Comparison between Soil-Water Characteristic Curves based on filter paper method and grain-size distribution/volume-mass prediction: An evaluation of input data stemmed from hydrometer and Cilas grain-size analyser.

. The current paper aims to present a comparison between soil-water characteristic curves based on the filter paper method and on a volume-mass estimation. Additionally, this research consists in comparing grain size distributions by sieving/hydrometer and by sieving/Cilas (particle-size analyser model 1092), when it comes to determining the soil-water characteristic curves by a volume-mass prediction on the Soil Vision software. As materials, were used two sets of colluvial soils sampled at the Campus Quinta do Órgãos – Brazil. The results showed that the soil-water characteristic curves related to the filter paper method or grain size distribution estimation are different. However, the tests performed by sieving/Cilas are more efficient than the results based on sieving/hydrometer, concerning its uses as input data for soil-water characteristic curves estimations. In conclusion, even considering the estimation method was not able to depict the same results such as obtained by the filter paper method, the use of Cilas is a procedure that can improve the quality of the predicted soil-water characteristic curve.


Introduction
When it comes to analysing unsaturated soils, the knowledge of drying (desorption) and wetting (sorption) processes [1] is considered a key factor for slope stability, flow, soil irrigation, and further analysis.
One of the most important approaches is related to the soil-water characteristic curve (SWCC). Since it is assumed [Williams (1982) ccc] as the relationship between water content and suction for a given soil, the SWCC is established by constitutive models [9]. However, the suction of soil under a given water content depends on its history of wetting and drying, and it is difficult to point out.
In order to determine the SWCC, the most common standard test method is based on filter paper [10]. The procedure is related to the suction equilibrium principle, in which two porous materials, when placed in contact, will lose and absorb water until the equilibrium of suction.
Even being a standardised method [11], the filter paper's based procedures have the equilibrium of suction as the major factor responsible for its time-consuming characteristic since its stabilisation time varies according to the level of suction measured and may vary from 7 to 30 days [12][13]. In this way, it is considered a high timeconsuming and because of this; The method is underused in comparison with the benefits it could input in the unsaturated analysis.
Thus, prediction models based on grain-size distributions are revealed as feasible options to obtain the SWCC in a reduced time. An advantage of this technique is that measuring soil grain-size distribution is much more practical and usual than techniques such as the filter paper method. They are easier to perform, and the results are obtained faster as the suction equilibrium time is avoided.
Because of this, the current research is focused on the comparison of a predicted SWCC (based on grain-size distribution -GSD) and the filter paper standard method. Further, it is tested the use of grain-size based on sieving/hydrometer or based on sieving/Cilas particle size analyser as input data for the soil-water characteristic curves estimated by using the Soil Vision software. nonlinear soil properties require several tests for the same specimen in order to depict reliable results. One additional obstacle hindering the determination of soilwater characteristic curves is that the same suction might be related to different water content. It happens because the drying path is generally greater than the wetting, showing that there is hysteresis. Thus, it is noteworthy that there are several curves for a given soil, depending on the wetting/drying path.

Theoretical background
A vast amount of empirical equations has been proposed to fit experimental data for soil-water characteristic curves (Fredlund, 2012 p.201 [1]). Even though numerous unimodal equations are exposed in the literature, there are an increasing number of equations that reveal a bimodal behaviour in the shape of SWCCs (Qi & Vanapali, 2015[14]).
The unimodal curves present only one desaturation branch ( Figure 1). In the same figure, it is possible to observe the effect of hysteresis since the adsorption is lower than the sorption. In the matter of bimodal soil-water characteristic curves, it is composed of two desaturation branches ( Figure 2). This type of curve presents two air-entry values (AEVs) related to macro (1st AEV) and micropores (2st AEV).

