Migration of Water and Salt in Confined Archaeological Complexes: Comprehensive Investigation Below the Third Courtyard of the Prague Castle

ABSTRACT

Archaeological complexes and open-air museums must navigate the delicate balance between their societal role and preserving the displayed heritage. Sheltering from atmospheric hazards is a common preservation strategy, yet it introduces new degradation challenges. This study focuses on the persistent material deterioration, heightened moisture content, and salt efflorescence observed in protected excavations beneath Prague Castle’s third courtyard. The objective is to identify and analyse primary sources and pathways of water infiltration, comprehend complex interactions with masonry, and propose conservation strategies to mitigate the degradation. Through detailed field observations, using various moisture sensors, hydrochemical and salt analyses, rainwater emerges as the primary source of damp masonry and the transporting agent of salt-forming ions. Two penetration mechanisms were observed: ingress through perimeter walls and capillary rise where groundwater is present. Evaluation of moisture measurement methods underscores the superiority of the microwave method for historic masonry and the uranine-probe method’s efficacy for capillary water detection. Addressing water infiltration and salt accumulation challenges, we recommend preventive measures: (i) prevent water infiltration, seepage, and ingress; (ii) reduce salt concentrations; and (iii) stabilise site climate to mitigate salt crystallisation transitions. These recommendations extend beyond the study site, offering valuable insights for similar confined archaeological complexes.

KEYWORDS:

• Confined archaeological complexes
• conservation strategies
• masonry degradation
• moisture content
• preservation
• salt efflorescence
• sheltering

1. Introduction

Archaeological complexes as a cultural heritage are vulnerable to various types of deteriorating processes, including salt weathering or biodeterioration. A strong precursor of such processes is water (Alfano, Palella, and Riccio 2023), which can act as a transporting agent of dissolved salts (Goudie and Viles 1997) or allow for microbial colonisation (Miller et al. 2012 and references therein). These archaeology museums are specific sites that display unearthed cultural relics or building ruins; they are often underground and (semi)-confined, but they were not designed this way. Sheltering such complexes aims to protect the sites from the outdoor environment, mitigate fluctuations in temperature and humidity, or reduce wind velocities. However, in confined and semi-confined spaces, the deterioration forces can be paradoxically amplified, leading to rapid degradation of built heritage. Many examples can be found worldwide, Table 1.

Table 1. Examples of confined and semi-confined archaeological museums.

Site	Location	Description	Reference
Cagliari	Sardinia, Italy	Multilayered complex from the Republican Roman period until the Middle Ages; remains of walls built with limestone blocks enclosing rectangular areas.	Carcangiu et al. (2015)
The archaeological site of the cathedral of St. Peter	Geneve, Switzerland	Underground archaeological complex of the cathedral’s history from Roman times.	Bonnet (1987)
Yoshimi Hyaku Ana	Saitama Prefecture, Japan	Underground archaeological complex, sacred tumuli, and megalithic burials of the uppermost social classes of ancient Japan created between the 3rd and 7th centuries CE.	Germinario and Oguchi (2021)
Megalithic Temples	Malta	Megalithic temples dating back to the Neolithic period.	Cassar et al.(2018)
Hal Saflieni	Paola, Malta	Hypogeum, or underground burial chamber, dating back to the Neolithic period.	Cutajar, Farrugia, and Micallef (2019)
Çatalhöyük	Anatolia, Turkey	Neolithic settlement dating back to the 7th millennium BCE, with interconnected houses and courtyards, and evidence of early urbanism.	Lercari (2019)
El Molinete	Cartagena, Spain	Archaeological Park located on the hill of El Molinete, with remains of Roman and Carthaginian fortifications, as well as a Roman theatre.	Manzano Fernández et al. (2022)
TerraCotta Warriors	Xi’an, China	Museum housing the Terracotta Army, a collection of terracotta sculptures depicting the armies of Qin Shi Huang, the first emperor of China.	Xia et al. (2023)
Akrotiri — Santorini	Cycladic island of Thera (Santotini), Greece	Archaeological site of the Bronze Age settlement buried by the eruption of the Thera volcano, preserving the remarkably well-preserved remains of houses, streets, and frescoes.	Doumas (2013)

Proper environmental control is crucial for protecting the relics displayed in these museums. This includes regulating heat transfer, water transfer, and indoor airflow control (Xia et al. 2023; Xu et al. 2023). However, underground archaeology museums face greater challenges in maintaining a stable environment compared to indoor display museums (Luo et al. 2015), which can result in the degradation of the heritage. Underground sites are characterised by natural or seminatural settings, which have different object-environment interactions compared to open-air built heritage (Germinario and Oguchi 2021). These sites tend to be very humid (Carcangiu et al. 2015; Germinario and Oguchi 2022) and are susceptible to salt weathering, involving efflorescence and crusts composed of mixed soluble salts, mostly hydrated.

Efflorescences and crusts, characterised by whitish coatings, alter the appearance of archaeological sites by forming on walls and artefacts, often covering extensive surfaces. Subflorescence, where salts crystallise below the surface (Vergès-Belmin 2008), can also occur, concealing the presence of salt and sparing the surface from visible alteration. However, subflorescence often leads to salt scaling and irreversible damage to historical materials. The chemo-mineralogical composition of formed salts depends on the interaction between the building material and the environment, including mineral dissolution and atmospheric deposition. The diversity of building materials also influences salt composition variability (Germinario and Oguchi 2021). Furthermore, chemical composition of infiltrating water, whether from above or below, also plays a decisive role in the subsequent salt composition and the entire salt weathering process.

In confined archaeological complexes, water serves as the primary driving force of salt weathering. Water enters porous materials either in the liquid phase from atmospheric precipitation, by capillary upwelling of groundwater or possibly from damaged sewage/water systems. Alternatively, water can enter gaseous phase through condensation or due to the material’s hygroscopic nature due to the presence of salts (Hall and Hoff 2021; Charola 2000; Sass and Viles 2022). Acting as a carrier, water transports salt-forming ions from leached building material components or other sources like the atmosphere, soil, or bedrock (Jiang, Guo, and Polk 2015). Moreover, water controls salt stability and cycles of crystallisation/deliquescence and hydration/dehydration by controlling the microclimate. Firstly, the microclimate in semi-confined archaeological museums is influenced by the rate of evaporation from the terrain/masonry. Secondly, the presence of water within masonry directly affects the relative humidity and temperature within the masonry pores, the actual place where the stability of the salt is determined. Understanding the specific thermodynamic behaviour of the involved salt phases can assist in predicting the severity and timing of salt weathering and resulting damage (Germinario and Oguchi 2022). Thermodynamic models, particularly ECOS-RUNSALT, are often used to estimate weathering caused by complex salt solutions, calculating salt volume based on temperature and relative humidity conditions (Godts et al. 2022; 2023; Menéndez 2017; Pintér 2022).

