Rainfall Effects on Pore Pressure Changes in a Coastal Slope of the Serra do Mar in Santa Catarina

This research aims to describe how rainfall can cause changes in the piezometric pore pressure and soil matric suction in a densely instrumented slope by the South BR-101 (Brazilian Numbered Highway), in the area of Morro do Boi, in the State of Santa Catarina, South Region of Brazil. The slope presented a history of instability movements instigated by intense rainfall, with debris accumulation on the highway and traffic interruption. The analyzed data are measured by six vibrating wire piezometers and eight electrical tensiometers attached to a datalogger, two conventional slope inclinometers and a rain gauge with an internal datalogger. A total of 2,552 readings corresponding to the vibrating wires and electrical resistance instruments, 29 inclinometers records and 7,143 rainfall records were collected over the first ten months of slope monitoring. The analysis results demonstrated that during the monitoring period there were no heavy rains. Three monitoring periods were identified by the frequency and intensity of rainfalls. The soil pore pressure monitoring instruments showed significant variations in the high frequency period and low intensity rainfall, and little variation in low frequency period and high intensity rainfall, which demonstrates greater runoff and little infiltration during the occurrence of more significant rainfall.


Introduction
The Serra do Mar is a mountain range which constitutes the most prominent orographic feature of the Atlantic edge of the South American continent, with approximately 1,000 km length, extending from the state of Rio de Janeiro to the state of Santa Catarina (Almeida & Carneiro, 1998).In these accentuated-relief regions, there are important Brazilian highways which are exposed to risks associated with mass movements, a consequence of the natural and anthropic conditioning (Montoya, 2013).
Although literature indicates that mass movements can be a result of many different factors, such as climatological and hydrological processes, geological characteristics, topography, vegetation, anthropogenic actions (garbage deposits, deforestation, changes in drainage or poor surface and deep drainage, cutting and embankment with expressive angles, overloading, design and change the route of highway) or of all these factors combined (Fernandes et al., 2001;Rahardjo et al., 2008;Zuquette et al., 2013;Carvalho et al., 2015), the role of rain in the events that cause slope instability is widely known (Brand, 1984;Brand et al., 1984;Lim et al., 1996;Rahardjo et al., 2001;Chen & Lee, 2004;Rahardjo et al., 2008;Zuquette et al., 2013).
The effects of rainfall on slope stability are a theme of interest as parameters and warning systems can be generated from rainfall data to prevent human and material losses (Montoya, 2013;Bandeira & Coutinho, 2015).The infiltration of rainfall into the ground develops positive pore pressures by raising the water table and reducing suction levels (Chen & Lee, 2004;Rahardjo et al., 2001Rahardjo et al., , 2008Rahardjo et al., , 2016;;Gerscovich et al., 2011;Advincula, 2016), and also generates a preferential flow through the fractures of the bedrock.Therefore, the infiltration resulting from rainfall and the subsequent variations in pore pressure determine the safety level of a slope (Gerscovich et al., 2011;Montoya, 2013;Carvalho et al., 2015).
This article aims to describe, analyze and discuss monitoring data from a research study on a highway slope, with a history of mass movements prompted by significant rainfalls, during the period from May/2012 to March/2013 (González, 2013).
According to Sestrem (2012), the slope had a history of instability characterized by movement and the consequent accumulation of debris on the highway and traffic disruption.An occurrence of mass movement -with the breakdown of this rocky slope and the removal of the soil top layer, causing soil and rock blocks to fall down on the highway lane -was recorded during the rainfall that occurred between November 20 th and 24 th , 2008.Besides this slope, several other highway points had ruptures in slopes resulting from the intense rainfall, which took place in the state of Santa Catarina during this period (CIRAM, 2016).These rainfalls fell on areas such as the Greater Florianópolis, Vale do Itajaí and North Coast of the state of Santa Catarina (Zuquette et al., 2013).Regions such as Blumenau and Joinville experienced around 1000 mm of rain in that month.The region of Vale do Itajaí has been subject to a total rainfall of approximately 600 mm between November 21 st and 24 th , 2008, according to CIRAM (2016).
The slope under study was stabilized (with nails, metallic mesh and a cap beam of root piles), in order to minimize future inconvenience to road users, after the catastrophic event of 2008.The need to better understand the mechanisms that may trigger accidents motivated the investigation and instrumentation of the slope (Fig. 2) to monitor the stabilization solution adopted (Kormann et al., 2016).