Fig. 2. A typical bimodal SWCC [Adapted from 14].
Besides experimental procedures (here, essentially comprised by filter paper method), indirect pedo-transfer functions (PTFs) [16] have been considered an alternative method when it comes to estimating of the soil-water characteristic curve. A PTF is a function that has its basis on elementary soil data such as the grainsize distribution (Fredlund et al. 2002 [1]). Some of the most useful PTFs are the researches presented by

Materials
The colluvial soils used in the tests have been sampled from two depths (Pt 01 -25 cm and Pt 02 -75 cm) located at Campus Quinta do Paraiso -UNIFESO (geographic coordinates 22°23'35.02" south and 42°57'40.78" west). The sampling point belongs to the Serra dos Órgãos, which is the local description for Serra do Mar. In the geological survey, the area comprises granite (monzogranite) and gneiss rocks. The rocks are constituted by a wide range of minerals such as quartz, muscovite, migmatites [19,20]. Further, regional and local faults are observed as consequence of geomorphological processes.

Methods
The method used consists of determining grain-size distribution curves by the standard method and the GSD revealed by the particle size analyser. Afterward, the soil-water characteristic curves were performed based on the filter paper. Additionally, the Soil Vision software was used to fit the GSD according to the experimental values. Further, it was also used to estimate the SWCCs based on grain-size distribution. Previously the test execution, the soils #40 and the fine particles were used in 1190 (Figure 33), with the measureme 0.04 and 2,500 microns. The Cilas 1 with the ISO 13320 standard for partic as well as the 21 CRR part 11 standard.

SWCC based on filter paper m
The soil-water characteristic curves we the filter paper method. For the tes samples ( Figure 4) for each soil. The intervals were obtained by two branch adsorption) and because of this, the ef was not evaluated. Further, there variation of the soil samples during the In order to fit the acquired curv website http://seki.webmasters.gr.jp/sw the method Seki [18] for a bimodal adju

SWCC based on estimated me
The estimation of SWCC was carried Vision software. As input data, were (standard and Cilas), the specific gravity the saturated gravimetric water content chosen the Fredlund bimodal fit [17] w source type coupled to the Ayra and method [18]

Fig. 5. Soils water characteristic c
The fitting method depicte experimental data, confirming ( Figure 6 a -b). ions performed presented ng the methods carried out Pt 01G and Pt 02G -Cilas itations was not possible to icle sizes. Due to this, the 001 mm (Figure 6 -a). On the GSD revealed by the method), it was possible to re 6 -b). sise that the particle size cles lower than 0,06 mm. e that the superior branches curves were not changed eving procedure. Both grain-size distribution curves w Vision software. The results (Figure 7 a a -b) depict the standard method was Unimodal fiting while Cilas coupled and Pt 02G), pointed out a Bimodal be the Pt 02G presented a noticea Furthermore, the amount of clay, silt, a changed for both soils, affecting the curve.  The results emphasise the SWCC when is performed wit the sieving/hydrometer GSD s b. and Figure 10 b.) For both soils, the p characteristic curves relat distribution, performed with revealed maximum suction va kPa. It can be considered un method reached a maximum of The first air-entry val comparison with the results filter paper method. Accordin values should be around 10 k observed values around 100 kP Regarding the second air-e micro-pores and to the b distribution obtained with Cil slightly better. However, composition, the analysis resu for both soils.  e better prediction of the th Cilas in comparison with standard method (Figure 9 prediction of soil-water ted to the grain-size h the standard method, alues higher than 1.000.000 nreal since the filter paper f 100.000 kPa. lues were different in of the tests based on the ng to the bimodal fit, the kPa.

Conclusion
The grain size distribution revealed a bimodal geometry for the Cilas coupled method, and it was not clear for the GSD results obtained by the standard method. It suggests accordance with (Qi et al. 2009 [14]), in which highlights "there appears to be greater difficulty in estimating the SWCC for silt-clay soils, silt-clay-loam soils, and silt-loam soils for both PTFs, although the predicted SWCCs look similar to the measured results" In a comparison of Arya-Paris (1981) [17] with Fredlund et al. (2002) [18], pedo-transfer functions, the solution based on Fredlund et al. (2002) [1] PTF performed slightly better than the Arya-Paris (1981) PTF for both soils.
Finally, Cilas particle-size analyser provides a better knowledge of the particle-size distribution of the soils, contributing to generate refined soil-water characteristic curves by using the prediction method on Soil Vision software.