To formulate a comprehensive conservation plan for an archaeological site, it is important to understand (Blavier et al. 2023; Ma et al. 2024) the sources and pathways of water migration, chemical composition of water, building material properties, salts present, and the prevailing microclimate characteristics (air temperature and humidity, air flow). It is equally important to understand the interplays between them (Blavier et al. 2023), to decipher the underlying processes, and to account for their spatial and temporal variations. Therefore, such a complex challenge is ideal to tackle on a specific site that allows for necessary environmental observations. Thus, employing case studies is always useful, as they not only provide practical site-specific insights, but also contribute to the wider pool of knowledge about similar archaeological sites and the potential challenges they present, thereby enhancing our ability to address similar issues in different contexts.

1.1. Objective/research questions

The present article addresses the challenge of high moisture content and the occurrence of salts using an example of confined archaeological excavations beneath the third courtyard of the Prague Castle. The objectives of this paper are as follows.

1. To determine the origin of water within the archaeological complex employing a multidisciplinary approach.

2. To describe the long-term and spatial variations in moisture content within the historical walls of the complex by comparing different methodologies.

3. To identify the presence of salts and establish their relationship with the water.

4. To suggest solutions for the conservation of the site.

The excavations studied serve as a representative example of underground constructed sites, which lends potential for generality to our findings.

2. Methodology + description of the area

The archaeological area beneath the third courtyard of the Prague Castle was established following extensive archaeological investigations led by Karl Fiala in 1920s. The Prague Archaeological Commission, headed by Karl Guth, and the State Archaeological Institute, led by Ivan Borkovský, played a pivotal role in this endeavour (Maříková-Kubková 2020). The area lies beneath the courtyard floor, upheld by a reinforced concrete slab (Figure 1), designed by Jože Plečnik. This slab is supported by columns anchored into the terrain and the underlying historic masonry. A specific focus of the study was on the central part of the area with visible water sources with the Romanesque chapel of St. Bartholomew (Figure 2), with the remains of a late medieval/early modern structure traditionally referred to as a bridge (see 2.4.3 Pillar). Currently, the area faces ongoing sandstone degradation attributed to various factors including settling of the courtyard slab, salt weathering, and/or expansion of the material due to moisture condensation.

Figure 1. Illustrative example of the confined archaeological complex with the massive reinforced concrete structure supporting the third courtyard of the Prague Castle.

Figure 2. Schematic map of the area with the location of the monitored sites.

To address the challenge of high moisture content and salt presence at the site, we chose a range of methodologies investigating possible water sources, measuring moisture in masonry over time and space, analysing the ion content of present water, and assessing salt presence in historical masonry (see Figure 2). Initially, we conducted a thorough site survey to identify possible sources of water such as leaks, drains, and wells (2.2) and analysed the measured materials (2.1). We then compared groundwater levels with precipitation events and performed a hydrodynamic test to determine the hydraulic characteristics of the subsurface (2.2). This was supplemented by a hydrochemical analysis (2.2), contributing to a more comprehensive interpretation of the water pathways. Additionally, to identify water infiltration origins including potential ambient humidity condensation, we regularly measured rock moisture content (2.4). For deeper insight into the moisture temporal evolution within masonry, we focused on a site characterised by extreme fluctuations in temperature and humidity (2.4.3 – the Pillar site), where we applied a series of techniques for moisture content measurements. This setup allowed us to assess the suitability of different moisture monitoring approaches. To complete the picture of moisture and salt dynamics, we analysed efflorescence salts and sampled masonry dust from selected depths for ion content (2.3).

2.1. Materials

The archaeological objects in the area below the third courtyard consist of several types of construction materials. The predominant material of most of the local buildings there is a stone called opuka, a sedimentary rock of local provenance (fine-grained calcitic siltstone) and, to a lesser extent, fine-grained arkosic sandstone (Kozlovcev et al. 2023). The stone in the buildings is dressed in regular blocks, typical of local Romanesque buildings, and bonded with lime mortar. The thickness of the masonry is variable, but, for example, the perimeter masonry of the church of St. Bartholomew is 80 cm thick. In addition to these inorganic materials, there are also valuable historical remains of wooden structures that have been heavily impregnated with carbolic acid. Others, such as partitional and perimeter brick walls, concrete and brick structures supporting the archaeological findings, and especially the reinforced concrete roof slab and its columns, are structures added during the twentieth century when the site was established as the archaeological area. Research focuses on porous building materials that are susceptible to degradation due to the presence of moisture and associated salt activation (Kozlovcev et al. 2023).

To characterise the ability of the material to absorb and transport moisture, selected physical properties were determined using standardised methods (Table 2). Bulk density (g/cm³), open porosity (vol. %), 24 h water absorption under atmospheric conditions (wt. %), capillary absorption coefficient (kg/(m²×√h) and water vapour diffusion resistance (-) were determined by EN 772–4: 1998, EN 13755: 2008, EN1015–18: 2002, EN 1015–19: 1998, respectively.

Table 2. Determined physical properties of opuka and sandstone units of the archaeological area.

Material	No. of samples	Bulk density (g/cm^3)	Open porosity (%)	24 h water absorption (wt. %)	Capillary absorption coefficient (kg/(m^2*√h))	Water vapour diffusion resistance
Opuka	10	2.01	24.60	12.32	2.6	33
	St. Dev.	0.18	6.44	2.97	1.2
Sandstone	3	1.86	28.03	9.46
	St. Dev.	0.43	1.98	3.31

2.2. Water sources

Water as a transport agent in confined archaeological sites can have three primary sources: precipitation from above, capillary rise from groundwater, or utility leaks. The excavations were first subjected to a visual survey to find suitable objects for groundwater monitoring, springs, or places with groundwater on the surface. To discover water sources in the area of interest, we monitored four boreholes (A, B, C, and D in Figure 2) to understand groundwater behaviour. Additionally, we also monitored three drains in the brick wall near the borehole A, the surrounding of the borehole A with occasional surface water, a pond below the drains, and water coming from a calthemite straw under the path (Figure 3) to determine the role of precipitation or possible utility leaks. Based on moisture monitoring, possible sources of water seeping were identified. In one of these potential sources, the collector area adjacent to the site, we performed a field physical investigation including the condition of the storm sewer outfall using a DURAMAXX Inspex-2000 inspection camera.

Figure 3. Location of the Pillar next to the Chapel of St. Bartholomew under the third courtyard of the Prague Castle. Seethe measurement points P1-P6on the right, the drains, borehole A, and the location of the calthemite straw.

The presence of groundwater beneath the third courtyard was monitored in all 4 boreholes. To determine the origin of the water and its potential to flow in the anthropogenic deposits, a pumping and subsequent recovery tests were conducted at borehole A. The short pump test was carried out on 20/8/2020 in borehole A using the SUBMERSIBLE 188 pump with which we pumped all the water out of the borehole in 18.5 minutes. The water level was measured using a handheld level gauge. Transmissivity was calculated from the recovery test using the Theis method (McElwee 1980).

