Influence of Weather, Soil Variability, and Vegetation on Seasonal Ground Movement: A Field Study

Abstract

Expansive soils change volume due to changes in moisture content, which results in ground movement. Lightweight shallow-depth structures such as pavements, residential footings, and pipelines can suffer distress as they face additional stresses from the moving ground. The soil reactivity (the ability of soil to expand or contract due to change in moisture content) and the interactions at the soil-atmosphere-vegetation boundary are two of the major contributing factors to the ground movement. The current methodology used in design to account for ground movement is based on limited field and laboratory observations. Aiming at a better understanding of the interaction at the soil-atmosphere-vegetation boundary and its correlation with ground movement, this paper presents results from a field study. The research site was located in a semi-arid climate area and consisted of moderately to highly reactive clay soils. Part of the site was vegetated with mature trees, and part was occupied by grass. The order of 10 s of mm difference was observed in the magnitude of movement across the site owing to site soil variability as well as differences in vegetation. The areas with large trees showed relatively lower ground movement over the study period compared to the area occupied by grass, which was attributed to the microclimate created by the tree canopy and the already established deeper drying of the soil near the trees.

1. Introduction

Nearly 30% of all surface soils in Australia can be classified as expansive [1,2]. These soils are also common in many other places in the world [2], including but not limited to the United States, Brazil, South Africa, India, China, Spain, Sudan and Greece [3]. Expansive soils shrink or swell due to changes in moisture content between dry and wet seasons and result in ground movement [4,5]. Such movements can apply substantial additional stresses to lightweight shallow-depth structures such as one or two-story residential building footings, pavements, water mains, sewer lines, stormwater drainage, and other underground utilities. It is challenging to design and construct on such soils. Maintaining assets is also challenging as asset performance degradation is directly related to ground movement driven by seasonal moisture changes (a function of local climate). Expansive soil-related damages to structures and infrastructure can exceed billions of dollars every year [6]. For example, the damage exceeded USD 15 billion/year in the United States of America alone in the year 2012 [7]. In Australia, more than 80% of the annual housing insurance claims have been attributed to ground movement [8]. The situation is expected to worsen due to climate change [1].

Many studies can be found in the literature that investigated the changes in stresses and stiffness of soil due to interactions that happen at the soil-vegetation-atmosphere interface [9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25]. Most of these studies focused on the pore water pressure cycles and related changes in soil strength and stiffness. Soil volume change and related ground movement have been paid limited attention.

Efforts have been made in the past to quantify the effect of expansive soil movement on structures and accommodate it in design since the 1960s [4,26,27,28,29]. Mitchell [28] showed that it is possible to account for expansive soil movement in design using empirical methods without heavily relying on experience. Later works [1,30,31,32,33,34] highlighted the importance of better understanding the influence of atmospheric boundary interaction and changes in climate on the behaviour of shallow-depth structures on/in expansive soils. Several previous works have related climate with depth of suction change, e.g., Hu et al. [35], Davenport [36], Walsh et al. [37] and Fityus et al. [38] generated depth of suction change (H_s) databases for different areas of Australia. Li and Guo [39] investigated a residential building site in Australia affected by seasonal ground movement combined with tree root desiccation and observed that tree roots can extract moisture from depths greater than H_s and magnify the ground movement and related stresses on the footing. Karunarathne [40] observed over a 2-year period that significant contribution to seasonal ground movement did not occur below a depth of 1.25 m in an area with a mild temperate climate. He also correlated monthly rainfall to ground movement. Similarly, Fernandes et al. [41] observed a correlation between meteorological events and ground movement. Karunarathne et al. [42] and Teodosio et al. [43] demonstrated the capabilities of a numerical approach to capture the important aspects of the soil-atmospheric boundary interaction problems in expansive soil areas. Despite these efforts, there are still uncertainties related to assessing/quantifying the effect of soil-vegetation-atmosphere boundary interaction on ground movement, and there are limited field studies available in the literature.

Many of the design guidelines in Australia, for example, embodied in AS2870 [44] and around the world, are developed based on personal experiences and limited field observations. Additional field studies are required for an improved understanding of the correlations among different factors contributing to ground movement potential.

This paper discusses a field investigation over three years to improve understanding of the correlation between seasonal ground movement and soil-vegetation-atmosphere boundary interactions in an area with a semi-arid climate. The site consists of unsaturated clays of high reactivity. Part of the research site is vegetated by grass, and other parts consist of mature trees.

