Estimating residual water content in air-dried soil from organic carbon and clay content

doi:10.1016/j.still.2014.09.021

Soil and Tillage Research

Volume 145, January 2015, Pages 181-183

https://doi.org/10.1016/j.still.2014.09.021 Get rights and content

Highlights

•
Residual water content in air-dried soil leads to a bias in concentration measurements.
•
Residual water is significantly correlated with soil carbon and clay content.
•
We derived functions to estimate residual water contents in air- and 40 °C dried soils.
•
The uncertainty RMSD) of the best model was a residual water content of 0.34%.

Abstract

Measurement of soil organic carbon (SOC) or other elements in air-dried soil samples leads to underestimation of mass fraction per unit dry soil (concentration) unless the residual water (RW) content is accounted for. However, RW measurements are time-consuming and costly and thus often neglected. The resulting bias can lead to dramatic erroneous results, especially when two differently treated datasets are compared. We therefore, derived pedotransfer functions to estimate residual water content in air-dried soil samples from: (i) SOC content, (ii) clay content and (iii) both variables together. The uncertainty of prediction decreased in the order (i)–(iii), with root mean squared deviation (RMSD) values of 0.64, 0.46 and 0.34% of RW content, respectively. These functions represent a potential step towards more harmonized and transparent SOC determination.

Introduction

The content of organic carbon and other elements in soil is often measured in air-dried samples which still contain some water. Dry weight (DW) usually refers to a sample where the residual water (RW) has been removed by drying at 105 °C. It is often unclear whether data reported in the literature refer to air-dry or DW samples. If a set of soil organic carbon (SOC) data is assumed to be on a DW basis when in fact it is on an air-dry basis, this results in systematic underestimation of SOC content. Therefore, RW needs to be determined in a subsample. However, it is time-consuming and thus costly to conduct this extra analysis for each sample, especially in large-scale case studies involving a huge amount of soil samples, while the bias might be relatively small. In some studies, the RW content is therefore determined only for pooled samples, e.g. for a site, and the value obtained is applied to correct all individual samples from that site, which adds uncertainty to each SOC measurement (Poeplau and Don, 2013). In other studies, RW may be considered negligible and not measured at all, introducing a bias to the sample set which becomes problematic when the study is compared with others. The most striking example is a repeated soil carbon inventory, where if RW content is accounted for on one sampling occasion but not another, the bias introduced may indicate a significant difference even if the actual SOC content remains unchanged. A residual water content of 3% does lead to an overestimation of around 1.5 Mg C ha⁻¹, assuming a SOC stock of 50 Mg C ha⁻¹. This is in the range of decadal effects of certain organic amendments, such as crop residue incorporation on SOC (Smith et al., 2005). A-posteriori determination of RW in samples is often impossible, so the alternative is to use a pedotransfer function (PTF). It is well known that soil water retention is largely a function of texture and organic matter content, determining the amount of fine pores and surface area on which adsorption takes place (Briggs and Shantz, 1912, Van Genuchten, 1980, Vereecken et al., 1989). Wäldchen et al. (2012) recently attempted to estimate clay content from the RW content of air-dried samples. Using 240 and 137 observations obtained using two different methods for determining soil clay content, they were able to estimate clay content with an uncertainty of 20 and 28%, respectively (scaled mean absolute error). However, this may not be sufficiently accurate for a parameter of such importance.

Our aim was to derive a PTF to replace extensive measurements of RW content, using a subset of data from the Swedish arable soil monitoring programme 2001–2007 (Eriksson et al., 2010).

Section snippets

Materials and methods

In the above mentioned programme, soil samples to 20 cm depth were taken at approximately 2000 sampling locations all over Sweden. The sampling was conducted following a fixed countrywide sampling grid. For each location, only one pool sample was analyzed. The observations can thus be considered independent and identically distributed. Organic carbon content was analyzed by dry combustion (LECO, St. Joseph, Michigan, USA). Particle size distribution was determined by a combination of wet sieving

Results and discussion

SOC content was significantly correlated with RW content (R² = 0.24) (PTF1; Fig. 2A). High RW contents tended to be underestimated and low RW contents overestimated. The uncertainty (RMSD) of estimation using SOC was 0.64%. Clay content was also significantly correlated with RW content, with much higher explanatory power (R² = 0.63) (PTF2, Fig. 2B), and with RMSD reduced to 0.46%. However, the best performance was achieved with a multiple regression combining clay and SOC content (R² = 0.81) (PTF3;

Acknowledgements

The Swedish arable soil monitoring programme is funded by the Swedish Environmental Protection Agency. For statistical support we thank Roland Fuss.