To further determine the origin of the water, we collected water samples (Figure 2 and 3) from the drains, pond, borehole A and calthemite straw, which were subsequently analysed for basic chemistry. The samples were filtered and those for cation determination were stabilised with HNO₃. Cations were analysed by inductively coupled plasma emission spectrometry (Thermo Scientific) and anions by liquid chromatography (Dionex ICS-2000). Chemical analysis was then evaluated using the PHREEQC geochemical programme. Due to the low yield of the water from a calthemite straw, the water was captured in a container and collected for analysis after approximately three weeks. In addition, a trench was excavated on February 1, 2022 by the west side of St. Bartholomew’s Church where the wall appeared the wettest Figure 3. The trench, due to the exposure, produced a concentrated short-term spring with a discharge in the order of millilitres per second. This water was also collected and analysed (Figure 2).

2.3. Salts in masonry

From the Pillar, the powder obtained during the depth drilling holes for moisture measurements (ER-Scan, see 2.4.1.) was analysed to determine the salt ions content in the material. The drilling for the probes was carried out at five different heights and the powder was collected up to a depth of 2.0 cm. In August 2023, two samples of salt efflorescences that formed on the surface of the stone were collected to identify the crystallised phases using powder X-ray diffraction (Malvern PANalytical Aeris). At the place under the efflorescence (height 1.07 m), another hole was drilled in three steps to a depth of 2.0 cm (0–1, 1–1.5, and 1.5–2.0 cm) with the aim of monitoring the distribution of salt ions under the salted surface. Finally, a mortar sample was taken at the same height and close to the site where the salt efflorescence formed. The concentrations of anions and cations were determined using ion exclusion chromatography (Dionex ISC-5000). The stone or mortar powders were dried at 40°C until they reached a constant mass before being mixed with ultrapure water and extracted for 24 hours to dissolve the water-soluble components contained in the sampled material.

2.4. Moisture in masonry

This section describes how the moisture characteristics were measured by each method. There were three types of sites (Figure 1): the Pillar (2.4.3; several measurement points along a vertical axis), Point measurements (2.4.4; point measurements on different materials spread throughout the entire archaeological complex), and Perimeter walls (2.4.5; grid measurements on selected walls). Note that the methods presented (2.4.1) are not applicable to all site types, and the specific measurement points are listed below and in the Table 3. Since many non-destructive techniques do not meet criteria for moisture measurement of masonry (Alfano, Palella, and Riccio 2023), we used a combination of different techniques: methods measuring electric resistance (see ER-GMR and ER-Scan below), electromagnetic methods measuring dielectric properties (see EM-Teros, EM-GMK, and Moist below), and directly measuring the presence of capillary water (see uranine-probe below).

Table 3. Description of methods to measure the moisture characteristics of masonry. *Measurement frequency is for Point measurements and Pillar; for Perimeter wall mapping, measurements were only taken on these days: 6/9/2021; 8/11/2021; 11/3/2021; 12/7/2021; 1/12/2022; 22/2/2023; 3/5/2023 and 16/8/2023. **Measurements with EM-Teros were completed on 3/21.

Method abbreviation	Device	Measurement accuracy	Measurement frequency*	Place of measurement
ER-GMR	Greisinger GMR 110	0.2%	irregular (~21 days)	Pillar (points P1-P6), Perimeter wall mapping
ER-Scan	Material Moisture Gigamodule (Scanntronik)	-	6 hours	Pillar (points P1-P6)
Uranine-probe	Probe prepared according to Weiss et al. (2020)	1 mm under ideal conditions	irregular (~21 days)	Pillar (points P1-P6)
EM-Teros**	METER TEROS12	3%	1 hour	Pillar (point P4)
EM-GMK	Greisinger GMK 100	-	irregular (~21 days)	Perimeter wall mapping
Moist	MOIST 210B + DM 0157 (Moist-15) or PM 02181 (Moist-30)	2%	irregular (~21 days)	Pillar (points P1-P6), Building materials (points M1-M41), Perimeter walls,

2.4.1. Methods of measuring moisture characteristics

ER-GMR: Manual measurement of electrical resistance by the Greisinger GMR 110 device with electrodes 26 mm apart, following measurement methodology by the manufacturer.

ER-Scan: Electrical resistance using Material Moisture Gigamodule (Scanntronik) with automatic recording. Electrodes were installed in predrilled holes with a diameter of 3 mm and 3 cm apart, sealed in a conductive paste provided by the manufacturer to ensure proper contact with the material. The ER-Scan was used on the Pillar (P1-P6) at three different depths below the surface, namely at a depth of 1–2 cm (hereafter referred to as S, “short”), 4–5 cm (M — “medium”), and 15–19 cm (L — “long”).

Uranine-probe: This recently introduced method (Weiss et al. 2020) for detecting capillary water (e.g., wet parts) is based on the reaction of uranine dye (sodium fluorescein) with moisture in the material, which causes the dye to change colour. A stainless-steel needle coated with the uranine powder was inserted into the 3 mm diameter predrilled holes into which ambient air was injected prior to measurement. Interpretation followed the methodology recommended by Weiss et al. (2020): when the relative humidity of the air was < 73%, the uranine-probe was left in the hole until the probe reading changed due to reaction with the capillary water, but not more than 15 minutes. The depth of the capillary water below the surface was interpreted as the clear green-red boundary. When the relative humidity was > 73%, the capillary water depth was interpreted as a boundary between indistinct red-green and green with droplets. If there was no capillary water present, a virtual depth of 105 mm depth has been prescribed for better data visualisation. The uranine-probe was used at the Pillar (P1-P6) site.

EM-Teros: Automatic measurement using the METER TEROS12. EM-Teros was used at the Pillar in the vertical level of point P4, where the electrodes were inserted into a narrow crevice between two stones that form the building material of the Pillar.

EM-GMK: Greisinger GMK 100 device was held against the surface during measurement following the manufacture’s guidelines.

Moist: This method refers to non-invasive manual measurement using the MOIST 210B device. We used two different probes with different depth ranges: DM 0157 with a range of 0–15 cm (hereafter referred to as Moist-15) and PM 02181 with a range of 0–30 cm (Moist-30). The probe was pressed to the surface to obtain the data; three values were measured each time and the average was taken for evaluation. Moist was used at the Pillar (P1-P6), Point measurements, and the Perimeter walls.

2.4.2. Calibration

Since most of the methods used (except for the uranine-probe) do not directly determine moisture content or presence, we performed the following calibration. For ER-GMR and Moist-15 and Moist-30, we used an authentic opuka sample from the archaeological site. The 70 × 50x20 cm sample was placed in a climate chamber where different humidity conditions were simulated, and both completely dry and saturated conditions were used. The moisture content was determined gravimetrically. By taking measurements at three points on differently treated surfaces of the calibration sample, the relationship between the moisture proxy (moisture index in the case of Moist-15/Moist-30 and reference values in the case of E-GMR) and the moisture content (wt.%) was derived. For the calibration of the EM-Teros, we used the calibration equation of volumetric water content for mineral soil provided by the manufacturers (METER) and by dividing the volumetric water content by the measured density of the opuka masonry, we obtained the gravimetric water content.