Lithological and geological aspects
The study area is characterized by the presence of two main lithological types: Morro do Boi's Migmatites and Nova Trento's Intrusive Suite granites.The suite is represented by an intrusive body in Morro do Boi's Migmatites, aligned in the NE-SW direction (CPRM, 2014).
According to CPRM (2014), the Morro do Boi's Migmatites extends in the northeast -southwest (NE-SW) direction, in a strip ranging in width from 1.0 to 1.5 km and to the south and east of the city of Camboriú.Its structure is mainly stromatic, often folded, where dark gray metabasic rock xenoliths are common, ranging from homogeneous bodies of massive aspect to finely banded.A major fracture system occurs in the body of Morro do Boi's Migmatites generated by NE-SW and NW-SE direction shearing and by sub-horizontal fractures, having as main effect the subdivision of the massif in blocks, which reduces its mechanical resistance.Additionally, due to the continuity and interconnection of fractures, the water easily flows within the massif.In complement to these conditions, there is a layer of silty sand soil on the slope.

Geotechnical characterization
The soil ranges from mature residual to young in depth, in areas of the slope which were not transported.However, in part of the monitored area, the top soil was identified as colluvial.Through field tests performed on the slope, including three holes of SPT (Standard Penetration Test) and five SM (Standard Penetration Test and rotary drilling), it was observed a superficial layer of silty sand soil with a thickness of around 3.0 m, complemented in some regions by the presence of blocks of rock.There is also a highly weathered layer of rock with a thickness of about 3.0 m over a layer of moderately weathered rock found at 6.0 m depth and with a thickness of around 3.0 m, which overlies the Migmatite, found from approximately 9.0 m depth.The depths of the field investigations were approximately of 12.38 m for the SM-01, 13.00 m for SM-02, 8.20 m for SM-03, 9.25 m for SM-04 and 10.70 m for SM-05.
The colluvium superficial soil presented N SPT from approximately 9 to 40 blows, increasing in depth along the drilling hole, characteristics of a medium compact to com-pact material.Below that layer, refusal was achieved, being false results due to the presence of rock blocks at some points.Through the SM field investigations, high percentages of RQD (Rock Quality Designation) were obtained from the samples with continuous recovery, characterizing an excellent quality of the rocky massif (RQD of 90% to 100%).As for the moderately weathered rock layer, the mean RQD values obtained were 60%, of reasonable quality.
During the geotechnical surveys it was also possible to observe the water level position, which was equal to 5.35 m for SM-01, 6.12 m for SM-02, 3.50 m for SM-03, 4.60 for SM-04 and 4.60 m for SM-05.Based on such water level depths, the quotas for the installation of pore pressure monitoring instruments were determined, specifically with respect to the deepest piezometers.A geological-geotechnical profile of the slope, which resulted from the compilation of the geotechnical investigation carried out in it, is presented in Fig. 3, including geotechnical monitoring instrumentation, such as inclinometers (INCL), piezometers (PIEZ) and tensiometers (TENS).
According to Massad (2003), in regions of humid tropical climate, the lithotypes which correspond to gneiss metamorphic rocks or with banded appearance give rise to predominantly silty and micaceous soil.For this purpose, soil characterization procedures were carried out to confirm that the weathered Migmatites found in the region result in this type of soil.
To evaluate and characterize the superficial soil properties in the monitored area of the slope, laboratory tests were carried out on four deformed samples collected from non-deformed blocks, in the top slope layer (Lazarim, 2012).Among the performed tests are: soil density -employing the procedure described by DNER-ME Standard 093 (1994) -, Atterberg limits -following the procedures described in the standards NBR 6459 (liquid limit) and NBR 7180 (plastic limit) (ABNT, 2016 b,c) -and particle size analysis of the material -according to the procedure described in NBR 7181 (ABNT, 2016 a).
The laboratory tests classified the top soil as silty sand, with particle density of approximately 2.66 g/cm 3 , average liquid limit of 32%, average plastic limit of 27% and average plasticity index of 5% (Table 1).With respect to particle size analysis (Fig. 4), the average percentages obtained were 4% clay, 27% silt, 61% sand and 8% gravel (Table 2).Direct shear strength tests for samples of colluvium soil collected at depths of 0.25 m to 1.27 m, presented mean friction angle of 34°and mean cohesive intercept of 2 kPa.The average specific natural weight for this material was equal to 16.20 kN/m 3 (Lazarim, 2012;Gonzalez, 2013).In addition, in situ permeability tests were executed at the colluvium surface soil, with values ranging between 4.47 x 10 -7 and 1.71 x 10 -6 m/s, in agreement with the granulometric analysis, according to Pretto (2014).
The slope geotechnical instrumentation aimed to observe the parameter changes such as positive pore pressure and matric suction, in order to check the oscillations of the water table and piezometric level, as well as the occurrence of negative pressure at the top soil.Therefore, six (06) vibrating wire piezometers were installed and eight (08) electrical resistance tensiometers were distributed in islands (Sestrem, 2012) or groups (Sestrem et al., 2015) connected to a datalogger for storing the resulting data.Additionally, two casings with inclinometers were installed to monitor possible horizontal movements of the soil mass as a result of the changes in the above parameters.A rain gauge was also installed to register the intensity of local rainfall and thus relate the monitored parameter variations with the recorded rainfall.A sketch of the instruments installed in the slope is presented in Fig. 5

