Unsaturated seepage analysis at the Guayabo National Archaeological Monument, Costa Rica

. The Guayabo National Archaeological Monument is considered one of the most important historical and political ceremonial centers of pre-Columbian Costa Rica, Central America, and it depicts the ingenuity and the quality of life of Costa Rica´s inhabitants between 800 BC and AD 1400. This site was named International Historic Civil Engineering Landmark in 2009 by the American Society of Civil Engineers (ASCE). Evaluation of the unsaturated flow at the Northwestern slope using a two-dimensional model was performed. It was determined from field and modeling that at a relatively shallow depth the soil is relatively impermeable; thus, producing a large amount of run-off that tends to deteriorate the archaeological structures, and induce landslides. As part of the site investigation, exploratory borings were performed, and piezometers were installed in the upper, middle, and bottom parts of the slope. A series of laboratory testing was also performed to obtain index soil and permeability properties. The soil-water characteristic used to develop the K-Curve was also obtained. Additionally, a groundwater model was created using the geotechnical model and a water balance analysis for the area, where different scenarios of recharge and precipitation were analyzed taking into consideration the observed data. The volume of slope run-off through towards the archaeological site was estimated and the areas where it emerges, as well as the field groundwater.


Introduction
The Guayabo National Archaeological Monument is considered one of the most important historical and political ceremonial centers of pre-Columbian Costa Rica, Central America, and it depicts the ingenuity and the quality of life of Costa Rica´s inhabitants between 800 BC and AD 1400. This site was named International Historic Civil Engineering Landmark in 2009 by the American Society of Civil Engineers (ASCE), because advanced civil engineering infrastructure, for the time, has been revealed by the restoration activities in the monument.
The restored hydraulics infrastructure of the pre-Columbian era can still store, conduct, and evacuate water from the hillside. However, the flow patterns along the Northwestern slope, shown in Figure 1, are still not well-understood, making it difficult to plan water control structures to protect the monument. Additionally, there is a risk that high-intensity rains will cause landslides that will put at risk the archaeological findings at the site.
Several experts in archaeological monuments agree that, if the pre-Columbian hydraulic system is restored, the runoff water and groundwater will be better controlled, and this will likely increase the slope's stability. [1,2,3]. Therefore, it is necessary to model the groundwater component, which represents a considerable contribution to the flow that affects hydraulics structures. These analyses are imperative to design future restoration activities aiming to re-establish adequate hydraulics storage and piping system so the country's cultural heritage can be preserved. The main focus of this research is to evaluate the behavior of groundwater through the northwest slope of the Guayabo National Monument using a twodimensional flow model. In order to construct this model, an estimation of the geological-geotechnical and hydraulic properties of the soil was performed by means of laboratory and field tests. Later on, areas most affected by water were identified.

Site Characterization
The soil types found in the study area are classified as sandy clays with high plasticity. The soils increase the fine percentage as a function of depth. In addition, falling head laboratory permeability tests were performed.
The case of study is located on a basal geological substrate, composed by a lahar [4] consisting of a mudflow, which can be classified as a sand slime with high plasticity and corresponds to a non-consolidated matrix with fragments of basaltic andesitic lavas. The site lithology is characteristic of soils with very low values for the suction range greater than 10 kPa. Soil sample 1 was taken from the toe of the slope and soil sample 2 is taken from the toe of the slope. The soil hydraulic conductivity was estimated using the van Genuchten & Mualem equation [7].   Laboratory testing was performed to develop the soilwater characteristic curve shown in Figures 5 and 6. As it was mentioned before, there are two types of silty sand present on the site, with an upper, more permeable layer and a lower, less permeable. These curves were used to develop the K-curve used in the numerical model to study the underground seepage on the slope. Because of the difference in fine content, two materials that exhibit the same soil classification and with relatively similar grain size distribution may present very different hydraulic behavior. Table 4 shows the fitting parameter used for the van Genuchten & Mualem equation, which was used to develop the K-Curve and the Soil-water retention curve. Since most of the measured points are in the wetter part of the soil-water retention curve the fitting parameters were selected using the database and recommended values for this type of soil in the RECT computer program of the Environmental Protection Agency of the United States of America [8].