A detailed geotechnical investigation was carried out, which included evaluation of shrink-swell indices, site classification, development of suction profiles and testing of electrical conductivity of soil solutions. Saturated hydraulic conductivity was determined in the laboratory and in the field, which has not been reported previously for expansive soils. Ground movement surveys were conducted at regular intervals across the site, and weather data were recorded using a weather station at a nearby location. A demonstration of seasonal ground movement estimation is provided, and evidence of the drying influence of trees is presented in Section 4 of the paper. Solute suction is shown to be significant in these saline soils. A comparison between available water for the soil and ground movement well away from trees highlighted the importance of the soil-atmosphere boundary interaction in modelling ground movements. The difficulties associated with estimating appropriate soil profiles and other properties and their influence on ground movement are also discussed.

2. Site Description

The site is in a suburb which lies 7.6 km south of Adelaide CBD in South Australia. The site is bounded by native shrubs and trees in a bushland setting (see Figure 1a) and relies on natural rainfall rather than irrigation. An old creek bed has been buried (since 1972), and stormwater is diverted to the drainage system around the park [45]. Dots in the Figure represent the near-surface ground movement monitoring locations placed at 8 m spacing on a square grid. The points are named A to H in East–West direction and 1-8 in the North–South direction. A1 in the lower left-hand (or South-West) corner, to H8, is the diagonally opposite North-East corner.

Shrinkage cracks are observed across the site during summer and autumn (Figure 1b). Grass cover is almost non-existent in this dry period. During winter and early spring, the surface is difficult to walk on without black mud from the clayey black earth (BE) sticking to footwear. The mean annual rainfall for Adelaide is 527 mm. A weather station close to Adelaide and approximately 9.2 km North to North-East of the site recorded annual rainfall totals of 471, 499 and 612 mm for 2020, 2021 and 2022, respectively. For comparison, the average rainfall in South Australia for the year 2022 was 310 mm, which was nearly 40% more than the last 120-year average [46].

Kikuyu grass and weeds inhabit the open area of the site. Extensive planting of a 20 m wide strip of the perimeter of the reserve was undertaken in the winter of 2007 to supplement approximately 20 established trees located mainly near the Northern and Western boundaries. The existing trees were River red gum, South Australian blue gum (Eucalyptus leucoxylon) and Lemon-scented gum (Corymbia citriodora). Newer plantings included Grey box, South Australian blue gum, Drooping sheoak (Allocauarina verticillata) and Southern cypress pine (Callitris preissii). The heights of vegetation varied over time. Lidar surveys of canopies in 2019 [47] indicated the trees on the southern border had canopy heights 6 to 11 m towards the front of a row of 10 to 14 m high trees; the western canopy height ranged between 8 and 19 m, and the eastern canopy height was 5 to 11 m.

Figure 1. (a) Plan view of the research site (highlighted by orange box) from Google Earth [48] on 7th August 2022: dots represent near-surface levelling points spaced 8 m apart, which are designated A1 in the lower left-hand (or South–West) corner, to H8 in the diagonally opposite North-East corner; BH = borehole (b) Shrinkage cracking observed in the summer of 2022.

Figure 1. (a) Plan view of the research site (highlighted by orange box) from Google Earth [48] on 7th August 2022: dots represent near-surface levelling points spaced 8 m apart, which are designated A1 in the lower left-hand (or South–West) corner, to H8 in the diagonally opposite North-East corner; BH = borehole (b) Shrinkage cracking observed in the summer of 2022.

Sheard and Bowman [49] performed a major undertaking to assess the soils and soil profiles of metropolitan Adelaide and the shrink-swell ability (herein referred to as reactivity) of the various soil types encountered. One borehole (BH) in the vicinity of the research site was reported [49], and it was located in the adjacent reserve, which is just above the top left-hand corner of Figure 1a. The first 3 m of the soil profile consisted of the Pooraka formation, which was described as red-brown earth soils which can be gravelly, and the clays may be impregnated with carbonate (i.e., calcareous). A layer of highly plastic clay was encountered between depths of 0.4 and 0.6 m. Below the red-brown earth, a transitional gley clay with occasional gravel clasts was observed extending to the exploration depth of 9.8 m. The plasticity of the clay in this layer varied, as did the colour. At 4 m depth, the clay was reported to be of intermediate plasticity (CI) but changed at greater depths to high plasticity clay/silt (CH-MH).

Soil samples were taken from 5 BHs in March and September 2019 and again from 3 BHs in June 2021. The March 2019 sampling was planned primarily to check soil profile variability and to assign site classification based on visual and tactile examination. This work was performed by an independent site classifier. The locations of the BHs for the September 2019 sampling are indicated in Figure 1 (BH 1, 2 and 3). The three BHs coincided approximately with the first three BHs of the earlier sampling, i.e., central (near BH2), NE (near BH1) and SW (near BH3). The remaining two BHs were in the SE (near BH4) and NW (near BH5) corners of the site. Descriptions of the soil horizons are provided in

Table 1. Soil horizons from a site classifier using the visual-tactile approach across five boreholes.

