Which Forecast to trust? Credibility and craftmanship in environmental modelling

Forecasts, here understood as quantitative predictions of the future state of a system, come in all shapes and sizes. The German "economic wise guys" forecast next year's economic growth, the evening news present the three-day weather forecast, climate scientists predict drought conditions in 2100, and ecologists forecast loss of diversity following land-use and climate change. The forecasts communicated to the public typically come in maps or simple line graphs or single numbers. Assumptions, data quality and forecast uncertainty are never communicated. Within each community, however, these are fundamental for the credibility of a modelling approach; the forecasts are often merely the spin on the actual science for a more widely disseminated publication. In some areas of environmental science, however, forecasts are all there is: a predicted time-trend, a map, emerging from a short analysis without thorough scientific evaluation of uncertainties, explicit statement of assumptions, discussion of potential errors or alternative scenarios. And, indeed, without any scientific merit. This kind of science is largely politically motivated, its scientific credibility is rather low. After giving some examples of the current range of environmental forecasts, I want to present a few lines along which to identify, as a person outside the actual field, the scientific quality and hence credibility of an environmental forecast.