A soil water retention curve (SWRC) is an essential soil physical property for analyzing transport and retention of water in a soil layer. A SWRC is often described as a single-valued function that relates the soil water potential <i>ψ</i> to volumetric water content <i>θ</i> of the soil. However, an <i>in</i>-<i>situ</i> <i>ψ</i> − <i>θ</i> relation should show soil water hysteresis, though this fact is often neglected in analyses of field soil water regimes while long-term <i>in</i>-<i>situ</i> soil water hysteresis is not well characterized. This study aimed at probing and characterizing <i>in</i>-<i>situ</i> <i>ψ</i> − <i>θ</i> relations. The developments of large hysteresis in the <i>in</i>-<i>situ</i> <i>ψ</i> − <i>θ</i> relations were observed only a few times during the study period of 82 months. Any of the large hysteretic behaviors in the <i>ψ</i> − <i>θ</i> relations began with an unusually strong continual reduction in <i>ψ</i>. The completion of a hysteresis loop required a recorded maximum rainfall. Because the study field had very small chances to meet such strong rainfall events, it took multiple years to restore the fraction of soil water depleted by the unusually strong continual reduction in <i>ψ</i>. While wetting-drying cycles had occurred within a certain domain of <i>ψ</i>, hysteretic behaviors tended to be so small that the <i>in</i>-<i>situ</i> <i>ψ</i> − <i>θ</i> relation can be approximated as a single-valued function of <i>θ</i>(<i>ψ</i>). These observed patterns of the <i>in</i>-<i>situ</i> <i>ψ</i> − <i>θ</i> relations were characterized by kinds of difference in d<i>θ</i>/d<i>ψ</i> between a drying process and a wetting process at a given <i>ψ</i>. Thus, more amounts of experimental facts about wetting SWRCs in parallel with drying SWRCs should be needed for correct modelling, analyzing, and predicting soil water regimes in fields. It is also necessary to increase our understandings about the long-term trends of occurrences of extreme weather conditions associated with possible change in climate.
A soil water retention curve (SWRC) is a relationship between the soil water potential ψ and volumetric water content θ of the soil. Mathematical features of SWRC have been used for classifying soil water in terms of energy state or plant-availability (Veihmeyer & Hendrickson, 1931; Hoover, 1950; Richards & Weaver, 1943; Veihmeyer & Hendrickson, 1948) . A SWRC can also allow to infer pore size distribution of a soil based on the theoretical relation between pore water pressure and radius of capillary pores (Washburn, 1921) . The concept of specific water capacity dθ/dψ is another aspect of the pore size distribution given by a SWRC, indicating how a soil is apt to dry or wet with the change in ψ under a certain moisture condition.
A SWRC is, thus, mostly described by a single-valued continuous function θ(ψ) for the convenience of the evaluation works mentioned above. Common procedures to obtain a θ(ψ) are laboratory tests using hanging-water columns or pressure-plate apparatuses (Dane & Hopmans, 2002) . Most of such procedures rely on a drying process through which water is removed from an initially saturated soil sample by increasing suction on water in the soil sample step by step. After the procedure gives a series of ψ − θ plots, a mathematical function of θ(ψ) is fitted to the measured ψ − θ plots so that the parameters in the expression of the θ(ψ) are identified. Classical studies that proposed mathematical functions for describing SWRCs are Campbell (1974) , Clapp and Hornberger (1978) , and van Genuchten (1980) . Once a θ(ψ) is identified, it can be used to typify a ψ − θ relation that would be observed for the soil in question.
However, a series of ψ − θ plots, or a ψ − θ relation, in a field may not be appropriately formulated by a single-valued function, because the field soil is incessantly experiencing wetting-drying cycles so that its ψ − θ relation may vary, which is known as soil water hysteresis (Hillel, 1998; Kutilek & Nielsen, 1994) . Soil water hysteresis was first discussed by Haines (1930) , and was later explained by the analogy named the “ink-bottle effect”. The ink-bottle effect can occur due to varying pore-diameter along a capillary pore, causing the difference in soil water potential between the water going into the pore and the water going out the pore. This effect may be found more readily among coarse-textured soils in a high-ψ domain when its mechanism is considered. Other mechanisms include the “contact-angle effect”. This effect is due to the hysteresis of the contact angle between an air-water interface and a soil-solid surface. The angle generally grows when water moves toward a dry solid surface and, contrarily, closes when water recedes, as measured for soils (Letey et al., 1962) and for glass beads (Laroussi & de Backer, 1979) . This effect can also induce the difference in soil water potential between a wetting process and a drying process.