Horizon	Pedology	Colour	Inclusions	USCS
1	Silt and Sand, clayey	Grey brown	Some fine roots	SC-SM
2	Clay	Grey brown to black	Trace of sand	CH
3	Clay	Grey brown orange	Trace of sand	CH
4	Clay, sandy silty	Grey brown yellow, cream in pockets	Angular gravel to 10 mm, highly calcareous	CI-CH
5	Clay, silty sandy	Grey orange yellow brown.	Moderately calcareous	CH
6	Clay, sandy silty	Grey brown yellow, cream in pockets	Angular gravel up to 10 mm	CI-CH
7	Clay	Grey brown orange, mottled green	Trace of sand, angular gravel to 5 mm	CH
8	Clay	Grey green mottled orange yellow	Trace of sand	CH
9	Clay, sandy silty	Grey brown yellow cream	Angular gravel to 10 mm, highly calcareous	CH

. Not all nine horizons were encountered in each profile, as indicated in Figure 2. The central profile captured just the first five horizons. Horizon 8 was only observed in BH1. Gravel up to 10 mm in size was encountered in horizons 4, 6 and 9, and smaller gravel was present in horizon 7. Horizons 4 and 9 were highly calcareous.

Soil reactivities were assigned based on visual-tactile examinations of core samples, which is a common practice in Australia for footing design on reactive clay sites [50]. To ensure the assessments of reactivities are reasonable and remains consistent over time and between assessor, laboratory tests are required to be conducted at regular intervals (one in every 50 tests). Essentially, soil plasticity is evaluated, and some appreciation of non-reactive inclusions is made. The black earth of horizon 2 was assigned a shrinkage index of 3.5% strain/pF, and deeper clays (horizons 7 and 8) were assigned 4.5% and 6.0 %/pF, respectively. The units of reactivity are a vertical strain for a unit change in suction. The logarithmic pF unit is commonly used in reactive soil studies. It is defined as: pF = log(cm of negative head) = log(suction in kPa) + 1.01.

With the assigned reactivities, characteristic site movements (y_s) based on the depth of suction change of 4 m. Estimates were 68, 70, 75, 65 and 50 mm for BHs 1, 2, 3, 4 and 5, respectively. The site would be classified as H2-D (highly reactive—deep-seated movements) as per AS2870 [44], although the maximum estimate of 75 mm is bordering between highly and extremely reactive site classification. This site classification is for a “normal” site, without consideration of nearby trees [50].

When trees are present on a house site, the Australian standard for the design and construction of footings [44] has guidelines for adding extra suction change to depth, H_t, which is greater than or at least equal to the depth of suction change (H_s). The knowledge base, however, is quite poor, and little is known about the water demand of the almost infinite variety of species in various climate regimes.

3. Laboratory Investigation

3.1. Atterberg Limits

Further description of the soil profile and its variability was afforded in June 2020 from inspection of the good quality samples that were taken for laboratory testing. Atterberg limits for these samples are presented in

Table 2. Atterberg limits and soil descriptions. USCS is the Unified Soil Classification System.

BH	Depth (m)	Liquid Limit (%)	Plastic Limit (%)	USCS	Soil Description
N	0.33	47	26	CI	CLAY, tan/yellow-brown, CaCO3 patches, moist
N	0.85	60	29	CH	CLAY, tan/yellow-brown, CaCO3 patches, some gravel to 5 mm mm
N	1.15	52	24	CH	CLAY, red-brown
N	2.1	43	21	CI	CLAY, reddish brown to yellow brown, occ. CaCO3 pocket
N	3.2	47	23	CI	CLAY, with fine gravels and occ. large stone (10 to 20 mm), grey brown, mottled yellow, CaCO3 pockets
S	0.4	23	13	CL	Black Earth, very dry, powdery/crumbly.
S	0.8	47	24	CI	CLAY, tan, CaCO3 patches, stiff and dry
S	1.3	55	24	CH	CLAY, red brown, very stiff, small pockets of CaCO3
S	2.1	42.5	20	CI	CLAY, red brown, minor CaCO3
S	3.1	45	24	CI	CLAY, silty, brown/yellow brown with CaCO3 patches and minor gravel

for BH North and BH South, along with soil descriptions. The more plastic CH clays, including the red-brown soil, lie less than 1 m below the surface and extend to 1.5 m or greater. Intermediate plasticity clays underlying the red-brown clay can be sandy and/or silty with inclusions of lime and some gravel. However, it should be noted that the Atterberg limits of the CI and CH soils did not differ greatly. Nevertheless, soil profiles vary across the site, as was reported in the exploration of the site in March 2019. Black earth was not evident in BH North, and the near-surface sample from BH South was less plastic than expected.

