Biotechnologie, Agronomie, Société et Environnement (Jan 2011)
Valorisation de la Carte Numérique des Sols de Wallonie et d'une base de données disponible en analyse de sols, dans le cadre de l'évaluation du risque de pollution des eaux souterraines par les pesticides
Abstract
Making use of the Digital Soil Map of Wallonia and of an available soil analysis database for assessing the risk of pesticide leaching to groundwater. This study shows how the Digital Soil Map of Wallonia and the associated soil database Aardewerk can be combined with the MetaPEARL model to assess the sensitivity of agricultural soils to pesticide transfer into underground water tables in Wallonia (Southern Belgium). MetaPEARL is based on an analytical expression which describes the concentration of leached pesticides at the soil profile bottom depending on available data on soil characteristics, climate and pesticide properties. The results show an important soil sensitivity to pesticide transfer with a coefficient of retention on organic matter (Kom) very weak (about 10 dm3.kg-1) or with a relatively high half-life time (DT50 about 60 days). In this case, the pesticide is weakly retained by soil organic matter or slowly degraded and then stays available in the soil and can be quickly leached during rainfalls. In other respects, the pesticide sensitivity to leaching is strongly correlated with soil texture and its organic matter content. It is also observed that the concentration of leached pesticide is strongly depending of the rainfall surplus (water flux into the soil or hydrous balance). A sensitivity analysis has shown that the model is very sensitive to soil thickness, to organic matter content, to the bulk density of the mineral fraction by textural classes and of the organic matter. Therefore, these pedological variables have to be assessed with the highest precision to avoid adding further uncertainty to the predictions obtained. On the basis of the most sensitive soil variables, the analysis of "spatial" uncertainty related to the results delivered by MetaPEARL, due to the consideration of an unique representative value by soil type and by region (deterministic approach), shows that this approach tends to under-estimate the concentration of pesticide leached in the soil, compared to the use of a stochastic simulation (probabilistic approach) which takes into account the soil characteristic variability within a given soil type. This stochastic approach allows also the calculation of a more relevant threshold of risk probability (percentile) which can be compared to an imposed standard in view to enhance a sustainable management of natural resources.
