Contexte et enjeux

Soils are a non-renewable resource over a human time frame. Human activities represent a major threat for soils through soil sealing, erosion, salinisation, acidification, pollution, etc. This has been recently recognised by the European Commission in a communication on soil protection (European Union Commission, 2002). Agriculture may be considered as a forcing variable on soil evolution, which acts on a time scale ranging from 10 to 100 years and which is thought to be non-reversible in many cases (Sombroek, 1990; Rounsevell et al., 1999; Montagne et al., 2009). However, the impact of this forcing variable on soil evolution at this time scale has been rarely studied and is poorly known (Tugel et al., 2005; Richter, 2006). It is hence crucial to understand and to predict the impact of agriculture on soil evolution, in order to propose agricultural practices allowing the management of soil resources in a sustainable way. This implies:
- Gaining knowledge about the soil processes that are affected by human activities;
- Determining the extent to which they are affected;
- Determining the rate of these processes;
- Modelling these processes.
The proposed project lies within this framework.

State of the art

In a review on soil modelling, Samouelian and Cornu (2008) showed that some of these soil processes are already well taken into account in models existing in the literature: for example, geochemistry and solute transfer is described in coupled models: e.g. KIRMAT (Gérard et al., 1996), MIN3P (Mayer et al., 2002), HP1 (Jacques and Simunek, 2005), and LEACHC (Huston and Wagener, 1992). Chadwick and Chorover (2001) stated that “significant research has been conducted since the 1960s to elucidate the kinetics and mechanisms of dissolution and precipitation at the molecular scale.” Numerous models do exist to model the dynamics of organic matter. Among others, Century (Parton et al., 1993) or RothC (Coleman and Jenkinson, 1995) constitute the most famous examples. Salvador-Blanes et al. (2007) reported that existing models of pedogenesis failed describing the horizon formation, because they did not consider redistribution of matter with depth. Moreover, Samouelian and Cornu (2008) identified two processes which are poorly modelled: bioturbation and lessivage – the latter being understood as the substantial translocation of fine particles (with size ranging from less than 2 µm to less than 10 µm according to different authors) from an horizon, called eluviated horizon, to another horizon, referred to as illuviated and responsible for the formation of horizons with textural contrast. Bioturbation and lessivage are both responsible for the redistribution of matter with depth. In a recent study, Finke and Huston (2008) developed a model called SoilGen1. This model is based on the LEACHC (Huston and Wagener, 1992) and the RothC 26.3 models (Jenkinson and Coleman, 1994) to which they added a module of bioturbation as well as an additional module on biogeochemical recycling by plants. With respect to lessivage, the authors concluded that: « clay transport (…) cannot be simulated due to the lack of process knowledge ». However, lessivage is among the most widespread processes in soils and it has been described in many soil types. It is notably the main process driving the formation of the so-called Luvisol (WRB, 1998) having an E-horizon containing around 15 % of the fraction less than 2 m and the Bt-horizon containing almost 25 to 30% of the same fraction. Several authors hypothesise that agricultural activities could have a crucial impact on their evolution (Khan et al., 2005; Kühn, 2003; Fedoroff, 1997). As lessivage depends both on the physico-chemical conditions found in the soil and on the flux of water percolating in the soil, all agricultural activities influencing on at least one of these two aspects may have consequences on this process. In this context, we decided to characterise, quantify and model the impact of different agricultural practices on a widespread soil process which has been poorly characterised and modelled so far: lessivage.

What is lessivage?

It is a transfer of fine particles in soil

It is characterised by the presence of clay coatings and cutanes at the microscopic and macroscopic scale (Jamagne, 1973)

Questions autour du lessivage

Quelle est l’importance des tranferts verticaux de particules en terme de fomation des sols ?

Si le lessivage est un processus important de formation des sols, est il toujours actif actuellement sous nos climats (Fedoroff, 1997)?

Quel sont les impacts de certaines pratiques agricoles courantes comme la réduction du travail du sol et l’apport de déchets organique sur le lessivage?

Quel est la vitesse de ce processus?