Hysteresis in soil water retention has been studied mainly through laboratory experiments (e.g., Topp & Miller, 1966; Topp, 1969 ), while several studies that used larger soil masses have also shown that hysteretic behaviors cannot be neglected (Watson et al., 1975; Beese & van der Ploeg, 1976; Tzimas, 1979) . However, so far, limited number of studies have reported the results of multiple years of monitoring about ψ − θ relations in fields. Royer and Vachaud (1975) carried out a series of independent measurements of changes in water content and soil suction during a year, and suggested that hysteresis is too important to be ignored. Basile et al. (2003) compared laboratory- and field-measured soil water retention curves, and reported that hysteresis is essential to interpret the difference between the laboratory-measured and the field-measured soil water retention functions. However, their datasets were composed of small number of ψ − θ plots. Contrary to these studies, Li et al. (2005) reported that field measured soil water retention curves showed negligible hysteresis. As described above, knowledge about in-situ ψ − θ relations seems partly contradictive.
Recent studies tend to monitor in-situ ψ − θ relations for longer periods. Bordoni et al. (2017) monitored soil water pressure and water content in fields for 58 months. They obtained in-situ SWRCs for pairs of drying and wetting processes for various periods, and confirmed that soil water balance was better reproduced by considering the in-situ SWRCs of the two kinds of processes. Zhang et al. (2022) simulated varying soil moisture conditions for precipitation events by using field-estimated drying and wetting SWRCs, and concluded that a simulation with a field-drying SWRC and a field-wetting SWRC gave better results than that with a laboratory SWRC. Based on these studies, the likelihood is that using at least a pair of drying and wetting SWRCs gives better estimations or predictions of soil water movement in fields. And, thus, the study on soil water hysteresis is in the stage of characterizing in-situ hysteretic behaviors by gathering as many cases of wetting-drying cycles in soil layers as possible.
Therefore, this study probed in-situ ψ − θ relations for 82 months by coincidently monitoring soil water potentials ψ and volumetric water contents θ, and characterized their hysteretic behaviors. The tested soil layers were volcanic ash soil layers that are typically found in eastern part of Japan archipelago. Volcanic ash soils are also primary land-covers in areas with volcanoes like the densely populated regions along the Pacific rim.
The study field was a meadow (36˚29'23''N, 139˚59'14''E) in the Utsunomiya University Farm in Moka city, Tochigi, Japan. From 2013 to 2019, Moka city had mean atmospheric temperatures of 14.0˚C, 13.6˚C, 14.3˚C, 14.1˚C, 13.5˚C, 14.7˚C, and 14.4˚C and annual precipitations of 1249, 1604, 1391, 1421, 1140, 1045, and 1458 [mm], respectively (Japan Meteorological Agency, 2024) . The meadow was 2.8 ha in size with its sides of 350 m from north to south and of 80 m from east to west, along the east side of which larch tree rows (Larix kaempferi (Lamb.) Carr.) had stretched as a windbreak. The grasses and forbs that had grown in the meadow were orchard grass (Dactylis glomerata L. cv. Natsumidori), tall fescue (Festuca arundinacea Schreb. cv. Kentucky 31), perennial ryegrass (Lolium perenne L. cv. Friend), and white clover (Trifolium repens L. cv. Huia) until July 2014. Then, the meadow was renovated for recovering yields and, in October 2014, orchard grass (Dactylis glomerata L. cv. Natsumidori), tall fescue (Festuca arundinacea Schreb. cv. Hokuryo), hybrid ryegrass (Lolium hybridum Hausskn. cv. Tetrelite II) and red clover (Trifolium pratense L. cv. Makimidori) were newly planted.
Two study sites have been placed at 2- and 8-m away from the larch tree row in the study field (
Volumetric water content θ and soil water potential ψ had been coincidently measured at each of the sampling points two to three times a week from June
2013 to March 2020. The values of θ were measured by using a capacitance-type soil moisture sensing system (Diviner 2000, Sentek Pty Ltd., Stepney, South Australia). This system was calibrated particularly for the study site (Iiyama, 2016) . The tensiometer tubes were permanently installed to evaluate the values of ψ. And air-pressure in the head space of a tensiometer tube was measured by a digital manometer (PG-100-102VP; NIDEC COPAL Electronics Corp., Tokyo, Japan), which was converted to ψ in accordance with the height of water column in a tensiometer tube.