ER-Scan calibration was performed as follows (Kalianková 2022): First, the measured electrical resistance data were converted to specific electrical resistance. Next, the value of the specific electrical resistance was calibrated to the temperature measured in the masonry using the relation of Weast, Astle, and Beyer (1989). The value of the correction factor, which depends on the amount of dissolved electrolytes in the aqueous solution, was chosen at 2.2% per K as the standard value according to Weast, Astle, and Beyer (1989). We then used the calibration equation provided by the study by Sass (2022) for natural sandstone.

2.4.3. Pillar

The “Pillar” was selected for long-term moisture monitoring in the masonry due to its specific humidity and temperature conditions. This site is a part of a late medieval or early modern bridge whose foundations extend below the present ground level (Figure 3) and lies in the most moisture-laden place under the third courtyard with groundwater present nearly year-round at a depth of 30 cm and periodically also at ground level. At the base of the Pillar there are relatively small temperature fluctuations compared to the rest of the archaeological complex due to the presence of water. On the contrary, the top of the Pillar (with a total height of 3 metres) experiences considerable temperature fluctuations throughout the year due to its proximity to the courtyard pavement above, which is fully exposed to the sun in the summer.

The measurement campaign on the Pillar lasted from August 2020 to April 2023. To obtain a basic overview of the spatial and temporal variations of the moisture content, a total of six different methods were used (ER-GMR, ER-Scan, uranine probe, EM-Teros, and Moist-15 and Moist-30; see Section 2.4.1 and Table 3 for details). The frequency of the measurements was taken several times a day for the automatic recording devices, while the manual measurements were taken approximately once every 21 days. Moisture characteristics were measured at a total of six vertical levels (P1-P6, unless otherwise stated) from 35 to 240 cm above ground level (Figure 3).

2.4.4. Point measurements

Forty-one measurement points (M1-M41, Figure 2) using Moist-15 and Moist-30 were selected to monitor the moisture content in relevant building materials (opuka, sandstone, brick, concrete, wood and ground). The objective of the assessment was to describe locally occurring microclimates or locations where signs of water ingress or the existence of moisture and salt related degradation were visible. Several measurement points per location and for each point, the condition of the material and its surface flatness were evaluated. Repetitive measurements were performed at regular intervals of 3 weeks. At the end of 2021, after a partial evaluation of the results, the Moist-30 measurement was discontinued, and Moist-15 was used in a measurement interval of 5 weeks.

To describe in more detail the distribution of moisture in the masonry where capillary rise likely occurs, three sites in the vicinity of St Bartholomew’s Church were selected. Its western wall (points M8-M12), its wall in an archaeological probe sunk below ground level on the south side of the church (points M13-M18), and the Pillar (2.4.3), Figure 3. For comparison, a measuring point (M1) was added to represent the relatively dry masonry near the entrance to the site.

2.4.5. Perimeter wall mapping

Moisture was regularly monitored at three specific locations on the perimeter wall (Figure 2). This brick wall separates the permanently uncovered part of the site from the remaining ground. The roof structure changes in the locations behind the wall where the concrete slab is already placed directly on the ground. From the situation, it can be estimated that on the hidden side of the wall there is altering backfill, remnants of historic structures and recent concrete structures that support the concrete slab and provide support against the lateral pressure of the surrounding soils.

The three monitored sections of the wall showed signs of repeated moisture penetration, and therefore repeated measurements were taken at a rectangular grid of points to better understand the mechanism of moisture ingress through the masonry. The monitored sections were located in the central part of the site, the north wall (Figure 8a), the west wall in front of St Bartholomew’s Church (Figure 8b) and the north wall near the entrance (Figure 8c).

Measurements were carried out every 3 months (from September 2021 to August 2023) using non-invasive methods (Moist-15, ER-GMR). The resulting moisture map (Figure 9) is a simplified model that assigns to a given square of the matrix the value determined at the measured point.

3. Results

3.1. Water sources

Infiltration of water from the pavement in the third courtyard, which penetrates the concrete structure, particularly during intense rainfall events, was observed throughout the archaeological site (Figure 4). Among the three monitored drains, one remained dry for most of the monitoring period, while the other two exhibited sporadic discharge ranging from 0 to approximately 0.05 l/s. During the short pumping test (19 min.) in borehole A, the water level dropped 12 cm and the total amount of water pumped was 47 l. After 12.5 hours from the end of pumping, the water level increased to 2.4 cm below the original level, indicating limited water inflow. The transmissivity, determined from the recovery test, is around 8–10 m²/day, which corresponds to low flow rates. A pond occasionally forms beneath the drain outlets, and the water persists nearly year-round due to this low permeability of the underlying sediment. Based on the position of the drains, the observed seepages are most likely to originate on the floor of the third courtyard, where water can infiltrate through the joints between the tiles.

Figure 4. Visible infiltration of water in selected locations in the archaeological area below the pavement of the third courtyard after an intense rainfall event.

During the monitoring period from 2020 to 2023, water was only observed in borehole B on one occasion. On the contrary, the borehole A consistently maintained a water level that was continuously recorded along with the drains and the calthemite straw (Figure 5). The graph of water level and rainfall shows an almost immediate (approximately one day) response in both the well and the pond to precipitation events, and in periods without rain, we observed a steady decline in water level.

Figure 5. The water level in the borehole A and the pond compared to precipitation events show the ~ 1 day lag in water level increase after rain events.

Based on the chemical analyses results, HCO₃⁻ emerges as the dominant anion (Table 4) in all instances where alkalinity was measured. The water from the trench and the borehole has K and Na as the dominant cations. The water in the pond has the chemical type K-Na-Ca-HCO₃. Water from the calthemite straw has the chemical type Na-K-HCO₃-Cl. The water of the calthemite straw also differs in its total mineralisation, which is 240 mg/l, while the other waters have mineralisation between 500 and 550 mg/l. This difference may be because the water from the calthemite straw is not sampled immediately due to its low yield but is sampled after three weeks of accumulation. During this time, massive carbonate precipitation occurs on the bottom and walls of the sampling container (clearly visible several mm thick layer). All samples have a high pH (9.9–11.9), so ion speciation is typical of an alkaline environment. Calcite, dolomite, and iron oxohydroxides precipitate from the environment. In addition, the solutions are supersaturated toward some silicates, but significant precipitation is not expected because of the extremely slow kinetics of silicate precipitation.