Piezometers
The piezometers are installed in two islands consisting of three instruments each; the upper island (Group 1) with depths of 8.60 m (PZE-04), 7.20 m (PZE-05) and 3.70 m (PZE-06) and the intermediate island (Group 2) with depths of 8.65 m (PZE-01), 6.40 m (PZE-02) and 3.90 m (PZE-03).It is important to note that these instruments were placed at an equivalent depth between the islands, with the deepest ones installed in the interface of rock and weathered rock layer, and the most superficial ones in the highly weathered rock layer.
For the determination of positive pore pressures, vibrating wire standard piezometers were used (Fig. 6a).Among the sensors available, it was selected the Geokon model 4500S (reading capacity ranging from -100 kPa to 350 kPa).These sensors present readings as frequency, the square of the vibration frequency being proportional to the pressure applied to the steel diaphragm (membrane), according to GEOKON (2012).
Prior to installation, a saturation procedure was necessary to prevent the presence of air bubbles inside the instrument.This procedure initially consisted of removing the porous tip and subjecting it to boiling.Then it was transferred to a larger vessel without the contact with the water being lost, so that it was repositioned in the body of the piezometer.The sensor was then stored and sealed.As a result, each instrument was read zero with the tip positioned at the bottom of the bottle with water.
The stages of installation of the piezometers began with the hole drilling.Then, the piezometer was positioned at the reading depth of interest.A bulb of sand (coarse and washed) with a height of 1.00 m was added to then remove the survey coating.Then, a seal with bentonite of 0.50 m thickness was realized, aiming to waterproof the region of the readings.Finally, the hole was filled to the surface.The cable was initially connected to a mobile reader unit for determination of preliminary readings.After that, all the cables were connected to multiplexers, these being finally connected to the datalogger, thus finalizing the automation of the readings.