Objectifs du projet

Characteriser les phases solides (taille nature réactivité) mobilisées par le lessivage

Quantifier l’importance du lessivage dans les sols en tant que processus d’évolution des sols

Quantifier sa cinétique

determiner les facteurs controllant le lessivage

Quantifier l’impact de l’homme et du climat sur le lessivage

Modeliser le lessivage.

Organisation du projet

Approche multi-isotope


L’objectif général de cette approche est de quantifier les processus de lessivage des fractions fines du sol et de pédoturbation, et leurs cinétiques, à l’échelle du solum et de caractériser et quantifier l’impact des activités agricoles sur ces processus pour des échelles de temps allant de la dizaine à la centaine d’années (voire au millier).


Plusieurs traceurs isotopiques naturels (210Pb, 10Be, 137Cs, 14C, 26Al), sont apportés aux sols par leurs surface et redistribués dans les sols par les procéssus bio-physiques. Ils permettent de suivre les flux de matière à différents pas de temps de par leur propriétés de demi-vie, l'historique de leur apport et/ou leur affinité préférentielle pour certaines phases solides. Ces isotopes sont pour l'instant utilisés séparément. Cette approche sera couplée à des analyses minéralogiques, géochimiques (isotopie du Si) et à la quantification des revêtements en microscopie.

Mesures sur ASTER (10Be, 26Al), à Poznan, Poland ou au NSF-Arizona-AMS-Lab , Tucson, USA (14C), au CEREGE par TIMS (206/207Pb) et LSCE à Gif-sur-Yvette ou au Laboratoire Souterrain de Modane (137Cs, 241Am et 210Pb)

Les Sites


Several factors or characteristics are known to impact particle transport (DeNovio et al., 2004; de Jonge et al., 2004; McCarthy and McKay, 2004) and, hence, probably lessivage in soils. The nature of the soil in terms of particle size distribution, mineralogy and geochemistry. Indeed, smectites are known to be the most mobile clay mineral; lessivage takes mainly place in soils within a pH range from 5.5 to 6.5 as minerals are stabilised by Al3+ or Ca2+ below or above this pH range respectively;
The characteristics of the percolating water (pH, ionic strength): in the case of lessivage, we can consider that the percolating water is rainwater with a pH of 5.5 and a low ionic strength.
The intensity of the percolating water is also an important parameter.
The structure of the soil.

We have decided:

i) to study the formation and the characteristics of the mobile fraction and its evolution with depth or with the different agricultural practices that modify the physico-chemical ambience of the soil solution

(ii) to investigate the respective contribution of exceptional climatic events compared to moderate rainfall events on lessivage

(iii) and to analyse the role of the soil structure on lessivage

i) The mobile fraction and its evolution

The impact of the physico-chemical properties of the soil water on the dispersion of the mobile particle fraction, on the interparticle organisation in the aggregates formed and on their transfer vs. deposition in the soil will be characterized in batch experiments.

ii) Contribution of exceptional climatic events to lessivage

Boulaine (1978) demonstrated that extreme events, such as storms, have a major importance on soil evolution. We propose to test this hypothesis for lessivage on soil monoliths in lab-experiments using the rainfall simulator available at INRA Orléans. Two types of rainfall events will be considered (i.e. storm events on a dry soil and a low-intensity rainfall events on a wet soil), the total quantity of water percolating the soil being kept constant in both cases, chemistry of the rainwater being as close as possible from real rainwater. Percolating solution (pH, EC, major anions and cations) will be monitored along the soil column during the experiment.
The experiments will allow locating and quantifying the fluxes of particles and water departing from and scavenged into the soil columns using different tracers (CEC, mineralogy and specific surface of the mobile fraction, particle size fraction measurements). To access the sensibility of these tracers, we propose to perform an analysis of sensibility mixing a small quantity of mobile fraction (operationally defined) of one material in that of the second material in order to determine the minimal quantity that can be identify.

iii) Role of the soil structure on lessivage

In lessivage, eluviation and illuviation obviously modify the porosity distribution of both eluviated and illuviated horizons, respectively. Interactions and feedbacks between soil functioning and soil structure are in that case of key importance for understanding and predicting soil evolution. However, these interactions and feedbacks are poorly considered nowadays and they are rarely modelled. We propose to characterise the structure of the soil monoliths used in the experiment and its evolution through time by tomography and microtomography.


