- journal_title：Vadose Zone Journal
- Contributor：Jos J. T. I. Boesten
- Publisher：Soil Science Society of America
- Date：2007-
- Format：text/html
- Language：en
- Identifier：10.2136/vzj2007.0067
- journal_abbrev：Vadose Zone Journal
- issn：1539-1663
- volume：6
- issue：4
- firstpage：793
- section：ORIGINAL RESEARCH
Zero-tension lysimeters play an important role in groundwater risk assessments for pesticides in the European Union. In these assessments, measured lysimeter leachate concentrations are usually used directly for decision making. When doing so, one assumes (i) that the lysimeter bottom boundary condition itself did not lead to underestimating field leaching concentrations, (ii) that the number of application years and the duration of the lysimeter study were adequate to measure the maximum concentration in time, and (iii) that the pesticide-lysimeter system considered was sufficiently vulnerable with respect to leaching. These assumptions were tested using simulations with a Darcian water flow model combined with a chromatographic pesticide leaching model. The scenario consisted of a layered light-textured soil cropped with cereals and of multiyear weather data. The groundwater level in the field usually fluctuated between 0.7 and 2.5 m depth. The lysimeter bottom boundary condition resulted in pesticide leaching concentrations lower than those calculated for the field system. Simulations showed that a lysimeter study of 2 yr was too short to measure the maximum leaching concentration for pesticides with organic-matter/water distribution coefficient values exceeding 40 L kg−1. The probability that a lysimeter study results in a leaching concentration below 0.1 μg L−1 by a coincidental favorable combination of pesticide–soil parameters was assessed by Monte Carlo simulations. This probability exceeded 20% for pesticides that would give leaching concentrations of 1 μg L−1 under field conditions. Therefore, it is advisable that for each lysimeter study, modeling be used to assess the likelihood that the leaching concentration would be below 0.1 μg L−1, considering all relevant systematic and random factors.