Table 4. Results of the hydrochemical analysis of the sampled water. The hydrochemical composition of precipitation in the vicinity of Prague was added for reference (Czech Hydrometeorological Institute 2022).

	date	Ion concentration (meq/l)								TDS (g/l)	pH
		Cl^−	SO4^2-	NO3^−	HCO3^−	Ca^2+	K^+	Mg^2+	Na^+
Pond under drains	20.08.2020	1.10	0.76	0.76	3.95	1.44	2.12	0.007	1.98	0.55	10.26
Borehole A	20.08.2020	1.15	0.77	0.78	3.38	0.96	2.12	0.025	2.08	0.51	9.89
Drain	15.10.2020	0.56	0.27	0.32	-	1.91	1.84	0.003	1.58	-	10.93
Surface water around borehole A	15.10.2020	0.62	1.23	0.39	-	2.00	1.93	0.003	1.78	-	10.42
Calthemite straw	17.12.2020	0.59	0.15	0.16	1.87	0.19	1.00	0.006	1.81	0.24	10.23
Trench	02.02.2022	0.40	0.57	0.18	5.52	0.72	2.00	0.001	2.02	0.55	11.85
Precipitation	2022 average	0.01	0.02	0.02	-	0.01	<0.01	0.004	0.01	-	5.29

3.2. Salts in masonry

The ionic water-soluble composition of the powder samples measured by ion chromatography (Table 5) indicates that the salts were not concentrated at the locations of the probes chosen for moisture content monitoring (see Figure 3) except at the place with the salt efflorescence (Table 6). The results show that at 1 m above ground, there is a higher concentration of sulphates in the surface layer, with a primarily sodium sulphate efflorescence (Table 7 and observed in other parts of the archaeological complex, unpublished). Under efflorescence, there is also a very high concentration of chlorides, which drops with depth, and a higher concentration of nitrates, which does not reach depth. However, the mortar has twice as many sulphate anions and a much higher concentration of nitrate anions than the stone.

Table 5. Results of the ion chromatography analysis, samples of stone from the Pillar. The results are presented as the weight percentage of the sampled dust.

Place	Height (m)	Depth (cm)	Ions detected (wt.%)
			Cl^−	NO3^−	SO4^2-	Na^+	K^+	Ca^2+	Mg^2+
Point P1	0.35	0–2	0.01	0.01	0.09	0.06	0.05	0.12	0.003
Point P2	0.73	0–2	0.01	0.01	0.02	0.03	0.04	0.09	0.002
Point P3	1.23	0–2	0.01	-	0.24	0.05	0.04	0.16	0.003
Point P5	1.75	0–2	0.06	0.04	0.08	0.03	0.03	0.17	0.007
Point P6	2.40	0–2	0.03	0.02	0.05	0.03	0.04	0.13	0.003

Table 6. Results of the ion chromatography analysis, stone and mortar samples at the place of efflorescence from the Pillar. The results are presented as the weight percentage of the sampled dust.

Place	Height (m)	Depth (cm)	Ions detected (wt.%)
			Cl^−	NO3^−	SO4^2-	Na^+	K^+	Ca^2+	Mg^2+
Stone under Efflorescence	1.07	0–1	0.61	0.15	0.20	0.24	0.03	0.12	2.67E–04
	1.07	1–1.5	0.34	0.06	0.06	0.11	0.02	0.07	7.45E–05
	1.07	1.5–2	0.26	0.05	0.06	0.08	0.02	0.05	1.22E–04
Mortar	1.07	0–1	0.21	0.82	0.43	1.12	0.04	0.01	1.31E–04

Table 7. Mineralogical composition of salt efflorescences (wt.%).

Height (m)	Thenardite Na2SO4	Gypsum CaSO4.2 H2O	Calcite CaCO3	NaHCO3	Hydroxyapatite Ca5(PO4)6(OH)2	Ca(NO3)2
0.94	74	14	11	-	-	-
1.07	38	30	2	2	5	1

3.3. Moisture in masonry

3.3.1. Pillar

Moisture, as expected, generally decreases with height from the wet terrain (with a gravimetric moisture content of 15–17% throughout the year) at the base of the Pillar. At the same time, the variability in moisture over seasons is almost negligible. The results of the respective methods also point to their methodological advantages and downsides.

Moisture content measured by Moist shows two distinct groups of points over the measurement period: a wetter area (points P1-P3) with a relatively high gravimetric moisture content typically between 5–10%; and a more elevated drier area of points P4-P6 with predominant moisture content ranging 0–3%. Similarly, ER-GMR also shows the driest point (P6, average moisture content 3%) located toward the top of the Pillar. However, other measurement points show high variability, with the highest average moisture content (10%) recorded at point P4. ER-GMR is probably also influenced by the presence of salts and the contact between probes and material (see Weiss and Sass 2022, and the discussion therein). Using the uranine-probe method, we observed liquid-capillary water on the surface of the material at the three lowest points and points P5 and P6 were often dry throughout the measurement depth (in Figure 9b shown as 105 mm depth). The two highest points also experienced large fluctuations from wet material, when liquid water was detected on the surface (in the case of P5) or at a depth of 2 cm (P6), to dry material, when liquid water was not detected even 10 cm below the surface. And although some observations may be affected by the accuracy of the method under high relative air humidity (see Weiss et al. 2020), these cannot account for the observed fluctuations of several cm. Therefore, these fluctuations are likely caused by fluctuations in relative humidity and the dissolution/crystallisation of the present salts. The kinetics of this process could explain the observed fluctuations in the depth of the capillary water. However, this method confirms the assumption that the measurement points at the foot of the Pillar are wetter due to upwelling, while those located higher are affected by drying due to evaporation, which is determined by the local microclimatic conditions.

3.3.2. Point measurements

The moisture content of the materials varies depending on their position relative to the site, the structure, and the surrounding terrain, here we present selected moisture measurements (Figure 6). Generally, in areas without moisture ingress, the masonry has typically moisture content between 0.5–3 wt.% but also below 0.5 wt.%. This contrasts with areas with active moisture (ingress or capillary rise) with moisture content of 3 to 9 wt.%. In places with a supposed capillary rise (M13-M19; M8-M12; M25-M31), the typical gradient was identified, where the moisture content was highest at the bottom of a structure and decreased with height. Generally, all the materials studied (opuka stone, concrete, terrain, and wood) show little seasonal variation in moisture content. The highest fluctuation of moisture content and its values belong to the Pillar (3.3.1) and in general, higher moisture content was detected in the area around the St. Bartholomew church. Here, there are some notable differences in moisture contents: on the east wall of the church (M13-M19), the moisture content is considerably lower than at the west wall and the Pillar, even though the measured points are below the level of the west wall and comparable in height to the Pillar site. There is an occasional rise in moisture content on the west wall (M8-M12) during the summer months.

Figure 6. Moisture evolution of the lower and upper points of the monitored profiles of the three opuka walls compared to the point without capillary moisture influence (point M1).

3.3.3. Perimeter wall mapping

The moisture map of the North wall section includes historic opuka-made masonry at the bottom and modern brickwork at the top. There is a concrete beam between the stone and the brick masonry. Both masonries contain soluble salts that are also visible as efflorescence on the surface (Figure 7) and show an elevated moisture content, which is rather well distributed, except for the concrete beam that divides it.

Figure 7. Moisture content measurement points in the North wall (a), West wall (b) in front of St. Bartholomew’s Church, and North wall near the entrance (c). Each point represents its surrounding area (depicted in squares) in the simplified moisture maps (Figure 8).

The brick perimeter West wall (Figure 7b) is almost entirely covered with multiple layers of leached portlandite (Ca(OH)₂) that has carbonated. The lower part of the wall is plastered (thickness approximately 10 mm) to a height of approximately 40 cm. Visible salt efflorescence is observed in several places around the joints at the bottom of the wall. Regular moisture measurements show three primary locations with persistently high moisture levels year-round, approaching the saturation of the bricks (above 8 wt .%; Figure 8b). Only at the top of the measured profile (approx. at 2 m) is the moisture content low or even very low. Again, this likely reflects moisture distribution behind the wall, suggesting the presence of water present within the materials. The extent of coverage by the leached lime layer indicates prolonged water seepage across nearly the entire area of the masonry in the past.