Tensiometers
As for tensiometers, they were distributed into three islands, all in colluvial soil; at the upper island (Group 1) with depths of 1.00 m (TENS-07) and 2.00 m (TENS-08), in the intermediate island (Group 2) with depths of 0.50 m (TENS-03), 1.00 m (TENS-05), 2.00 m (TENS-06) and 3.00 m (TENS-04), and in the lower island (Group 3) with depths of 1.00 m (TENS-01) and 2.00 m (TENS-02).Installation depths of piezometers and tensiometers followed the water level found at the geotechnical surveys, in order to obtain records of the increases and decreases of the positive and negative pore pressure values, as well as the advancing wetting front through the soil.
For the determination of negative pore pressures, conventional tensiometers were used, model model 2725A from Soil Moisture (Fig. 6b), composed by the following components: porous ceramic cup, plastic tube body and a vacuum meter.Measurement of the negative pressure (vacuum) was automated by means of a transducer coupled to the tensiometer.The instrument reading capacity ranged from 0 kPa to -100 kPa.
Prior to installation, the tensiometers were prepared and assembled in laboratory where initially the porous stones were submitted to a saturation procedure.To this end, they were immersed in a container containing water and subjected to the removal of air in a desiccator with silica and vacuum pump.In parallel to this, the inside of the tensiometer tube was washed with water and detergent.This procedure aimed to remove particles and possible fat spots that might favor the formation of air bubbles and, therefore, alter the suction values read.After saturation of the porous stone tips and cleaning of the interior of the tensiometer tubes, they were fitted according to the desired lengths.All the connection threads between the tube extensions were installed with o-rings to ensure complete sealing of the tensiometer, preventing the entry of air and the formation of bubbles, avoiding the phenomenon of cavitation (expansion of air bubbles), according to Soil Moisture (2011).
After being assembled, the tubes were subjected to a suction process, with the ceramic tip being immersed in a vessel with boiled water and the other end connected to a pump.This procedure allowed the removal of as much trapped air as possible in the wells (Jones et al., 1981apud Marinho, 2005).
Once the tubes were completely filled by water, they were connected to the reservoirs, which were also filled with boiled water.Then the upper end of the reservoir was pressed so as to inject water into the tube to fill it completely and eliminate any remaining bubbles.Once assembled and tested, they were prepared for transportation to the field.In order to avoid the loss of saturation of the porous stones, in addition to possible leaks, they were immersed in water and protected with a plastic bag, according to the recommendations provided by the manufacturer.
For field installation, it was applied a hand drill.Prior to the positioning of the instrument in the drilling, its tip was placed in contact with a mixture of water and previously sieved local soil (#40).This mixture was also used to fill the hole, ensuring the system sealing and avoiding infiltrations into the tensiometer.
It was also necessary to verify the calibration of the analog tensiometer, an accessory supplied by the manufacturer hermetically sealed at sea level.When installed at a higher elevation, as in the case of the present work, the pointer on the gauge display may have a reading other than zero, resulting from a lower atmospheric pressure.
Finally, the portion of the tensiometers positioned above the soil surface was protected with a 100 mm diameter PVC tube filled with soil from the site, trying to avoid possible problems, such as accidental impacts and bending of the tensiometer tube.In addition, all tensiometers received additional concrete-based protection and an external metal shield of 300 mm diameter.

Inclinometers
For the monitoring of horizontal displacements in the slope, two conventional inclinometer tubes were installed (Fig. 6c), anchored in Migmatites, at a depth of 12.38 m (INCL-01) and 13.00 m (INCL-02), placed in the middle and upper islands, respectively.They were installed into the drilling holes of SM-01 and SM-02, respectively.
The installation sequence of each inclinometer started with placing the access tube in a hole with a diameter of 100 mm, with the respective depth of INCL-01 and INCL-02.An aluminum tube with a diameter of 80 mm and four diametrically opposed slots was used to guide the instrument (torpedo) during the readings.The tube was inserted in the hole, maintaining the alignment of the grooves according to the main axes of displacements of the slope, that is, a plane perpendicular and another parallel to the highway.After complete installation of the pipe/tube, the space between it and the walls of the bore was filled with cement grout and bentonite (1:10) upwardly through the injection hose.Finally, a protective box with padlock was installed at the surface, and a concrete base was also executed, in order to prevent any damage caused by work operations and vandalism.

Rain gauge
The rain gauge installed (Fig. 7) with tipper buckets was model TB4/0.2 from Hydrological Services, whose readings are obtained by a datalogger model ML1-FL.This system has a maximum reading intensity of 700 mm/h and a resolution of 0.2 mm, being able to record in its memory the date and time of the occurrence of rain, with a storage capacity of up to 100 thousand events with a resolution of 1 second, according to Hydrological Services (2011).
The chosen pluviograph has its operation based on a tipping system.Whereby, a metal bucket of 200 ± 0.3 mm in diameter accumulates the precipitations and, when its capacity is reached (0.20 mm) tipping occurs.At this point, the data collector system records the date and time of this occurrence.Between the bucket and the measuring system, there is also a metal screen with the purpose of preventing the passage of objects that could obstruct the system (leaves, branches).The data collector has a reading capacity for rains with intensities between 0 and 500 mm/h (lower than that of the datalogger), temperature range from -20 to +70 °C, and accuracy of ± 2% for intensities between 25 and 300 mm/h ± 3% for intensities between 300 and 500 mm/h, according to Hydrological Services (2011).The definition of the position of the pluviograph considered that it should guarantee a representative reading of the pluviometric indices at the site.It was positioned as close to ground level as possible, avoiding sloping terrain.In addition, there was the need to position it in an area protected from strong winds and obstacles.Another problem that could occur was the absorption of rainwater from the soil around the sensor.Thus, to avoid such interference, it was decided to install the instrument at a distance of approximately 1.20 m from the ground.It was chosen to position it within the stabilized area.It should be emphasized that care with vegetation that grows in this location should be taken, thus not only serving the pluviometer, but also making it possible to read the inclinometer installed in the same local area (INC-01).
The installation began by driving a vertical nail at the chosen location (of the same model used for the stabilization solution) so that its tip was approximately 40 cm above the surface.A circular base was then positioned on such bar, leaving it centered in concrete.Finally, the pluviograph was installed on three screws that allowed leveling, by means of the adjustment of the nuts guided by the bubble level contained in the equipment.After the pluviograph was installed, a test operation was performed in which the hopper tip was initially pressed a few times to check if each movement was being logged, and whether the tilt mechanism was operating freely.According to Hydrological Services (2012), the instrument is factory calibrated and the only maintenance procedure required is cleaning, whereby the following items must be checked: trap filter, siphon, bucket interior, upper surface of set screws, fastening screws (which must be lubricated after cleaning) and screens against insects.