3.2. Shrink Swell Indices

Samples were taken with 50 mm Shelby push tubes on two occasions in 2019 (March and September) to determine the shrink-swell indices (I_ss) to study the influence of seasonal suction differences across the site. Boreholes BH1 and BH2 provided the samples in September 2019, while the samples in March 2019 were taken near these same locations. Shrink-swell tests were conducted following AS1289.7.1.1 [50]. A summary of the shrink-swell data is presented in

Table 3. Shrink-swell indices.

BH	Soil Type	Depth (m)	Water Content (%)	Initial Suction, s (pF)	Swell Strain (%)	Shrink Strain (%)	Iss (%/pF)	Mean Iss
1	BE	0.35	30.9	3.41	0.2	8.2	4.6	4.7
1	BE	0.3	16.5	4.93	11.4	2.4	4.5
2	BE	0.7	32.6	3.91	0.7	8.4	4.9
1	RB	2.0	23.8	3.66	1.3	5.0	3.1	2.6
2	RB	1.9	21.6	4.20	0.5	4.3	2.5
2	RB	1.9	21.5	4.24	1.8	3.8	2.6
2	RB	0.75	15.1	4.68	3.0	2.2	2.1
1	Other	2.2	24.5	4.01	0.1	3.2	1.8	2.0
1	Other	3.0	19.5	3.87	0.2	2.1	1.2
1	Other	3.25	25.0	3.85	0.3	4.4	2.5
2	Other	2.95	21.8	3.90	0.2	3.9	2.2
2	Other	3.0	21.9	4.06	0.9	3.7	2.3

, and the shrink-swell indices have the unit % vertical strain for a pF change in suction, although suction is not measured in this test. The shrink-swell index is simply the free vertical strain divided by an assumed suction change of 1.8 pF. Free or laterally unrestrained vertical strain is the sum of the magnitudes of half the swelling strain from the oedometer swell test and the shrinkage strain to an oven dry state.

In Table 3, the data have been arranged according to soil type. Initial total suctions were measured with a Wescor dew point meter and PT-55 sample chambers following AS1289.2.2.1 [51]. Although just three samples of black earth were available for testing, the repeatability of shrink-swell indices was very good, given the differences in initial total suction (3.41 to 4.93 pF or a difference of 1.5 pF or ~8 MPa). The shrink-swell indices of the black earth varied between 4.5 and 4.9 %/pF. As expected, differences in suction between the two sampling dates decreased with depth; red-brown soil samples differed by up to 1.0 pF, while a much smaller difference of 0.2 pF was observed for the deeper soils (listed as “Other” in Table 3). Given the differences in inclusions between samples, particularly for the deeper soil samples, a high degree of variation might be expected. The average value for the deeper soils was 2.0 %/pF with a standard deviation of 0.5.

Considering the means and standard deviations, design values of I_ss could be taken as 4.9 %/pF for the black earth, 3.0 %/pF for the red-brown clay and 2.5 %/pF for the remaining soils to a depth of 3 m.

3.3. Soil Salinity and Osmotic Suction

Soil salinity was assessed on samples taken from the North and South boreholes. Electrical conductivity (EC) was measured on soil solutions prepared at a moisture content of 500% (EC_1:5). The results of the EC testing are presented in

Table 4. Soil salinity.

BH	Depth (m)	*EC1:5 (mS/m)	Level of Salinity [54]
N	0.33	47	Slightly saline
N	0.85	93	Highly saline
N	1.15	160	Highly saline
N	2.1	178	Highly saline
N	3.1	102	Highly/moderately saline
N	3.3	100	Moderately saline
S	0.4	42	Slightly saline
S	0.8	175	Highly saline
S	1.3	214	Severely saline
S	2.1	166	Highly saline
S	3.1	215	Severely saline

*EC_1:5 is the electrical conductivity of soil solution measured at 500% water content.

. The salinity level was determined according to the Western Australian Department of Primary Industries and Regional Development guidelines [52]. Salinity levels were generally high at both borehole locations between 0.8 and 3 m depth. Levels reached “severely saline” in BH South at 1.3 and 3.1 m depth. Salinity affects total suction through osmosis; total suction is regarded as having two components, matric and osmotic, or solute suction (ψ_s). The level of solute suction is dependent on the moisture content and the dominant salts within the soil. Various empirical estimates of osmotic suction exist [53]. Equation (1) from Mata et al. [54] has been applied herein to estimate osmotic suction at field moisture content in kPa:

ψ_s = 0.019(EC_fmc)^1.074

(1)

where EC_fmc is the electrical conductivity at field moisture content in μS/cm.