Figure 8. Moisture maps of the North wall section (a), West wall in front of the St. Bartholomew’s Church (b) and of the North wall near the entrance (c) with a colour scale of the moisture content.

The moisture map of the North wall near the entrance shows the main locations of moisture seepage (Figures 7 and 8c). The brick perimeter wall exhibits elevated to very high moisture content throughout the monitoring period. Partial drying occurs during the year when conditions are suitable. The areas of brickwork that cover the concrete column are less damp, while the lower right corner is characterised by the highest moisture. The masonry surface is covered with a layer of precipitated calcium carbonate. Its thickness is much lower compared to the other surfaces in the site. This may indicate some relatively recent changes in the penetration of moisture through the layers of the roof/pavement and materials behind the wall.

4. Discussion

4.1. Assessment of the moisture measurement methods

In our investigation of masonry moisture, we have used various methods measuring electric resistivity, dielectric permittivity, or the presence of water. The microwave-based device emerged as the most reliable based on the repeatability of measurements, which also corresponds to findings in other studies (Weiss and Sass 2022). Both sensors with claimed measurement depths of 0–15 and 0–30 cm demonstrated sufficient accuracy. An advantage of this electromagnetic method is its noninvasive nature, which allows the sensor to be placed directly on the surface without the need for material interference. The local flatness of the surface to which the probe is attached is important for the quality of the measurement. We can thus endorse the use of electromagnetic techniques with a deeper penetration depth (at least several cm), where the inaccuracy due to surface roughness is relatively reduced.

Electric resistance, even cost-effective, cannot be recommended for use in similar settings as also reported by Alfano, Palella, and Riccio (2023). A crucial prerequisite for correct measurement is a good contact between the material and the electrode (Phillipson et al. 2007). Although we established measures for good contacts, that is, a conductive paste or a firm and repeated pressing of electrodes against the material’s surface, our results show that these measures were insufficient. This can be seen especially in the case of ER-Scan (P1-S in Figure 9) and on the variability of the ER-GMR data (Figure 9). And although a promising approach has been proposed by Cacciotti (2020), who applies an epoxy-graphite conductive sealant, this is a significant drawback of all electric resistance methods. Another issue with using the electric resistance as a proxy to water content is its relationship with salts, since dissolved ions greatly increase electric conductivity. This can be observed in the results from ER-GMR (Figure 9) where the highest apparent moisture content is recorded in number 4, where the highest salt concentration was detected, not the highest moisture content as derived from other methods, which was at the foot of the Pillar. Thus, we would not recommend the use of electric resistance devices (both destructive and non-destructive) in areas where one cannot reasonably assume that salts have a negligible presence. Thus, their use in confined archaeological complexes is very limited and, in general, not recommended.

Figure 9. Results of moisture measurement from the Pillar: Moist, Uranine-probe, ER-Scan, and ER-GMR. For uranine-probe measurement, note that when capillary water was not present,a virtual depth of 103 mm has been prescribed for better data visualisation.

To determine the presence of water and thus identify potential salt crystallisation zones, the only low-destructive method involves the use of probes employing a reactive dye (uranine-probe; Weiss et al. 2020). However, it is worth noting that the deployment of these probes in heritage preservation efforts may present challenges, as the current version of the probes may cause staining of the examined material. However, we anticipate that this drawback will be addressed soon (web.natur.cuni.cz/uhigug/moistureprobe).

4.2. Water sources

The hydrological environment surrounding Prague Castle is distinctly urban, heavily influenced by underground structures such as utility line tubes. For example, the water pipelines that provided Prague Castle with drinking water in the past, although not in use, still may allow preferential flow that could affect groundwater in the castle grounds. Additionally, the presence of tunnels built directly into the rock mass poses another unmapped issue. These tunnels contribute to the leaching of the historic environment and often allow water to flow, especially towards the southern part of the Prague Castle ridge.

The sources of water and subsequent dampness of the masonry in the investigated archaeological area probably originate from the paved surface of the third courtyard of the Prague Castle. This has been directly demonstrated by visual observations after heavy rains. The accumulation of water in the vicinity of borehole A comes from the studied drains that drain the sediment beneath the tiles. Incomed water accumulates in this terrain depression, creating a local aquifer in the sedimentary bedrock of the site. Since the permeability of the compacted anthropogenic sediments is low, water flow through the sediments is notably slow. Therefore, this also implies that the relatively rapid responses of the water level in the borehole A and in the pond to precipitation are unlikely to be caused by the propagation of the hydraulic response in the sediment, as it is well explained by preferential pathways from the pavement of the courtyard to the drains. The possibility of another dominant source of water such as seepage from a water supply or sewerage system (that has not yet been located) is also unlikely, as the water level in borehole A steadily decreases during dry periods and the possible system in the vicinity of the drains has been checked for leakage.

The hydrochemical analyses also indicate that the water in all sampling locations has a similar origin, most likely rainwater from the third courtyard. This water enriched with bicarbonate ions dissolves calcite contained in the cement binder and is then enriched with anthropogenic sediments. The concentrations of individual ions are similar in all objects, with a more pronounced difference in the mineralisation of calthemite straw. However, this is clearly due to carbonate precipitation at the bottom of the sampling container. The relatively lower concentrations of most ions in the sample in the trench than in the well tend to indicate a historical origin for the ions measured.

4.3. Salts in masonry

The combination of methods allowed for the monitoring of ions along their likely pathway. Since it can be safely assumed that the present groundwater comes from precipitation at the study site and since sewage leaks are unlikely, we can assume that the journey of water and ion-forming salts begins with atmospheric deposition on the pavement of the third courtyard, be it rainwater, snow, particles, or aerosols (Figure 10). This water then infiltrates through cracks and joints on the tiles in the concrete slab that supports the courtyard floor. Rainwater enriched with bicarbonate ions dissolves calcite contained in the cement binder of the concrete structures, and calcium ions dissociated in water increase the pH of the water. The water then becomes enriched with nitrate and potassium in the anthropogenic sediments. In normal waters, sodium dominates over potassium, which is not the case here, indicating the possibility of historic hearth dissolution, where potassium can be expected to be present in ash in a 100:1 ratio with sodium (Pitter 2009). Another source of potassium and sodium is human or animal urine, which may be of both recent (sewage leaks) and historical origin. Similarly, elevated nitrate, which typically originates from the decomposition of plant or animal remains, can be both recent and historical. The anthropogenic sediments also enriched the water with chlorides (residues of human activity and animal husbandry) and sulphates (oxidation of pyrite, a mineral highly abundant in the bedrock, air pollution). The latter could also come from the brick clay used to produce bricks in perimeter walls (Viani et al. 2018). This mineral-rich solution penetrates the perimeter walls or rises by action in the area of archaeological excavations into the structures.