Instrumentation data
The slope instrumentation data collected during the first ten months of monitoring -from May 1 st , 2012 until March 1 st , 2013 -were compiled and analyzed.The automatic data collection from the piezometers and tensiometers provided readings every hour and, later, these readings were grouped so that the results were converted into daily average values.The readings of the tipping bucket rain gauge (PLUV-01), located in the intermediate island, were registered every time it reached 0.20 mm of rain, and the data was stored in an independent datalogger.
During the monitoring period, 7143 rain gauge readings, 2552 readings from piezometers and tensiometers and 29 total readings from both inclinometers were collected.
The obtained data were classified as continuous time series data, which can be interpreted with specific technical statistics.This classification was based on the characteristics of the obtained data, which showed a sequence at regular time intervals during a specific period (Latorre & Cardoso, 2001).
The data interpretation was based on several graphical representations of the time series to determine an ascending or descending trend, the influence of time -stationarity -and any discordant observations -outliers -(Gonzalez, 2013).
The time series were compared with the rain events, for example, to establish a relation between the positive and negative variation of pore pressure parameters and the rainfall events that occurred in the analysis location.It is important to observe that for the analysis of time series, the first step is to model the phenomenon to be studied for describing its behavior and thus evaluate which factors influenced its variations and behavior (Latorre & Cardoso, 2001).
For the definition of rain intensity, the classification system by CIRAM (2016) was considered, whereby the intervals for accumulated rainfall per hour (mm/h) were classified and defined, in a general manner.In this classification, the authors considered: drizzle rain (C mFra ) for rainfalls between 0.25 mm/h and 1.00 mm/h; light rain (C Fra ) for rainfalls between 1.00 mm/h and 4.00 mm/h; moderate rain (C Mod ) for rainfalls between 4.00 mm/h and 16.00 mm/h; heavy rain (C Fo ) for rainfalls between 16.00 mm/h and 50.00 mm/h, and violent rain (C mFO ) for rainfalls equal to or greater than 50.00 mm/h.
With respect to the classification of accumulated daily rainfall (mm/day), intervals were determined based on the definition of quantiles, evaluating the history of rainfall by the probability of occurrence (Xavier & Xavier, 1987;Leite et al., 2011;Souza et al., 2012).The analysis was performed with an updated database of the rain gauge installed in the study area, as shown in Table 3.

Results and Discussion
The time series resulting from the monitoring of geotechnical instrumentation were analyzed initially considering the independent variable (rain) and the relationship of this parameter with the variations in positive pore pressure and suction values, according to Lim et al. (1996) e Rahardjo et al. (2008).