References (9)

G. Piñeiro et al.
How to evaluate models: observed vs. predicted or predicted vs. observed?
Ecol. Model.
(2008)
C. Poeplau et al.
Sensitivity of soil organic carbon stocks and fractions to different land-use changes across Europe
Geoderma
(2013)
R. Andersen
Modern Methods for Robust Regression
(2008)
L.J. Briggs et al.
The wilting coefficient for different plants: and its indirect determination
Govt. Print. Off
(1912)

There are more references available in the full text version of this article.

Cited by (15)

Effects of soil pore structure on gas diffusivity under different land uses: Characterization and modelling
2024, Soil and Tillage Research
Soil gas diffusion significantly affects soil respiration and greenhouse gas emissions, and is strongly determined by soil structure (mainly pore structure) and water content. Given the challenge of quantifying soil gas diffusivity and pore structure, the effect of soil structure on gas diffusion under different moisture conditions remains elusive. Here we measured soil gas diffusivity ( $D_{s} / D_{0}$ ) of intact and repacked soils under four typical land uses, including irrigated cropland (IC), native grassland (NG), native forest (NF), and wetland (WL), at different soil matric potentials (ψ, –1 $\leq$ ψ $\leq$ –1500 kPa) using one-chamber method. The pore structure of intact soils was quantified using micro-computed tomography (µ-CT) and soil water retention curve. The $D_{s} / D_{0}$ for repacked soils with different bulk densities (1.2–1.5 g cm⁻³) and clay contents (14–35%) were further determined. Land use significantly affected $D_{s} / D_{0}$ with NF > NG > IC > WL, and disturbing soils via repacking generally decreased $D_{s} / D_{0}$ , especially for NF. Increased bulk density and clay content decreased $D_{s} / D_{0}$ , particularly at low ψ (–50 $\leq$ ψ $\leq$ – 1500 kPa), under which $D_{s} / D_{0}$ was controlled by volume of pores with diameter > 6 µm and particularly 6–300 µm. By contrast, $D_{s} / D_{0}$ at relatively high ψ (–1 $\leq$ ψ $\leq$ –10 kPa) was driven by volume of pores with diameter > 300 µm. Furthermore, an empirical model of $D_{s} / D_{0}$ was developed by incorporating a scale parameter, which quantifies the effects of soil pore structure on gas diffusivity. The developed model predicted the $D_{s} / D_{0}$ under different land uses and moisture conditions better than most currently used models. Overall, this study reveals how soil structure quantitatively affected gas diffusivity under different land uses and moisture conditions, and provides a feasible model to predict soil gas diffusivity, which is critical for accurately estimating and predicting greenhouse gas emissions from soils. This study suggests the importance of considering soil structural parameters derived from soil water retention curve and/or µ-CT in predicting gas diffusion across different soil types and land uses.
Determination of soil water retention curves from thermal conductivity curves, texture, bulk density, and field capacity
2024, Soil and Tillage Research
The soil water retention curve (SWRC) is frequently expressed using the van Genuchten (VG) model, which has four parameters: saturated water content (θ_s), residual water content (θ_r), α, and m (1–1/n). Soil thermal conductivity (λ), which is linked to the hydraulic properties of unsaturated soil, has been a proxy variable used to estimate SWRC. In this study, we present a new approach to estimate the VG model parameters. Parameters θ_s, α and m are calculated from the information of soil texture, bulk density (ρ_b), and a measured water content at field capacity (θ_fc, at −33 kPa or −10 kPa), and θ_r is estimated from the thermal conductivity versus water content curve, λ(θ), based on similarities between SWRCs and λ(θ) curves. The new approach was evaluated with laboratory and field measurements on 23 soils of various textures, ρ_b values, and θ values. Results showed that for repacked core samples, intact core samples, and in situ field soils, the new approach estimated SWRCs with average root mean square errors (RMSEs) of 0.042, 0.030, and 0.049 m³ m^‐3, respectively. The new approach offers a quick and effective way to estimate SWRCs accurately with measured λ(θ) curves, texture, bulk density, and θ at field capacity.
Pattern of spatial variability in granulometric and mineralogical characteristics across an ecologically sensitive sub-tropical wetland (Bihar, India)
2023, International Journal of Sediment Research