Figure 10. Graphical representation of the ion transport paths from the third courtyard, through the concrete slab and the archaeological terrains, to the investigated masonry. Photos by Jan Gloc and Studiobarcelona via depositphotos.com.

In the case of the Pillar, we can assume that at its foot, a solution chemically similar to that in borehole A is being taken up by the rise. This solution in the borehole is supersaturated to several phases, and considering crystallisation kinetics, we can expect fractionation of the dissolved salts along the height of the Pillar. According to Arnold (1982) who proposed a fractional crystallisation model, low-soluble salts would crystallise just above the ground (calcium carbonate and gypsum), followed by the most deteriorated zone caused by recrystallisation cycles of moderately soluble salts such as sodium sulphate and gypsum. Above, only highly soluble salts such as nitrates and chlorides are expected. This also corresponds to our findings with carbonates (calcite, aragonite, dolomite) likely crystallising perhaps even below the terrain due to their high saturation indexes in the borehole water (1.5 for calcite in borehole A). Going up the Pillar, the presence of efflorescences and water-soluble salts in the surface layer of the masonry was detected only at a height of about 1 m — the height of the upper edge of the wet area (Figure 11), the wet-dry boundary. Their higher abundance on protrusion parts of the masonry is most likely due to preferential drying of parts that stand out from the surface, since the drying front tends to be smooth (Weiss, Slavík, and Bruthans 2018). Disintegration of the material was observed in the mortar joint at the wet-dry boundary of the Pillar, and also in other parts of the site, the flaking of disintegrated stone due to salinisation and crystallisation of the salts present was observed.

Figure 11. Two main types of water presence: (I) rising damp at the Pillar site and (II) ingress through walls at the perimeter walls. The red highlights show the likely presence of salt efflorescence. Note that the graph shows the boundary between the dry surface layer and the capillary zone and that in cases when no capillary water was recorded, a virtual depth of 103 mm depth has been prescribed for better data visualisation.

The efflorescing sodium sulphate that was identified by XRD in the form of thenardite is likely to form mirabilite in the cold season when the temperature is low and the relative humidity is high (deliquescence humidity at 20°C is 95.6%, Steiger and Asmussen 2008). Scherer (2004) argues that when the anhydrate (thenardite) dissolves under ambient conditions, the resulting solution is highly supersaturated relative to the decahydrate phase (mirabilite) and this phenomenon makes sodium sulphate one of the most destructive salts in nature (Scherer 2004). Doehne (1994) published that the damage is not caused by hydration pressure (which is the pressure that would be required to prevent a 314% increase in molar volume when anhydrate is converted to decahydrate) because thenardite cannot continuously absorb water and expand like a sponge when hydrated; instead, thenardite dissolves and recollapses like mirabilite. Thus, the damage is a consequence of the crystallisation pressure exerted by mirabilite (Flatt and Scherer 2002), and this process operates whenever sodium sulphate-containing material is subjected to wetting and drying cycles. Damage has then been found to occur during the wetting cycle rather than during the drying cycle.

4.4. Moisture in masonry

4.4.1. Spatial layout

The moisture situation in opuka masonry and other building materials shows two main types of water presence, pointing to two forms of long-term water movement: (I) rising damp and (II) ingress through walls (Figure 11). The first type (I) indicates uptake, with the highest moisture content near the present terrain and decreasing the higher the measurement points. This moisture distribution is typical of rising damp in walls and has been well studied and described in the literature on moisture dynamics in masonry (see, e.g., Hall and Hoff 2021). This situation has been observed at most of the studied opuka locations, except for the perimeter walls (Figure 2). The elevated moisture content around the St. Bartholomew church can be well explained by the higher groundwater level that is present within tens of centimetres below/above the surface (as seen in borehole A). The relatively high moisture content on the west wall of the St. Bartholomew church (M8-M12) is also likely caused by moisture that seeps from the perimeter wall side of the site (Figure 9b) and enters the church opuka masonry from below through the ground.

The moisture within masonry around St. Bartholomew church also suggests that the aquifer around the borehole A is localised, extending only a few meters from the borehole: The moisture distribution on the east wall of the church (M13-M19), indicating rise, is considerably lower than at the west wall and the Pillar, even though the measured points are below the level of the west wall and comparable in height to the Pillar site. The occasional rise in moisture content on the west wall during the summer months can be attributed to heavy rainfalls.

4.4.1.1. Rising damp

In the rising damp type of water presence (I), the Pillar site is a typical example. Here, we had groundwater level control and measurements of both periodic and continuous moisture content. Given the consistent presence of liquid water, the highest moisture in masonry was measured at the point closest to the ground. We would expect a semi-continuous decrease in water content with the height, but here we can distinguish quite a sharp line at about 130 cm above the ground between the lower wet part of the Pillar and the higher dry part of the Pillar, which can be clearly seen from the results of Moist and uranine-probe. This observation cannot be explained by the retention curve of the material, as one would expect a higher rise for opuka/sandstone (Van Genuchten 1980; Weiss and Sass 2022). The most likely reason for the sharp wet-dry line is a combination of (a) sharp front theory (Hall and Hoff, 202 1; Ghanbarian-Alavijeh et al. 2010), which sets the rising liquid water height as the result of competition between capillary uptake from the ground and evaporative loss along the exposed surface. And the second reason is likely (b) the heterogeneity of the wall, which can effectively act as a hydraulic barrier. Therefore, we can assume that in similar sites with masonry of variable porosity, a clear wet-dry boundary may be more typical, as in the case of our Pillar.

4.4.1.2. Ingress through walls

The second type of water presence (II) was observed on the mapped perimeter walls and was characterised by an unstructured layout of the moisture content. This observation suggests that the rising damp is not the only or even the dominant water movement and that in this case the moisture comes from undefined locations behind the walls. This interpretation applies to all the perimeter walls, but in the case of the West (Figure 9b) and the North (Figure 9a) walls, capillary uptake also plays a role at least in the redistribution of the moisture. Since seasonal changes in indoor climate have little effect on moisture content (Figure 9), the observed changes in moisture (Figure 9) are likely to be caused by the moisture development behind the walls.

The two types of water presence presented above predefine the locations where salts tend to precipitate. Salts usually crystallise from a supersaturated aqueous solution at the place of increased evaporation (called ‘evaporation front’; Weiss, Slavík, and Bruthans 2018). The position of the evaporation front is a result of the relationship between the rate of liquid water inflow from the inside of the wall and the rate of evaporation (Mls 2022). Therefore, the source of incoming water and the magnitude of the water inflow determine the place of salt crystallisation. In the second (II) type of perimeter walls, the evaporation front is expected to occur mostly on the surface of the wetter parts, hence salt precipitation is likely to occur at the whole wall, depending on the liquid water intake from behind the walls. In the first type of rising damp (I), the evaporation front is expected to occur on the surface at the wet-dry boundary described above; hence, efflorescence is likely to be present (and is present in our case). The rising damp is expected to cause fractionation of the dissolved salts, with the default case being for the salts to be distributed in height according to their solubility (see Gonçalves 2022 and references therein).