Rainfall events
Through the data series, three different rain periods were observed, delimited by the intensity and magnitude of the events.The first period -between the months of May and July, 2012 -was characterized by magnitudes of 196.60 mm, 203.40 mm and 260.00 mm, although with moderate intensities -around 19.40 mm/h, 11.60 mm/h and 14.60 mm/h.It means that rainfall events resulted in prolonged rain over the days, with the most significant ones classified as moderate and heavy rain, according to the classification of CIRAM (2016).
In the second period -between August and November, 2012 -the accumulated rainfalls per month had a lower magnitude, with 49.20 mm for August and 61.80 mm for September, with peaks of 4.00 mm/h and 11.60 mm/h, respectively.These events were considered moderate rain-falls, according to CIRAM (2016).It was observed that October had an outstanding record, in which isolated rains of great intensity and magnitude reached 174.20 mm accumulated rainfall with a peak of 13.40 mm/h.In this month, two significant events were recorded -which influenced the instrumentation readings behavior, hence an increase in the month-accumulated value: they occurred on October 11 th , with an 80.60 mm accumulated rainfall, and on the 22 nd , with 40.40 mm accumulated.
It is important to highlight that -due to failure in the instrument between November 8 th and December 12 th , 2012 -November recorded only 0.40 mm, hence November to the beginning of December were without valid records.
In the third period -between the months of December, 2012 and February, 2013 -the rains had great magnitude with intensity that reached 112.00 mm in 14 days of rainfall recorded for December and with a peak of 29.60 mm/h.For January the record was 109.60 mm in 17 days of rainfall and a peak of 38.00 mm/h.For February the record was 216.60 mm in 22 days, reaching a volume peak of 38.00 mm/h.The precipitation related to these three months was characterized as heavy rainfall, according to CIRAM (2016).
In order to observe in detail the magnitude and behavior of rainfall, without generalizing the monthly accumulated values, it is shown in Fig. 8 the rain distribution throughout the monitoring period based on daily rainfall values.Furthermore, it is presented the water level readings at the slope during the same period of time, where the records were made once a month manually.It can be observed water level variation ranging from 6.98 m to 9.80 m in INCL-01 and 9.05 m to 9.97 m in INCL-02.Unexpectedly, water level depths were deeper than the values determined during the geotechnical survey, as equal to 5.35 m for SM-01 (INCL-01) and 6.12 m for SM-02 (INCL-02).
In this study, 263 out of 304 days had records during the monitoring period.These rainfall events were classified according to the quantiles technique (Xavier & Xavier, 1987;Leite et al., 2011;Souza et al., 2012), with the results shown in Table 4. González et al. the others from Drizzle (C mFra ) to Strongest Rain (C mFo ), ranging from 3.04% to 16.73%.
Surface runoff basically occurs when the rainfall intensity overcomes the infiltration capacity.Under this concept, evaporation and evapotranspiration during the rain are negligible.Considering the in situ permeability test, with values between 4.47 x 10 -7 and 1.71 x 10 -6 m/s for the surface soil layer, it is possible to establish the equivalence to 1.61 mm/h to 6.16 mm/h of rainfall intensity.In this range and by the rainfall classification made by CIRAM (2016), moderated rainfall will have surface runoff.

Piezometers
As for piezometers installed at the intermediate island, it was observed a behavior of significant variations during the first (May to July) and second (August to November) monitoring periods, until the readings stabilization in the third period (December to February) as shown in Fig. 9. On the other hand, the piezometers installed at the upper island did not show significant variations in readings behavior, as shown in Fig. 10.The daily accumulated rainfall and the total accumulated rainfall during the monitoring period can be observed and compared with the piezo-  meter records (Figs. 9 and 10).In general, the piezometer presented good agreement with the daily accumulated rainfall, showing an increase in positive pore pressure as the rainfall occurs and a decrease during the period with less rain.
The instruments placed at greater depths showed the greatest pore pressure variations, with peaks up to 7.00 kPa, minimum reading equal to -3.80 kPa and an average reading of 0.17 kPa for the piezometer PZE-01 (installed at a depth of 8.65 m), as can be observed in Fig. 9.The maximum reading was equal to 2.90 kPa for the piezometer PZE-04 (installed at a depth of 8.60 m), while the minimum reading was equal to -3.10 kPa with an average reading of -0.53 kPa, as can be observed in Fig. 10.These high rise behaviors were justified by the water table level increase during continuous rainfall periods which occurred from July 15 th to 31 st , 2012.As the water level depth recordings were deeper than the values determined during the geotechnical survey, the deeper instruments were reading intentionally the capillary fringe.Furthermore, these instruments could capture an eventual elevation of the water table following the occurrence of a very intense rainfall as happened in 2008.
The piezometers installed at an intermediate depth, such as PZE-02 (6.40 m) and PZE-05 (7.20 m), as well as the most superficial ones, PZE-03 (3.90 m) and PZE-06 (3.70 m), presented lower readings variation.The records of such instruments varied around zero, even though they were located in sites with different slopes and elevations, demonstrating that the wetting front was parallel to the slope and that the instruments were located above the water level.Conversely, the values corresponding to PZE-02 pre-sented reduced reading intervals, characterizing abnormal behavior (Fig. 9).