When the devices for measuring θ and ψ were installed at each of the sampling points, triplicate undisturbed soil cores of 50 cm3 and disturbed soil samples were taken. The soil cores were used to determine θ under −0.1, −0.98, −9.8, and −98 [J∙kg−1] in ψ by applying the hanging-water-column and the pressure-plate methods (Dane & Hopmans, 2002) . Soil bulk density ρd was also determined with the soil cores by the gravimetric method in which the soils are dried at 105˚C for more than 24 hours. Soil particle density ρs was measured with the disturbed soil samples in accord with Japan industrial standards “JIS A 1202” using Gay-Lussac-type pycnometers. Soil texture was also determined through the particle-size analysis (Gee & Or, 2002) .
Soil physical properties of the soil samples were listed in
The values of θpF3 indicated that more than 60% of the entire soil volume is filled with water even at ψ = −98 J∙kg−1, though the soils nearer to the tree rows
| sampling points | soil texture | ρd [Mg∙m−3] | ρs [Mg∙m−3] | θpF0 [m3∙m−3] | θpF1 [m3∙m−3] | θpF2 [m3∙m−3] | θpF3 [m3∙m−3] |
|---|---|---|---|---|---|---|---|
| A-40 | Clay Loam | 0.612 (0.011) | 2.836 (0.026) | 0.786 (0.007) | 0.736 (0.031) | 0.654 (0.011) | 0.610 (0.007) |
| A-100 | 0.519 (0.006) | 2.808 (0.011) | 0.800 (0.025) | 0.762 (0.036) | 0.698 (0.017) | 0.650 (0.015) | |
| B-40 | 0.600 (0.023) | 2.811 (0.053) | 0.754 (0.008) | 0.734 (0.006) | 0.671 (0.024) | 0.609 (0.031) | |
| B-100 | 0.602 (0.023) | 2.777 (0.028) | 0.768 (0.025) | 0.739 (0.004) | 0.681 (0.010) | 0.651 (0.011) |
were likely to be drained more easily. This feature is found commonly on a Kanto Loam layer. In general, subsoils of a Kanto Loam layer have large amount of non-free water (Takenaka, 1963) so that sometimes more than half of the pores is undrained even under a moisture condition of the permanent wilting point (Yamazaki et al., 1963) . Also, large fraction of free-water is retained in poorly-continuous or dead-end macropores within the massive soil structure and, sometimes, lowering ψ to −300 J∙kg−1 is required for draining out the free-water fraction (Takenaka et al., 1963) .
The time-series of water potentials ψ and volumetric water contents θ are shown in
timescales of months occurred several times. On the other hand, the values of ψ in the sampling points B-40 and B-100 had minutely fluctuated mainly in the range above −10 J∙kg−1, suggesting that the soil layers were moist on most measurement dates and their pore-water pressure easily went up in response to almost every rainfall.
In the site A, the largest reduction in θ occurred from April to August in 2018. In this period, the values of θ at A-40 and A-100 had decreased from 0.6 m3∙m−3 to 0.4 m3∙m−3 or less with concurrent continual declines in ψ. Particularly at A-100, θ had stayed around 0.45 m3∙m−3 for almost one year even after ψ recovered above -10 J∙kg−1. A definite chance of recovery in θ in the site A happened on October 12 in 2019 when the typhoon Hagibis passed Moka city with daily rainfall of 209.5 mm∙d−1, including the recorded maximum hourly rainfall of 27.5 mm∙h−1 in the study period (Japan Meteorological Agency, 2024) . After that, the values of θ in the site A regained to above 0.55 m3∙m−3.
In the site B, ψ dropped down most apparently in May in 2015. At B-40, this reduction in ψ caused the decrease in θ to less than 0.5 m3∙m−3. After that, it took more than 4 years for this depth to restore the value of θ to around 0.6 m3∙m−3, though the value of ψ at this depth migrated around the same level as found before 2015. At B-100, where the level of ψ was almost always higher than -10 J∙kg−1, θ had never been below 0.6 m3∙m−3 in the entire study period, suggesting that the demand for soil water at this depth had been too weak to withdraw substantial amount of soil water from the soil layer.
water content θ for the site A. Each ψ − θ plot was made by linking a value of ψ to the value of θ that was obtained with the ψ in the same sampling date. The laboratory-measured ψ − θ plots, as shown in
The in-situ ψ − θ relations were plotted consistently in the domain lower in θ than the laboratory-measured ψ − θ relations, implying that they rarely follow the primary drying SWRC obtained through a laboratory experiment. The trajectory of the in-situ ψ − θ plots showed hysteretic behaviors with various size and temporal scales.