This equation is approximate as it assumes an average valence of common salts in soil. A linear extrapolation of EC_1:5 to the EC value at field moisture content was made. It has been found that the back extrapolation can overestimate the EC of pore water by 18 to 21% [55] Consequently, the EC_1:5 values have been reduced by 20% before applying Equation (1). The resultant solute suctions are presented together with total suction results later in Section 4.2.

3.4. Hydraulic Conductivity

Saturated hydraulic conductivity (K_sat) is commonly used as the starting point from which the hydraulic conductivity reduces as the matric suction of soil increases and the soil becomes unsaturated [56]. The K_sat of laboratory soil samples (K_sat-lab) was determined using a Rowe and Barden cell [56,57]. Hydraulic conductivity was determined in the field as well (K_sat-field) using a Guelph permeameter. Laboratory tests were performed on each sample at different overburden pressures and at back pressures, usually limited to 35% of the applied vertical pressure. A minimum vertical pressure of 25 kPa was applied. Variations of K_sat-lab with changes in applied pressure were relatively small, and a single value was chosen for plotting. K_sat-lab was determined for both vertical flow and horizontal flow conditions. For the horizontal flow condition, a central hole of 10 mm diameter was drilled carefully. Filter paper lined the hole, and dry granular soil particles, 1 to 2 mm in diameter, were placed and tamped inside to form the vertical drain.

The results of the vertical and horizontal K_sat-lab tests are presented in Figure 3a as a plot against the depth of sampling. The results of the laboratory tests were indicative of low plasticity clays generally, falling within the range of 3 × 10⁻⁰⁹ to 2 × 10⁻¹⁰ m/s. In terms of vertical K_sat-lab, the most permeable soil (1.8 × 10⁻⁰⁹ m/s) was the clay of intermediate plasticity at 2.1 m depth from the North BH, and the least permeable (2.1 × 10⁻¹⁰ m/s) was the highly plastic red-brown clay at 1.3 m deep from the South BH. With respect to horizontal K_sat-lab, the most permeable soil (3 × 10⁻⁹ m/s) was the Black earth soil, just below the surface in the South borehole, while the least permeable soil (2.0 × 10⁻¹⁰ m/s) came from 1.35 m depth in the same borehole, from the highly plastic red-brown clay. The Black earth from the South BH was classified as low-plasticity clay (see Table 2).

Horizontal K_sat-lab matched reasonably well the vertical hydraulic conductivity for the samples from BH North (refer to Figure 3b). The ratio of vertical to horizontal K_sat-lab ranged between 0.8 and 1.7, with an average of 1.3. The samples from BH South had ratios of ~1 between depths of 0.8 and 1.4 m, but horizontal K_sat-lab exceeded vertical values at three other depths. The ratio of vertical to horizontal permeability ranged between 0.2 and 1.1 for BH South, with an average of 0.5.

“Field saturated” hydraulic conductivity (K_sat-field) was determined with a Guelph permeameter for comparison with laboratory tests. This device measures the combined effect of water flow in both the horizontal and the vertical directions, although only relatively shallow testing is possible with the Guelph permeameter. The results of the field tests are presented in Section 5.

4. Atmospheric Boundary, Vegetation, Suction Change and Ground Movement

4.1. Weather

A weather station was installed and operated since 1 August 2021. The weather station was located ~900 m South-West of the research site. Recordings were made at five-minute intervals. Recorded parameters included temperature, relative humidity, relative air pressure, wind speed and its direction, rainfall and solar radiation. A plot of daily rainfall is presented in Figure 4. The yearly rainfall for 2022 was 760 mm, which is significantly higher than average. A further 394 mm of rain had been recorded in 2023 by 28 August 2023. The maximum daily rainfall was 67 mm on 22 June 2023. Heavy and persistent rain events are generally expected over winter and spring (June to November). Rain was less frequent and intense over the summer.

The daily reference Potential Evapotranspiration (PET) was estimated from the weather data for clipped grass using the method prescribed by ASCE-EWRI [58]. PET reached a peak of ~11 mm/day over the summer months and dropped to a low of 1 mm/day at the start of winter (see Figure 5).