Tensiometers
The tensiometers showed a variation trend similar for them all during the first monitoring period (May to July) until the beginning of August, with values between 0 to 10 kPa.After the initial phase, it was observed a change in the data provided by the instrument closer to the surface.More specifically, there was an increase in the suction values for the TENS-03, located at a 0.50 m depth at the intermediate island, with minimum and maximum readings equal to 3.20 kPa and 77.97 kPa, respectively (Fig. 11).As the slope conditions changed over time due to the vegetation growth, the instrument installed closer to the surface became more susceptible to reading changes after rain events, which reflected on the quick variation on the suction records, as can be noticed by the difference between slope vegetation covering in April 2012 and March 2013 from Fig. 12.
The most significant rainfall event for the decrease of suction in the TENS-03 (0.50 m depth) was February 7 th to 11 th , 2013.In this period, the suction measured decreased 66.6 kPa in five days and the previously daily accumulated rainfall associated was 66.80 mm (February 8 th ).Like this episode, there were two significant decreases of suction in this instrument.In January 2013, between 4 th and 14 th , there was a decrease of 42.72 kPa in eleven days with an associated daily accumulated rainfall of 52 mm (January 6 th ).The last event occurred, from October 11 th to 14 th , 2012, in which the suction measured decreased 39.2 kPa in four days with previously daily accumulated rainfall of 80.60 mm (October 11 th ).The other tensiometers also installed at the intermediate island (Fig. 11) had records with small ranges between minimum and maximum values, yet with higher variations starting from November, 2012.The minimum and maximum readings of the tensiometer installed at the greatest depth, TENS-04 (3.00 m), were 0.12 kPa and 16.66 kPa, respectively, with an average of 6.32 kPa, from May, 2012 to March, 2013.TENS-05 (1.00 m) had minimum and maximum readings of 1.09 kPa and 21.13 kPa, respectively, and an average reading of 7.38 kPa.TENS-06 (2.00 m) recorded a minimum reading of 0.17 kPa, a maximum of 24.84 kPa and an average of 6.78 kPa.
The reading variations of the instruments can be interpreted according to their location on the slope, that is, their positioning in the islands.For example: TENS-01 (1.00 m) and TENS-02 (2.00 m), installed at the lower island (Fig. 13), starting the reading variations in December, 2012 -month with intense yet disperse rainfalls.TENS-01 showed increases in the suction levels going from minimum readings of 4.04 kPa to maximum readings of 79.46 kPa with an average of 53.15 kPa.TENS-02 presented minimum readings of 9.77 kPa to a maximum reading of 68.84 kPa and an average of 38.70 kPa.These measurements were associated with their location on the steeper portion of the slope, which is more exposed to sunlight.
On the other hand, at the upper island (Fig. 14), the TENS-07 (1.00 m) and TENS-08 (2.00 m) presented low and constant suction values, ranging from 3.36 kPa to 13.32 kPa for TENS-07 and from -3.58 kPa and 6.42 kPa  for TENS-08, with average values of 5.97 kPa and 2.09 kPa, respectively.This island is less exposed to sunlight so that the local humidity can be much more preserved.
The daily accumulated rainfall and the total accumulated rainfall during the monitoring period can be observed and compared with the tensiometers records (Figs. 11,13,14), where the daily accumulated rainfall shows more influence than the total accumulated rainfall in tensiometers variation readings.Overall, the tensiometers presented an increase in the negative pore pressure (suction) during the period with less rain and a decrease as the rainfall occurs.
It can be observed that, in the instruments located at 1.00 m depth and less (TENS-03 at 0.50 m depth according to Fig. 11), daily accumulated rainfall over 40 mm caused also variations to the readings.In TENS-05 (Fig. 11), TENS-01 (Fig. 13) and TENS-07 (Fig. 14), for example, the events occurred in January and February, 2013, had similar effect, with abrupt falls after rainfall events, but with different range.As the instruments are installed at  deeper layers from the soil surface, they do not suffer such significant variations.
In the data presented (Figs. 9,10,11,13 and 14) it can be observed that the records were interrupted during the period of November 8 th to December 12 th , 2012, usually resulting from problems with the acquisition at the datalogger system.