4.4.2. Time evolution

Overall, all the materials studied (opuka stone, concrete, terrain, and wood) show little seasonal variation in moisture content. This is also the case for the studied Pillar, contrary to our initial hypothesis of higher seasonal moisture variability in its upper parts due to the relatively high temperature fluctuations. Materials not affected by rising damp contain on average tenths of a weight percent more moisture in summer than in winter. These small variations are likely related to the humidity sorption properties of the materials and the ambient (air) moisture content. The absolute moisture content is higher in the summer months and reaches approximately 15 g/m³, compared to approximately 4 g/m³ in the winter. Relatively stable moisture content was thus observed at the upper points of the walls where rising damp occurred. The moisture content of the lower measurement points affected by rising damp also remained almost constant throughout the measurement period and seasonal variations were negligible. Exceptions were the two lowest measurement points on the Pillar, which can be explained by the effect of the fast capillary uptake after the occasional flooding of the vicinity of the borehole A.

The observations also allow for some interpretation of the long-term moisture-time evolution that extends the measurement period. The study area is partially covered by a leached lime layer, which could be especially observed on the perimeter walls. This shows that in the past there had been a long-term seepage of water over almost the entire area of the perimeter walls, which had since ceased. The current situation and the moisture detected in some areas of the walls indicate to some extent the effectiveness of the repair works carried out to insulate the concrete roof/pavement over the site in the last decade.

5. Suggested solutions

The water seepage poses a significant threat to the structural integrity of architectural elements, particularly masonry, within the archaeological site beneath the Prague Castle’s third courtyard. The observed seepage results not only in occasional extreme wetting, but, more critically, initiates a prolonged transport of anions and cations from archaeological terrains to the masonry. In response, we advocate for three primary preventive measures: (i) avoiding water infiltration, seepage, and ingress; (ii) reducing salt concentrations within the site; and (iii) stabilising the site climate to minimise salt crystallisation transitions.

1. Since the main source of water likely originates on the floor of the third courtyard, it is imperative to block direct leakage through the concrete ceiling by addressing structural vulnerabilities in the tiles and concrete structure. Addressing the cause of the leakage would potentially be the most effective intervention. However, if the causes of leakage cannot be eliminated, then the inflow of water into historic structures should be stopped or at least reduced. The above-ground parts of the masonry are particularly problematic in terms of degradation, where water evaporation and drying out occur. An appropriate solution may be to divert, or even drain, the incoming water. This can be applied, for example, at the location of the Pillar, where a significant amount of water enters via the drains into the terrain and into the masonry. A direct collection of water from the drains or the creation of a collection sump and its automated pumping should help partially reduce the overall moisture load.

2. The systematic removal of salt efflorescence is also a recommended aspect of preserving the archaeological site. These salt crusts on masonry surfaces should be regularly removed to prevent fallout on the ground and recontamination of masonry. Since certain salts can crystallise and revert to an aqueous form depending on relative humidity and temperature, the effectiveness of regular cleaning as a desalination method is crucial. Removal is recommended during periods preceding drops in relative humidity (spring months) and increases in humidity (early autumn). Collaboration with a chemist-technologist is essential for exploring potential chemical interventions in the mechanical removal process.

3. To prevent damage to the masonry caused by the crystallisation of salts already present, a strategic approach involves minimising wetting/drying cycles and decreasing the drying rates at the site. Especially, stabilisation of the site’s microclimate is paramount to minimise the year-round transition of sodium sulphate from its anhydrous form (thenardite) to decahydrate (mirabilite) — see, e.g., Tsui, Flatt, and Scherer (2003) for details. Furthermore, ventilation is essential in areas prone to condensation on the ceiling slab during winter due to microclimatic conditions. However, careful consideration is required when designing new vents to mitigate potential risks to masonry, since air passage can accelerate material evaporation and shift crystallisation pressures beneath the surface, posing a greater threat to materials than surface crystallisation.

6. Conclusions

Archaeological complexes are often exposed to various deterioration processes, such as salt or frost weathering. Sheltering is intended to protect them from weather conditions, although it can paradoxically lead to increased degradation. Therefore, it is crucial to understand the underlying processes that lead to the degradation and to implement conservation strategies that consider the potential negative effects of sheltering. An example of a protected cultural heritage site is the archaeological excavation under the third courtyard of the Prague Castle, where there is a persistent problem of material deterioration, increased moisture content in the walls, and noticeable salt efflorescence. In this article, we investigate the origin of the water and salts, observe their spatial and temporal variability, and finally propose a suitable solution for the conservation of the investigated site.

The dominant source of water for the wet parts of the masonry in the archaeological area is rainwater. This water seeps through the paved surface of the third courtyard, flows through the concrete slab and the archaeological terrain, and subsequently enters the masonry walls in two ways. The first is the accumulation of water in a local aquifer suspended in low-permeability sediments, from which it rises into the masonry by capillary action. The second way is ingress through walls that are in direct contact with the anthropogenic sediments.

As rainwater flows through the concrete structure, it is enriched with bicarbonate ions and calcium, which increases its pH (up to 11). In the anthropogenic sediments, the water is then enriched in nitrate and potassium (indicating the possibility of historic hearth dissolution), chloride (residues of human activity and animal husbandry), and sulphate (pyrite oxidation and air pollution). This mineral-rich solution then penetrates in two ways: by ingress (perimeter wall sites) or rises by capillary action (Pillar site). At the perimeter wall sites, we observed an unstructured moisture layout, indicating that the moisture comes from undefined locations behind the walls. In the latter case, the moisture values are generally relatively lower at high measurement points above the ground with a sharp drop in the moisture content at a height of 130 cm. Here, the solution undergoes fractional crystallisation, and the wet-dry boundary on the surface leads to efflorescence, particularly of sodium sulphate, forming potentially highly damaging salts.

The multidisciplinary approach also allowed evaluation of the methods used to measure moisture in historic masonry and their advantages and disadvantages. In summary, the microwave method (high frequency electromagnetic waves) appears to be the most suitable, while the electrical resistance devices have complications with low contact between the probe and the material to be measured, as well as a false increase in material moisture due to the signal from dissolved salts. The uranine-probe method proved to be suitable for detecting capillary water but is of limited use in heritage conservation because it causes staining of the material being tested.

In response to the challenges of water ingress and salt accumulation studied, we propose three fundamental preventive measures: (i) averting water intrusion, seepage, and ingress; (ii) reducing salt concentrations within the site; and (iii) stabilising the site climate to mitigate transitions in salt crystallisation. These recommendations apply not only to the investigated site but should also be considered in other confined archaeological complexes.

Acknowledgments

The authors express their profound gratitude to Jana Maříková-Kubková and Iva Herichová, along with their colleagues at the Institute of Archaeology of the Czech Academy of Sciences, for their hospitality at Prague Castle and their invaluable support and collaboration. The authors also wish to acknowledge the significant contribution of Kateřina Kalianková and Kristýna Kotková in conducting some of the measurements. This work was supported by the Technology Agency of the Czech Republic under Grant TL03000603 and supported by the Center for Geosphere Dynamics [UNCE/SCI/006].

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was supported by the Technology Agency of the Czech Republic under Grant [TL03000603].