4.2. Suction-Depth Profiles

Total suction (s) profiles for September 2019 were developed from borehole samples and are presented in Figure 6. The suction profiles indicate shallow wetting; BH1 was wetter than BH2 to about a depth of 3 m. Suction values at greater depths were similar. The deep equilibrium suction below 2.5 m for this site can be interpreted as being 800 kPa or 3.9 pF according to the suction profiles for BH1 and 2. Suctions at BH3 present a different picture. Being close to the tree line suctions at depth are consistently higher than either those from BH1 or BH2. Below 3.5 m, the difference is approximately 0.24 pF all the way to 6 m depth. This is also the depth and trend indicated by Cameron [59] and Cameron and Beal [60] due to groups of trees in the prevailing climate in Adelaide. Soil suction is largely constant in the zone below normal drying activity and corresponds to the wilting point of the trees. The depth of drying in Adelaide may approach 6 m for trees in competition for moisture [60]. The difference in total suction observed at this site is towards the lower limit of the range reported previously for South Australia of 0.23 to 0.4 pF.

Suction profiles have been determined subsequently at other locations. A suction profile for June 2021 is presented in Figure 7 and indicates drier conditions in BH South relative to BH North. Visualizing the extrapolation of the PET diagram of Figure 5, June 2021 was not the peak of the wet period, and water infiltration can be slow in clay-dominant soil profiles. BH North, at this time, had experienced wetting to a depth of approximately 1 m, but levels of suction between 1 and 3 m indicated seasonal drying had not been recovered. Solute suctions have also been plotted for these two locations, which were estimated from electrical conductivity data, as explained earlier in Section 3.3. The solute suctions are significant at this site. The total suction from the North BH in the top metre and at depth was not much greater than the solute suctions, indicating near saturation of the soil. The average difference in suctions was 0.13 pF with a range of 0.08 to 0.15 pF. The difference in total and solute suction for the South BH varied from a maximum of 0.29 pF at 0.8 m depth to a minimum of 0.14 pF at 3 m. The average difference was 0.22 pF.

4.3. Ground Movements

Near-surface movements across the site relative to a 6 m deep BM have been monitored every 2 to 3 months since November 2019 by licensed surveyors using a Leica digital level. Sixty-four concrete pads buried 0.35 m below the soil surface and covered with an irrigation valve box were established on a grid of 8 m spacing over an area of 3100 m². The locations of the pads are shown in Figure 1. Concrete pads were 0.1 mm thick and 0.4 m in diameter and were anchored to the soil with a vertically placed 16 mm diameter bar, 400 mm long, set in the concrete and extending 0.15 m below the pad into the soil. The top of the bar served as the level survey point on the grid. The research site is an open field and undergoes regular mowing and other maintenance activities, and so council vehicles frequent the area. The concrete pads helped provide a reasonably stable movement measurement location and were not significantly affected by the movement of maintenance vehicles.

Movements have been highly variable across the research site, as shown in Figure 8. The maximum movement at any pad since the start of the level surveys has varied between 10 mm (pad B5) and 50 mm (pad E7). Twelve pads experienced differential movement greater than 40 mm, most notably within the western tree canopy (pads B1, D1, E1 and F1) and away from any trees at pads E5 to E7, C6, D6 and H6 to H7. A plot of the movement against time for B5 and E7 is provided in Figure 9. An additional point closer to the trees, A2, is also depicted; the maximum difference in movement at A2 was 40 mm. The patterns of movement indicate greater wetting and drying in 2021/2022 than in the previous cycle. The maximum seasonal movement was observed at E7 between late autumn of 2022 and near the end of winter in 2023.

Average movements along the canopy lines are provided in Figure 10 for points A1 to A8 and A1 to H1. The movements are compared with the average movement of points D5 to H5, which are well away from the trees. Away from the trees, seasonal movements were significantly greater; near the tree line, the established vegetation had already dried the soils, and the vegetation canopy afforded some protection from extreme seasonal wetting and drying of the near-surface soils, which was also observed by Sun et al. [61] in a field experiment in an outer Melbourne suburb in the State of Victoria.

Across the site, the difference in movement between pads at a given point in time was striking, reaching a maximum of 73 mm between pads G1 and G8 (maximum to minimum), which was recorded on 18 March 2022. G1 was located inside the canopy line, while G8 sat outside it. G1 had heaved 35 mm, while G8 had settled 38 mm on this same day. The variation in ground movements is likely to be due to the differences in initial moisture content/suction, variability of soil profiles (Figure 2) and difference in microclimate at the site.

As BH1 in September 2019 was close to level point G8, where a maximum differential movement of 41 mm had been recorded between August 2021 and March 2022. The total suction profile was determined near this point on samples taken on 30 March 2023. The two suction profiles are shown in Figure 11 below, which indicate seasonal activity to a depth of 1.5 m. Estimates of ground movement were made using these suction differences. The design values of I_ss suggested in an earlier section were adopted. The adopted soil profile was a compromise between the soil profiles determined from the two sampling dates, with black earth extending to 0.6 m and the red-brown earth starting at 1.4 m depth. The soil layer between the black-earth and the red-brown earth was assigned an I_ss of 2.0 %/pF. The estimated movement at 0.35 m depth over the time period shown in the suction plots was 21 mm. As the level surveys had not begun in September 2019, the level surveys cannot be used directly; however, looking at a similar period for BH1 in 2020, the movement of G8 was 25 mm.