The large hysteresis observed at A-100 from 2018 to 2019 drew a clockwise loop on the graph. In the first half of 2018, the value of θ of A-100 continually decreased from 0.6 to 0.4 m3∙m−3 with ψ going down from −1 J∙kg−1 to −100 J∙kg−1 and, then, stayed around 0.42 m3∙m−3 by October in 2019. When the typhoon Hagibis passed Moka city on October 12 in 2019, the value of θ of A-100 jumped up to 0.6 m3∙m−3, and the ψ − θ relations merged into those obtained before 2018 again. Contrary, at A-40, hysteretic behaviors were not so clearly observed as at A-100, because the value of ψ at A-40 fluctuated more frequently and minutely than that at A-100 so that the scales of wetting-drying cycles were not enough large to make a clear hysteresis loop of ψ − θ plots.
water content θ for the site B in the same manners as
In the site B, the values of θ were ordinarily higher than those in the site A during the whole study period so that any drying process did not made progress so far as those in the site A. Particularly at B-100, θ varied within a very limited range around 0.65 m3∙m−3 while ψ fluctuated around −1 J∙kg−1. Thus, the ψ − θ relation at this depth was plotted in such a narrow band-like region on the graph, and can roughly be approximated by some single-valued function θ(ψ). At B-40, the series of ψ − θ plots showed a clockwise hysteretic spiral, instead of a loop, in 2015. In the first three weeks in May in 2015, θ decreased monotonously from 0.63 to 0.45 m3∙m−3 while ψ lowered from −8 to −72 J∙kg−1. After that, the ψ − θ relation went back and forth repeatedly in the domain of −100 J∙kg−1 < ψ < −1 J∙kg−1 with 0.43 m3∙m−3 < θ < 0.60 m3∙m−3 till the end of 2015. Then, the ψ − θ relation was going upward very gradually for 4 years and, as a result, the ψ − θ relation obtained in 2019 were plotted in the region obviously higher in θ than those in 2015.
The facts found in
The sub-graphs A-100 in
The observed features of how a ψ − θ relation shows hysteresis can be schematized in
On these assumptions, the factors affecting how hysteretic behavior of a ψ − θ relation appears should be (f1) the width of each of the domains of ψ, and (f2) the size of difference in dθ/dψ between drying and wetting processes in each of the domains of ψ. For example, when a soil has a broad domain- (ii), soil water hysteresis of the soil should become apparently negligible when wetting-drying cycles of the soil are repeated well within the domain- (ii) as found at B-100 in this study (“ex1” in
On the other hand, when wetting-drying cycles are repeated solely in the domain- (iii), a ψ − θ relation should gradually shift downward on the graph, and finally, can be found migrating around the primary wetting curve as shown at A-100 and B-40 in this study (“ex2” in
this type of downward shift in ψ − θ relation can be restored when the soil experiences a wetting process that reaches the domain- (i) as observed at A-100 (“ex3” in
A large hysteretic loop can emerge when a ψ − θ relation passes both the domain- (i) and the domain- (iii), namely, when both a demand for soil water strong enough to make ψ go into the domain- (iii) and a water supply to the soil intense enough to regain ψ to the domain- (i) occur. And it is likely that hysteresis loops appear more clearly in a soil that has SWRCs with larger difference in dθ/dψ between the wetting and drying processes in any of the domain- (i) and the domain- (iii).
A volcanic ash soil as used in this study is likely to be typical of a soil with a broad domain- (ii) and, thus, can show an apparently “hysteresis-less” ψ − θ relation for years. A ψ − θ relation for a soil with a narrow domain- (ii), on the other hand, is expected to shift more frequently toward a moister or a dryer region of θ. In addition, a soil in a climatic region with more distinct a pair of rainy season and dry season in a year may have a ψ − θ relation with more drastic soil water hysteresis.