Seepage Analysis
An underground seepage analysis requires an understanding of the flow directions and magnitude. Figure 6 shows the overview of groundwater flow patterns considering equipotential curves every 2 m. In this figure, it is important to notice that the hydraulic gradient of the water mantle in the area is approximately 0.52 m/m, quite high considering the tendency of groundwater behavior. Therefore, it is expected that the flow rates would be quite high in this zone. It can be noticed how the gradient of the equipotential curves decreases as the height of the slope decreases, i.e., upslope the gradient is high, so the speeds are lower in this zone. In the lower part, due to the sudden change of topography, the curves start to space out, i.e., there is a hydraulic lower gradient. Fig. 6. Flow pattern along the slope. [7] Three different cross-sections (profiles 1, 2, and 3) along the slope were analyzed using the finite element groundwater seepage analysis capabilities of the Slide ® software from Rocscience. Two different scenarios: unsaturated field conditions for November, and unsaturated field conditions for November plus the calculated infiltration. The first scenario showed suction values of 19.61 kPa, and the second scenario showed values up to 24.5 kPa. It was observed from the models that the suction values are representative of near saturation at the mid-height of the slope and then increase again in the lower part. Table 5 shows the flow rates calculated for the three profiles at the top, middle, and bottom of the slope. Figure 7 shows the profiles selected for modeling. It is worth mentioning that the software capabilities allow for the consideration of the infiltration rates estimated from the water balance analysis obtained from the Schosinsky Method [6], but does not allow to consider the runoff water. This must be estimated otherwise. Fig. 7. Profiles selected to perform the modeling. [7] The models revealed that despite the presence of high precipitation, there is an unsaturated zone near the surface at the top and toe of the slope, so it is expected that the water that emerges in the middle part of the slope will turn into runoff water that recharges the surface water bodies identified on the field. Profile 1 shows a sharp decrease in the middle part of the slope, which is consistent with field observations. It means that this zone does not have groundwater recharge, but most of the water becomes part of the runoff water. Groundwater flow increases again in the lower zone, so there is a section between the middle and lower zone of the slope where the water can infiltrate again. Profile 3 showed similar behavior, but the midheight estimated flow drop is not as big. Therefore, it can be concluded that the superficial groundwater flow rate  The volu the one calc run-off need flow rate.
Speed pro appears in t slope. In add and the midd that ground r these speeds water. The sp the equipoten speeds of 3.8 2 shows spe cm/day.
The perm towards the the permeabl completely i high and the while the low because of samplings.
In the u gradually inc less permeab quickly in p profile is larg The midd permeability runoff water difficult.
It should permeability with precipit expected to because of in allow confirm abruptly with so the zones below this la the lower zon in Table 5 the natural moisture contents measured from soil samples obtained during the geotechnical exploration.
The upper stratum has a saturated permeability value of 2 E-1 cm/s, while the underlying stratum has a value of 1.79 E-4 cm/s. Therefore, the lower stratum prevents water infiltration because both layers have hydraulic conductivities with a different order of magnitude, suggesting that infiltration happens only in the upper stratum.
Both soil-water characteristic curves obtained displayed a typical behavior of fine soils. The initial section of the curve is very sensitive to the moisture changes produced by the presence of air into the pores. Therefore, the curves obtained described the hydraulic behavior of sampled soil strata when moisture changes occur on site. A wider range of suctions at drier water contents must be obtained in future research, in order to develop soil-water characteristic curves and K-curves that are more representative of the soil present in the site.
The equipotential curves modeled show a greater gradient in the upper zone of the slope than in the lower zone producing higher speeds that tend to favor pluvial erosion and reduce infiltration.
Flow rates of 84.97 liters/day were obtained in profile 1 at the toe of the slope; 3.0 liters/day for profile 2, and 0.91 liters/day for profile 1, under conditions without precipitation. This contrasted with values obtained in the upper slope zone of 3.31 liters/day in case of profile 1, 11.13 liters/day for profile 2 and 12.46 liters/day for profile 3. Flow rates at the bottom of the slope in profile 1 did not show significant variations as a function of precipitation. Consequently, the rain infiltrates very little in this section, draining superficially from the upper zone of the slope.
Discharge areas in the restored archaeological zone of the monument have a small contribution to groundwater; however, the greater volume of water that reaches this area is through runoff water.
The groundwater flow occurs in the most superficial layers, while the infiltration to the low permeability silt layer is reduced.
The shallow depth of the layer where the water can infiltrate causes that under heavy rain the soil infiltration capacity gets reduced, consequently runoff is the dominant mechanism through the surface of the hillside and has high drag capacity.
The degree of saturation in the hillside increases from the lower stratum towards the surface due to the low permeability materials at the bottom of the profile acting as a seal. These changes in the degree of saturation have a significant effect on the seepage pattern in the hillside as the soil nearest to the surface has higher suction values.
The high volume of water that floods the toe of the slope and the archaeological structures at the bottom, is due mainly to runoff water and not groundwater, because high suctions in the surficial soils and high-intensity precipitation in the area impede proper rain infiltration.