Floodplain wetlands are dynamic systems undergoing frequent hydro-sedimentological changes. Tropical/subtropical wetlands have even more pronounced climatic and biological influences on the sediment dynamics. In the current study, a protected subtropical wetland in Bihar (India) has been assessed for its sedimentary and mineralogical status. Sediment characteristics and associated granulometric parameters are discussed to highlight the wetland's hydrogeomorphic character, energy dimension, and the depositional processes taking place. Statistical grain size parameters like graphic mean, graphic skewness, graphic kurtosis, and their interrelations studies have helped in understanding the sources of deposited materials. Multivariate statistics have identified, sand fractions, measures of central tendencies (mean, median, and mode), percent organic matter (%OM) and percent carbonates (%carbonates) as primary factors explaining the maximum variance in the data. The non-parametric Kruskal–Wallis test has identified %sand, %clay, electrical conductivity (EC), and %carbonates as the major factors responsible for significant spatial variability. The grain texture analysis shows that the majority of samples fall in the brackets of clayey silt; silty sand; and sand, silt, and clay. Common mineral components associated with these are quartz, feldspar, clay minerals, etc. The range for the mean grain size lies between 3Φ to 9Φ (Φ is diameter of the sediments) where most of the grain size distributions are symmetrically to fine symmetrically skewed, indicating intermediate energy to adjust to environmental conditions. The granulometric and physicochemical characteristics of the sediment samples also indicate the directional distribution pattern of sediment along the hydrological pathway. Future studies concentrating on the fluvial discharges of water and sediment will be highly insightful in reconstructing the past net sediment accumulation processes in the Baraila Wetland.
Contribution of organic carbon to the total specific surface area of soils with varying clay mineralogy
2023, Geoderma
Citation Excerpt :
Kirschbaum et al. (2020) determined the OC contribution to apparent SA (determined from water sorption at RH = 0.42) using New Zealand soil samples collected from different depths that were characterized by poorly interstratified hydroxyl interlayered vermiculates and mica-vermiculite clays, with a minor contribution of kaolinite, quartz, and feldspar They found that OC contributed approximately 0.43 m2 mg C−1, which translates to 4.3 m2/g per %C. According to Kirschbaum et al. (2020), this estimate of OC contribution ranges from 5.06 (Poeplau et al., 2015) to 7.1 m2/g per %C (Parfitt et al., 2001). The differences between the cited studies and the present work can be attributed to the differences in clay content, clay mineralogy (as can be seen from the different soil groups here), and the hydrophilic or hydrophobic nature of the OC present (Hermansen et al., 2019).
Soil specific surface area (SA) reflects the quantity and quality of the soil mineral and organic fractions. For soil samples that have similar clay mineral types, clay content and or organic carbon (OC) contents are good predictors of SA. However, the magnitude of the OC contribution to SA for different soil types is unclear, particularly since SA varies depending on which method or probe molecule is used to measure it. Consequently, we set out to (i) quantify the contribution of soil OC to total SA for soils with varying clay mineralogy, and (ii) assess the effect of probe molecule (EGME or H₂O) and water sorption direction on the contribution of OC to the total SA. We utilized 330 soil samples (clay = 1 to 89 %; silt = 2 to 78 %; OC = 0.03 to 34.9 %) that were grouped according to clay mineralogy and OC content. The total SA was measured using EGME adsorption (SA_E) and water adsorption and desorption (SA_H2O). The contribution of OC to SA was examined using regression and partial correlation analyses that combined clay, silt, and OC as explanatory variables. Results showed that SA_E values were strongly correlated to SA_H2O, except for OC-rich soils where SA_E was significantly lower than SA_H2O. Apart from montmorillonite-rich samples, OC contribution to SA_E was smaller than for SA_H2O. Organic carbon had a positive contribution to desorption SA_H2O for all sample groups, except for montmorillonite-rich samples; OC contribution was 7.5, 10.7, and 13.9 m²/g per %C for illite-rich, kaolinite-rich and OC-rich samples, respectively. There was no significant effect of water sorption direction on the OC contribution to SA_H2O. Partial correlation analyses that accounted for clay and silt contents confirmed a negative contribution of OC to SA_E or SA_H2O for soil samples dominated by montmorillonite clay minerals. We can conclude that for the soil types investigated here (except montmorillonitic soils), OC has a positive contribution to total SA, and the magnitude of the contribution depends on the clay type.
