Questions complexes posées par l'analyse coût-efficacité pour l'évaluation des mesures destinées à améliorer la qualité de l'eau | Challenging issues in Cost-Effectiveness analysis when assessing measures for water quality improvement

Cost-effectiveness analysis is particularly useful in comparing mitigation measures by providing a rationale for decisions to be taken at the level of agricultural areas for drinking water abstraction. Effectiveness of measures can be assessed for medium-term goals (pressure, practices intensity or risk of transfer) by the use of indicators. However, models are required for assessing final goals (impacts). In addition, questions of spatial and temporal heterogeneity make devising environmental mitigation programmes a complicated exercise. This is because the characteristics of agricultural production (soil types, slopes, farming systems, proximity to streams...) can vary hugely across a river basin as well as stochastic climatic events within the implementation period of measures. Furthermore, the precise extent of the damage caused to the environment by the use of pesticides is difficult to assess (delay between applications and the appearance of any quantifiable effects, origins of breakdown products). We used the SWAT model for predicting the long-term impacts of mitigation measures on water quality. Farming systems heterogeneity is addressed by using bio-economic modelling to appraise marginal and total costs of implementing measures. Traditionally the difficulties caused by data availability bring to model “representative” (average) farms and “type” (modal) farms. By using these, geographical information on farm plots and management practices is lost. We overcome these problems by calculating costs at the hydrological response unit level (HRU) which is the spatial unit used by SWAT to calculate flows of water and nutrients. Costs and Effectiveness are best calculated at common appropriate spatial and temporal scales. The sub basin and watershed level appeared to be the more relevant spatial scale while the hydrological simulation period is the common temporal scale used for quantifying total costs. Ratios of discounted sum of annual costs (¤) divided by the concentration reduction (g.l-1) summarises results into single useful quantitative indicators for selecting measures. They can be calculated at different scales within a river basin. However, these ratios need to be handled carefully, because of the deterministic approach. Our modelling framework for cost effectiveness analysis of measures has been applied to different case studies (drinking water abstraction areas) within the Garonne river basin in terms of their applicability, transparency and capability for decision making for reducing water pollution by pesticides. Our results show that it is possible to classify scenarios based on their cost-effectiveness, which can be represented graphically (maps with spatially distributed cost-effectiveness ratios or scatterplots) allowing for rational discussion between stakeholders.