Point G8 lies approximately 9 m from the eastern canopy line and 10.2 m from the nearest tree trunk of a group of trees with canopy heights averaging 9 m. Allowing 1.2 m of trunk height, the distance of separation between the levelling pad and the tree group relative to the average tree height is 10.2/10.2 or D/HT = 1. At this distance of separation, the Australian standard guidelines [44] would suggest a reduction of the maximum drying movement to just 50% of the estimated value. However, the suction profiles do not indicate any deep drying commensurate with that seen from BH3 in Figure 6.

4.4. The Influence of Available Water on Ground Movement

Available water for soil replenishment is simply the difference between rainfall and PET, less run-off. To allow for run-off, daily rainfall was held to a maximum of 15 mm, the assumed daily capacity for soil infiltration [40]. At this site, rainfall exceeded 15 mm on just 18 of the 760-day-long monitoring period. In Figure 12, the monthly available water has been plotted against time. The average movement of the central levelling pads along column 5 (D5 to H5) has been plotted in this Figure using a secondary axis. The similarity of the shapes of the plots confirms the strong relationship between water availability and ground movement. There is a noticeable lag between available water and ground movement, which has been observed in other research [40,62]. The time lag can be explained by the slow infiltration into uncracked soil and the sealing of shrinkage cracks during the wet seasons. As illustrated by Figure 11, significant seasonal soil movement is generated by changing the state of suction to at least a depth of 1.5 m at this site. From the plots, the time difference (delay between peaks of monthly available water and peaks in movement) approaches 2 months. A linear relationship was attempted between available water 60 days prior to ground movement measurement and observed ground movement, which required interpolation of weather data. The coefficient of determination of the regression (or R²) was 0.93, indicating a strong correlation. Further data are needed, however.

5. Hydraulic Conductivity Laboratory vs. Field Observations

Hydraulic conductivity is needed to better understand the reactive soil movements in light of developments in unsaturated soil mechanics. Laboratory determinations on the intermediate and high plasticity clays of this site from a depth of about 3 m produced typical values of hydraulic conductivity, which fell in the lower part of the expected range of 1 × 10⁻¹⁰ to 1 × 10⁻⁷ m/s [63]. However, a great disparity was revealed between field and laboratory-determined hydraulic conductivity values (see Figure 13 for field results near BHs South and North). Two rounds of field testing were conducted over autumn (1) and winter (2). The field testing gave values orders of magnitude higher than the laboratory-based values. Field testing does not guarantee soil saturation, and the term “field saturated” was invoked by [49].

A United Kingdom study involving intermediate and high plasticity clays in embankments and in cuttings [64] also found wide-ranging field values of hydraulic conductivity, which were higher than expected, and some limited evidence of disparity between field and laboratory values. Shallow testing of a cutting in London clay (CH) with a Guelph permeameter saw variations of approximately 3 × 10⁻¹¹ to 4.5 × 10⁻⁸ m/s over just three tests. The depth of testing was 0.2 to 0.6 m. Consistently greater conductivities were recorded using a double-ring infiltrometer, which the authors put down to the considerably greater wetting zone in this test and the greater likelihood of encountering defects such as fissures and more permeable seams in the soil. Laboratory determination of hydraulic conductivity on 38 mm diameter triaxial samples of the London clay from depths of 0.4 to 2.9 m averaged about 1 × 10⁻¹⁰ m/s, and conductivity decreased gradually with depth. In other test sites, variations of K_sat-field up to 5 orders of magnitude were observed between different field test methods. Some of the reasons for this variability included spatial soil variability, differences in soil macrostructure, and wetting initially unsaturated soils, which may or may not have bypass zones.

In the case of the field site in this paper, drying shrinkage cracks in the reactive soils are an obvious macrostructure feature that affects hydraulic conductivity [65], as are slickensides. Theoretical accounting of soil cracking on hydraulic conductivity has been proposed by Fredlund et al. [66]. However, knowledge of soil cracking patterns and variations of crack widths with depths is required to estimate the volume of cracking within the affected soil profile. A study of crack patterns at the site and crack depths may be undertaken in the future.

