breakdown during normal grading tests. Terzaghi K (1958) noted that Javanese volcanic residual soils had characteristics very similar to Kenya soils in terms of the clays forming aggregated particles that distorted the value of the soils liquid limit and were difficult to break down. Whilst it is difficult to draw parallels between the colluviums and the residual soils this ability of such soils to behave similar to soils with a large granular fraction, and consequently high cohesion, but readily break down under specific conditions should be considered in any future research into slope stability problems in Java. The values obtained from pre-consolidated drained triaxial tests included low angles of shear resistance within the range 51 to 150, and relatively high values of soil cohesion. They may reflect an orientat~ion of the soil structure from the effects of the slow creep. The angle of internal friction is clearly related to the soils cohesion, see Figure 6. This shows the intercept between shear angle and cohesion for what are considered to be samples of colluvial near planes of failure. Figure 6 shows the increased range of plasticity for samples recovered from shallow depths and this is reflected in the range of soil moisture values. A correlation also exists between the soils liquidity index and horizons of failure within the soil profile. She., ngle Plasticity i0de. Shea, angle (per ceet) a 100 Tao 10 30 50 70 80 '(Srnectite) 60 ~~~~~8 z40 * 1216 20 0~~~~~ - 2 20 Vemiculile 24 Liquidity mnde. 0.00.4 0.2 0 0.2 0.4 0.8 02[ [1-Zones of failure Fig 6: Range of soil properties 5. GROUNDWATER INFLUENCE ON SLOPE FAILURE Rainfall conditions at a number of slopes are shown in Table 5. The information suggests that the total amount of annual rainfall had some influence on the number of slides. Also it was established that at least 50% of reported slides occurred during or after periods of heavy rain when daily rates exceeded 100mm. The rate of infiltration of such rain into the slopes is also relevant to slope stability. This together with groundwater flow conditions were determined from dye tracer tests. Such tests at four sites, within the upper slope area, provided samples of the dye, at dilution levels of between 0.3 and 2.3 parts-per-million recovered in bore holes placed up to 200 metres down slope. Intrinsic permeability, estimated from the dye tracer recovery, was as high as 5,000 in/hour in some tests indicating that fissure and natural pipe flow was a 263 1c .rt.t1 significant factor in groundwater movement. The distribution of dye concentration on an unstable slope is shown in Figure 7a and provides a basis for estimating groundwater flow with depth. The relative position of the spring line agrees with observations made at the surface. In hydro-morphological terms it is also significant that, as previously identified by Deere D.U and Patton F.D (1971), failures are generally within such groundwater discharge zones. Table V. Features of rainfall at specific sites see Figure 1. Fluctuations in pore pressure over a period of one year are shown in Figure 8. Within the mid-slope zone rapid transients in piezometric levels, at the beginning of the wet season, (about November, see Figure 8) are noticed and attributed to the priming of permeability in the soils. An explanation by Bouwer H (1978), is that entrapped air in the soil mass blocks conductivity paths and causes permeability to be less than when the soil is fully saturated. Consequently an initial build up of pressure or head of water is needed to overcome this. SPECIFIC LANDSLIDE SITES ANNUALRAINFALL SEASONALRAINFALL SITE ONE 1983. 2,329mm 1982/A3. 1984. 2,753mm 1983/84.1984/185. SITE THREE 1983. 2,390mm 1902/03. 1984. 2,340mm 1983/84.1984/AS. SIRE FIVE 1982. 1,643mm 1982/83. 1983. 1,862mm 1983/64. DAILY RAINFALL RATESDAYS EXCEEDING SOmm/24 HOURS SITE ONE THIRTEEN SITE THREE TWENTY FOUR 2, 6 44mm1,'81 Smm 3 , 222cmi2., 353mm2 , 2 90mmn2 ,01 7mm I , 388ec2 ,51 3mm DAYS EXCEEDING 10Omm/24 HOURS SEVE N FIVYE NUMBER OF LANDSL IDES NANETWONINENINETHREENINENINEsix PEAK RAINFALLPERIODS246mm/2 DAYS 168mm/1 DAY The recovery of dye from an apparent slip plane was also achieved at one site, Figure 7b, indicating the preferential looveinent of groundwater along failure planes. The'soil samples, of what was a coarse gravelly material, showed a high concentration of the tracer, Subsequently an inclinometer in the same borehole confirmed movement at the depth the tracer had been recovered. -E0 N 'Aample recovery (P1PM, Poaes per eillioe( Ecoespim (A) ;Distrrboti.e of flow S h ole m0 o l e o - 7 S ., '..~~~~~~ mla d e p t 'h' o Emo ple (81, Recovey (ro slidiog plane Borehole 2 Fig 7: Results from dye tracer tests 5.1 Pore pressure and groundwater at Tomo site: The hydrological pattern affecting slope behaviour has been determined from three instrumented sites on a failing slope at Tomo, upper slope 2 '71 1984K Mid slope 1.4E 1.0t; 1.8 -- K1.2.1 4 ~T.~05X 8.5S .1984 .. Lowe, slop,_ 3.5 -- ' - 11 -- ' Z 9 0 8 .0,75 ` 0 ,25 .. ...._ 1984 Fig 8: Piezometer levels at one site 5.1.1 Upper stable slope area: This zone is above the road on an area of steep volcanic breccia which overlies shale at a depth of 8.75m. within this area there were considerable fluctuations in pore water pressure at depths up to seven metres. Infilit-ration peaks were recorded on at least three occasions during storms, when open standpipe piezometer water reached to almost the maximum levels indicating pore water pressures in excess of 15 kN/in 2 at a depth of 1.7 metres, 50 kN/mn 2 at 5.5 metres, and 60 kN/m' at 7.15 metres. In the shale there is little response and its riot normally associated with the infiltration wetting front. 5.1.2 Mid-slope area: Within this area of the slope the phreatic level generally remained high for most of the year and piezomretric levels rise above the tops of the standpipes frequently during the wet season. Figure 9 the left hand side, shows the variation in piezometric pressure and rainfall together with transient increases in groundwater pore pressure for three piezomreters. The right hand side of Figure 9 illustrates increases in pore pressure and a five day pattern of rainfall. Piezometers in the shale showed little response in what is obviously a relatively impervious horizon. The uppermost 300 mm of -colluvium, within the A horizon of the soil mass is also relatively impervious as is evident from the flooded paddy conditions on this and similar slopes. Hydraulic conductivity is therefore mainly confined to a narrow horizon which extends to a depth of 5 metres. In this respect the limited depth of the soil profile is probably more relevant than the basin morphometry in determining groundwater movement on this part of the slope. Soil void ratios are in the range of 0.8 and 1.2 and therefore with natural moisture 264 0 Scale 100. Sample emtry Sprn~g liep?---.--- ---- PPM