These expectations should be confirmed with further collections of in-situ ψ − θ relations for various combinations of soil types and climatic zones. Particularly, knowledge about wetting SWRCs, in addition to drying SWRCs, should be accumulated to analyze and predict soil water regimes in fields in more precise manners. At the same time, how often large hysteresis loops are clearly drawn depends on how frequently the land encounters extreme weather conditions. So, it is also necessary to gain our understandings about the long-term trends of frequencies of extreme weather conditions associated with possible change in climate.
For characterizing hysteretic behaviors of in-situ ψ − θ relations, soil water potentials ψ and volumetric water contents θ had been monitored coincidently in a meadow field for 82 months. The tested soil layers were volcanic ash soil layers, as an example of primary land-covers in areas with volcanoes like the densely populated regions along the Pacific rim. Laboratory-measured ψ − θ relations were also obtained by using undisturbed soil samples for the comparison with the in-situ ψ − θ relations.
The in-situ ψ − θ relations were plotted consistently in the domain lower in θ than the laboratory-measured ψ − θ relations, implying that in-situ ψ − θ plots mostly follow scanning SWRCs and rarely follow the primary drying SWRC.
Large hysteretic behaviors in the in-situ ψ − θ relations were observed a few times during the study period. The large hysteretic behaviors began with unusually strong continual reductions in ψ. Once the value of θ decreased significantly by the strong continual reduction in ψ, the in-situ ψ − θ relation tended to reside in a low-θ region on the ψ − θ graph. To complete a hysteresis loop, a recorded maximum rainfall was required. In the sampling depth that had not experienced such a large amount of water supply in a short period, the ψ − θ relation failed to complete a loop and, instead, had developed a spiral trajectory on a ψ − θ graph for at least four years. In either way, it was suggested that it will take many years to recover a loss of soil water once a significant fraction of soil water is depleted by an unusually strong continual reduction in ψ. In case that wetting-drying cycles had occurred solely within a certain domain of ψ during the entire study period, hysteretic behaviors were so small that the in-situ ψ − θ relation can be modeled by a single-valued function of θ(ψ).
This study simplified the features of soil water hysteresis into the following three assumptions: (a1) The domain of ψ is divided into three types of sub-domain, namely, (i) the domain with a drying process of gradual dθ/dψ and a wetting process of steep dθ/dψ, (ii) the domain of an apparently “hysteresis-less” ψ − θ relation, and (iii) the domain with a drying process of steep dθ/dψ and a wetting process of gradual dθ/dψ. (a2) Any scanning curve is drawn inside the loop of primary drying and wetting curves. (a3) Any scanning curve has a similar figure of a corresponding primary curve, and is asymptotically merging into the corresponding primary curve until the process will reverse its direction.
These features of soil water hysteresis suggest that the factors affecting how hysteretic behavior of a ψ − θ relation appears should be (f1) the width of each of the domains of ψ, and (f2) the size of difference in dθ/dψ between drying and wetting processes in each of the domains of ψ. To confirm these characterizations, more amounts of experimental facts about wetting SWRCs, in addition to drying SWRCs, should be accumulated to analyze and predict soil water regimes in fields in more precise manners. At the same time, it is also necessary to increase our understandings about the long-term trends of occurrences of extreme weather conditions associated with possible change in climate, since the frequency of large drying- or wetting-shifts in an in-situ ψ − θ relation must depend on the frequency of extreme weather conditions that the field encounters.
The author thanks Mr. T. Shiozawa, Mr. E. Saito, Mr. K Seino, and Mr. N. Yamaguchi at the Utsunomiya University Farm who have managed the study field. The author also thanks Mr. T. Hirai of the former student in the Graduate school of Agriculture, Utsunomiya University, and Mr. H. Ajito, Ms. M. Anzai, Ms. K. Arakawa, Ms. M. Cho, Ms. N. Goka, Mr. T. Hoshino, Mr. D. Iimura, Mr. J. Kaneko, Mr. T. Kurata, Mr. K. Ogasawara, Mr. Y. Sakanishi, Mr. H. Shoji, Mr. H. Suto, Mr. S. Takaki, Mr. K. Tezuka, Mr. N. Totsuka, Mr. K. Tsukagoshi, and Mr. Y. Usui of the former students in the Faculty of Agriculture, Utsunomiya University, for their works in the field and in the laboratory.
The author declares no conflicts of interest regarding the publication of this paper.
Iiyama, I. (2024). Characteristics of In-Situ Soil Water Hysteresis Observed through Multiple-Years Monitoring. Journal of Geoscience and Environment Protection, 12, 162-175. https://doi.org/10.4236/gep.2024.125010