6. Soil Reactivity—Visual Tactile vs. Laboratory and Effect of the Initial Moisture State

Shrink-swell indices in the laboratory tests differed from those predicted in the initial visual and tactile assessment. The black earth had a minimum I_ss from 3 tests of 4.5 %/pF, which was considerably higher than the assigned value of 3.5 %/pF. The estimate for the red-brown earth (described as grey-brown orange) was within the upper end of the tested range at 3.0 %/pF. The estimates of I_ss for the deeper soils were reasonably close (2 to 2.5 %/pF) except for BH1 at 3 m and 3.25 m depth, which should correspond to horizons 7 and 8, respectively, according to Figure 2. The specimen at 3 m yielded 1.2 %/pF, compared with the estimated value of 4.5 %/pF for horizon 7. The specimen from 3.25 m depth yielded 2.5 %/pF, compared with the estimated value of 6 %/pF for horizon 8. However, the soil description of the shrink-swell specimens was a little different than that described by the site classifier and may not have been representative of horizons 7 and 8. The shrink-swell sample from 3 m depth was described as clay, sandy, brown and orange mottled, with some gravel up to 12 mm in size and calcareous pockets. The subsequent sampling at much the same location and 3.25 m depth was similarly described but as slightly sandy clay, calcareous and containing some small gravel. From these two descriptions, the horizons that were sampled were more likely 6 and 7, which were assigned reactivities of 2.0 and 4.5 %/pF.

In summary, the visual-tactile approach requires long-term local experience and calibration with occasional laboratory testing as required by [50]. This exercise has reinforced the spatial variability of the soil profile across this site and the difficulty in assigning an assured site classification. Given the underestimation of the black earth, a site classification of E-D (extremely reactive—deep-seated movements) would apply. It should be noted that for subdivisions in greater Adelaide, the highest level of classification applies from the visual-tactile examination of continuous samples from 3 boreholes [44].

Testing of the CH black earth samples at significantly different initial moisture contents and suctions showed little impact on I_ss values. The initial total suctions were 4.93 pF in autumn, which reduced to 3.41 pF in spring at this same location, a difference of 1.5 pF. The autumn sample swelled appreciably (11.4%), but the companion sample shrank relatively little. The soil from the sampling in spring barely swelled, but the shrinkage sample realized −8.2%. Plots of the sample strains in Figure 14 highlight how different the responses were. Nevertheless, the shrink-swell index was not significantly affected: 4.5 %/pF compared with 4.6 %/pF. This is just one example where the initial moisture content did not matter.

7. Conclusions

This study reports on observations from a field study aimed at a better understanding of the interactions that take place at the soil-atmosphere-vegetation interface and the influence of other associated parameters. A range of field and laboratory tests have been conducted, along with monitoring of ground movement across a relatively large area with differences in vegetation, i.e., Kikuyu grass and weeds (approximately 3100 m²) surrounded on three sides by 20 m wide strip of mature trees with up to 19 m canopy height. The conclusions from the study are summarised below.

• Significant differences in changes in ground movement over time were observed across the site (Figure 8 shows a range of 10 to 50 mm), highlighting the effect of vegetation and local heterogeneity as evidenced by the subtle differences seen in the soil profiles (Figure 2).
• Seasonal ground movement was reduced generally in areas covered by tree canopies, which was likely due to the long-established deep soil desiccation by the trees and the microclimate created by the tree canopies. Evidence of deep soil desiccation down to 6 m due to the trees was provided by the three suction profiles from September 2019 shown in Figure 6; BH3 near the trees was consistently drier by 0.24 pF below 3.5 m.
• The atmospheric boundary interaction at the soil surface, represented by monthly available water, correlated well with the recorded ground movement well away from the trees around the site, with a time lag of 2 months or so before the soil responded to water availability or deprivation (Figure 12).
• The apparent depth of seasonal suction change was observed to be 1.5 m (Figure 11) in an area where the local climate can be classified as semi-arid, and the design depth of suction change used in house footing design is 4 m.
• The suction profiles near levelling pad G8, which is reasonably close to a group of established trees (D/HT ≅ 1), showed little, if any, evidence of drying by the trees (Figure 11), in contrast to current design guidelines [44].
• Solute suction estimates based on electrical conductivity testing of 1:5 soil-water solutions gave quite high values when compared with total suctions and were within 0.08 and 0.29 pF of corresponding total suctions values (Figure 7).
• Differences of more than three orders of magnitude were observed between hydraulic conductivity values measured from field and laboratory tests (Figure 3 and Figure 13) and are likely to be due to the presence of cracks and fissures in expansive soils, local heterogeneity and the difficulty fully saturating soil in the field.

8. Further Research

Further research at this site is anticipated, which will include a study of shrinkage crack volumes during the dry period, soil water retention characteristics, moisture content monitoring at depth, preferential moisture flow paths and numerical modelling of ground movements with consideration of the atmospheric boundary interaction with the soil surface. The study of tree water uptake of a selection of trees is being undertaken by academic colleagues at Flinders University in Adelaide.