Inclinometers
As for the monitoring of horizontal movements of the soil mass, INCL-01 located at the middle portion of the slope (intermediate island), presented stable accumulated reading values of less than +/-2 mm between the base and the top (Fig. 15).There have been two perceivable yet subtle areas of horizontal displacement accumulation, at depths of 2.5 m and 5.0 m, however, they were not significant.
INCL-02 located at the upper island (Group 1), also presented stability in its readings, with accumulated displacements lower than +/-2 mm (Fig. 16).The distortions observed in both instruments can be attributed to: (i) accommodation of the top silty sand soil layer as a consequence of hole drilling for tube installation, and (ii) torpedo readings.
The axes of the inclinometer tubes corresponded to the direction in which its casings were positioned in relation to the slope.Therefore, axes A and B corresponded to movements which are perpendicular and parallel to the slope, respectively.
In slope stability studies, the movement's magnitude and relevance are considered according to the horizontal displacement speed, with creep being the slowest process, with displacement rates of 15 mm/year (Cruden & Varnes, 1993).For both inclinometers (INCL-01 and INCL-02), the accumulated horizontal displacement measured over the slope monitoring period did not reach what is considered to be a soil creep phenomenon.The data point out the stability of the monitored slope, attesting to the adequacy of the stabilization structure implanted in situ.

Conclusions
The analysis of rainfall readings in this study demonstrates that, during the monitoring period, there were not any events of great magnitude (highest record equal to 260 mm in July, 2012), as opposed to those recorded in 2008, which accumulated approximately 1000 mm in November.There were also no heavy rain events during the monitoring period, which led to little significant variations of rainfall readings.Throughout the first three months of monitoring (from May to July, 2012), it was possible to note low intensity rainfalls which lasted for long periods.Starting from the months of August and September, 2012, the events were scattered, with low rainfall values.This can be characterized as a drier period in relation to the previous quarter.In the last months of monitoring (December, 2012 to February, 2013), the rainfalls had higher hourly intensity during reduced periods of time.The records measured were equal to 38 mm/h in January and February.This type of rain events tends to produce greater runoff and less water infiltration in the soil.Consequently, the positive pore pressure levels remained relatively stable.
The piezometer and tensiometer responses were in accordance with the daily accumulated rainfall; for example, an increase of piezometric levels and a decrease of suction values were observed after rain periods and, additionally, a decrease of positive pore pressure values and a suction increase were observed after periods without rainfall records.It is worth to highlight that the local soil type, characterized as silty sand, and the high fracture of the underlying rock, contribute to allow a faster drainage of the slope, thus reducing the increase of positive pore pressures.The piezometers showed a certain tendency towards stabilization of readings as a result of a few events of great inten-sity and heavy rainfalls.The piezometers installed at greater depths showed more significant readings, especially during July, 2012, which was characterized by low intensity rainfalls that lasted for long periods.This corresponded to an increase of the piezometric level, as a result of infiltration and rainfall accumulation.
Due to the vegetation growth over the monitoring period, the suction values tended to increase for the tensiometers installed closer to the surface, in the intermediate and lower slope islands (Groups 2 and 3).It was also verified that the suction readings in the study area decrease with location depth, which corresponds to the expected humidity profile for the active, non-saturated zone.As for the horizontal displacements, the readings analyses indicate stability, with values ranging up to +/-2 mm.

Figure 1 -
Figure 1 -Location of the study area.

Figure 3 -
Figure 3 -Sketch of the geological and geotechnical slope profile with instrumentation (WL = water level).

Figure 5 -
Figure 5 -Sketch of the instruments installed in the slope.
It is possible to observe that 44.11% of the rainfall events were classified under the category Dry Day (D S ) and 270 Soils and Rocks, São Paulo, 40(3): 263-278, September-December, 2017.
Figure 8 -Daily rainfall accumulation and water level monitoring at the study site.

Figure 9 -
Figure9-Readings of piezometers installed at the intermediate island.

Figure 10 -
Figure 10 -Readings of piezometer installed at the upper island.

Figure 11 -
Figure 11 -Readings of tensiometers installed at the intermediate island.

Figure 13 -
Figure 13 -Readings from tensiometers installed at the lower island.

Figure 14 -
Figure 14 -Readings of tensiometers installed at the upper island.
Figure 15 -Readings of accumulated displacements from INCL-01 at the intermediate island.

Table 3 -
Daily rainfall determined by means of the quantile technique.

Table 4 -
Results of the classification of rainfall events during the monitoring period according to the quantile technique.