Evaluating Mediterranean field vulnerability for soil and goundwater agrochemical contamination

Comunicación oral presentada en la Session: Soil and Water, del citado Congreso, celebrado del 1-4 de junio 1992, en Upsala, Suecia.

Agricultural production within the Mediterranean regions has strongly increased during the last decades as a result oC, amongst other factors, grace use of agrochemical compounds: fertilizers and pesticides. This increase has become beneficial in many ways, but these is a clear evidence oC adverse effects oC modem agricultural practices 00 soil and groundwater quality. In the framework Di MicroLEIS: Microcomputer-based Mediterranean Land Evaluation Information System, an expert evaluation model (ARENAL program), is developed as knowledge-based system that allows lo predict! the relative vulnerability of the different fields to agrochemical compounds, in terms of soil and groundwater diffuse contamination. Soil factors (texture, salinity, pH and CEC) and related agricultural land features (precipitation, physiography and water conditions) are combined with management system criteria (farming system, artificial drainage and water extraction), through a microcomputer expert model. The expert knowledge was captured into the Automated Land Evaluation System (ALES), through computer-based decision trees. ALES system is a framework for evaluators lo build their own expert model in accordance with the FAO Land Evaluation Guidelines. Four steps can be distinguished when ALES is used for assessing the relative field vulnerability: (1) formulating field Imitations in relation lo the studied process on soil and groundwater agrochemical pollution; (2) sleeting relevant land and management characteristics, which have an important influence on the process being considered; (3) defaming land and management factors, "key" parameters, and deducing these from the relevant characteristics or diagnostic criteria using decision trees. Security levels of the criteria are determined for each field unit. (4) combining land and management factors using a decision tree lo infer relative vulnerability I according to four c1asses, for each field unit. Finally, ARENAL represents a compiled version of the Basic program elaborate from the expert model previously built. 466 ARENAL model was applied to selected benchmark agricultural areas from Andalucía region, comparing the effect of the management system changes lO the vulnerability classes. Vulnerability class oC each oC the selected areas was easily computed by application of the evaluation program The 36 per cent oC the total number of selected units was evaluated as oC None Vulnerability Class, basically due lo the very large storage capacity Coc chemical compounds oC (he representative soils. On the contrary, the 45 per cent was classified as of Moderate or Severe Vulnerability Classes. Similar applications oC this export evaluation model can help to develop new agricultural management systems which tack care oC the environmental quality. The computer program on diskette is provided as an integral part of MicroLEIS Version 3.1 package.

Peer reviewed