Lessivage, was never modeled to our knowledge. However models of particle transfer exist in the literature (Jarvis et al., 1999; Simuneck et al., 2005; Majdalani et al., 2007). Such models are all coupled with a water transfer model that can either consider uniform flow, or more complex flow pathway considering non equilibrium water transfer based on physical factor like dual porosity (Mobile Immobile models MIN) or dual permeability model. The second possibility is mainly used when preferential flow is dominant. In most cases water and particle transfer models are coupled using one output of the water transfer model (usually the flux q or the water content theta) as an input in the particle transfer model. Concerning the particle transfer model, there is a large body of literature devoted to model particle fate in porous media, but models considering in situ particle mobilization (i.e. of particle originally present in the soil) by unsaturated flow are scarce. More generally particle mobilization mechanisms are still poorly understood. For example, to date, the relative importance of the different mechanisms governing the particle mobilization is still unknown and some factors identified as influencing particle mobilization (e.g. the chronology of irrigation, Majdalani et al., 2008) are not taken into account in any model. As a consequence, existing models are not generic, and most of their parameters cannot be determined from independent measurements but are estimated by inversion on the experimental particle breakthrough curves. Based on this analysis, in this project, we propose to use, as water transfer model, either uniform or non-equilibrium, depending on the importance of the preferential flow in the lab-experiments. Therefore, the input data necessary to parameterize these two types of model will be acquired along the experiment. As an example, Hydrus and MACRO necessitate the unsaturated soil hydraulic functions while for KDW, three parameters have to be estimated thanks to hygrograms obtained independently on the same soil core for various irrigations intensities. Whatever the model chosen, the obtained simulations will be compared to experimental hydrographs. As particle transfer model we propose to use our own model (Majdalani et al., 2007) originally proposed for contexts in which preferential pathways are dominant, but which hypotheses hold for uniform flow. In this model, two different mechanisms describe mobilization during transient and steady flow. Transient flow mobilization is proportional to water flux acceleration during the flow onset, while, when steady flow is reached, the detachment rate follows a first order kinetics and depends on the amount of particle already mobilized. The parameters of this model will be estimated through particle breakthrough curve inversion. There are two difficulties to consider when attempting to model lessivage with a colloid transport model: the evolution of the soil structure though time and the evolution of the stock of potentially mobile particles. Most of the models hypothesize that the stock of potentially mobile particles is of finite size. The size of this stock is usually estimated prior each rainfall event. In Majdalani et al. (2007), the size of the stock S(z) is estimated once, as a function of the soil core depth z. Initial values of S(z) for a rainfall event n is that reached at the end of the rainfall event n-1. This features reflect only partially reality: if the stock of particles is depleted as lessivage proceeds, the stock of particle may be regenerated during wetting/drying (Kaplan et al., 1993) or freeze/thaw cycles (Worrall et al., 1999), or when mobilized particle uncover a new stock of potentially mobile particles (Majdalani et al., 2007). However, there is no regeneration process of the stock of potentially mobile particle neither in Majdalani et al. (2007) nor to our knowledge in any other colloid transport model except an attempt in Jarvis et al. (1999). The implementation of such a regeneration mechanism may proved to be essential in order to describe mobilization occurring over tens to thousands years, i.e. lessivage. Soil structure modification may also need to be considered. Indeed as lessivage proceeds, over tens to thousands of years, the structure of both the elluviated and illuviated horizon are expected to change. These modifications are not taken into account by the current models of colloid transport that are used on shorter time scale (one to several rain events). Nevertheless, linking, in a model, lessivage, soil structure and hydraulic property modifications is a research topic by itself and may prove to be difficult. We thus propose an alternate way to implement this feedback loop based on data assimilation of hydric properties.

projet.txt · Dernière modification: 2011/03/15 10:31 par scornu