Estimation of soil water retention curves from soil bulk electrical conductivity and water content measurements
2021, Soil and Tillage Research
Measurement of soil water retention curves (SWRCs) is time consuming, and there is no single laboratory device available to measure a SWRC over an entire range of relevant pressures. The van Genuchten (vG) model is commonly used to characterize the shape of the SWRC. Bulk soil electrical conductivity as a function of water content, σ(θ), has been used to estimate hydraulic properties of unsaturated soils, thus making it possible to relate σ(θ) and SWRC. The saturated and residual water content values, θ_s and θ_r, can be estimated from soil bulk density and particle size distribution. In this study, we present an approach to estimate vG parameters m and α from σ measured at saturated and residual soil water contents, as well as σ(θ) values measured at intermediate water contents. A thermo-time domain reflectometry (thermo-TDR) sensor is used to measure σ and θ of the same soil sample volume. SWRCs for three soils (Glassil 530 sand, Tennessee silt loam and Illinois clay loam with bulk densities ranging from 1.52 to 1.67 g cm⁻³, 1.05 to 1.25 g cm⁻³ and 1.05 to 1.2 g cm⁻³, respectively) are estimated from σ(θ) measurements and compared with direct SWRC measurements obtained with a tension table and pressure plate extractors. Additional comparisons are made using data obtained from the literature. The proposed method to estimate SWRCs performs well when compared to direct SWRC measurements (with an average RMSE and an average bias of 0.041 cm³ cm⁻³ and 0.008 cm³ cm⁻³, respectively). Results indicate that the new σ(θ) based method accurately estimates SWRCs.
A conceptual model of carbon stabilisation based on patterns observed in different soils
2020, Soil Biology and Biochemistry
In principle, greenhouse gas emissions can be offset by increasing soil carbon stocks. Full utilisation of that potential, however, requires a good understanding of the controls on carbon stocks to identify factors that can be modified through management changes and distinguish those from factors that are inherent soil properties that cannot be modified. Here, we present a conceptual model of protected (or stabilised) carbon stocks in soils based on observations from two farms in New Zealand, and from a combined soils data set from observations from throughout New Zealand. These data showed that
- 1)
  When other factors, such as climate, plant cover and pasture management, were identical, soil carbon stocks were highly, and linearly correlated with the soil's specific surface area estimated from soil water adsorption.
- 2)
  The slopes of these relationships decreased with soil depth.
- 3)
  Extrapolation of the relationships to zero specific surface area resulted in relatively small intercepts on the soil carbon axis. These intercepts decreased with soil depth.
- 4)
  The intercepts were considered to correspond to unprotected labile carbon, with highest contents near the soils surface where most carbon inputs are received by soils.
- 5)
  Together, these observations implied that virtually all protected carbon in the analysed soils was protected by the soil matrix rather than biochemically, and that mineral surface area was the functionally relevant key attribute that defined the soils' protective capacity.
- 6)
  It implied that protected organic carbon,  $C_{p}$ , in a soil can be described as:  $C_{p} = k C_{i n} A_{m} / f (T, W, pH, A l, \dots)$ , where k is a simple constant,  $C_{i n}$  is the total carbon inflow rate into the soil,  $A_{m}$  is specific surface area, and f(T, W, pH, Al, …) is a specific turn-over rate of protected carbon as a function of temperature, soil water, pH, aluminium concentration, or any other factors apart from soil texture that may affect soil-carbon turn-over rates.
These observations improve our understanding of the important carbon-protection mechanisms in the soil, with significant implications for the optimal manipulation of carbon input rates into different soils to maximise overall soil carbon storage. They imply that overall carbon storage of soils could be enhanced by physically transferring any available carbon from soils with low to soils with high specific surface areas.

View all citing articles on Scopus

View full text

Short communicationEstimating residual water content in air-dried soil from organic carbon and clay content

Highlights

Abstract

Introduction

Section snippets

Materials and methods

Results and discussion

Acknowledgements

Ecol. Model.

Geoderma

Modern Methods for Robust Regression

The wilting coefficient for different plants: and its indirect determination

Govt. Print. Off

Short communication
Estimating residual water content in air-dried soil from organic carbon and clay content