A Fast and Easy to use Method to Forecast the Risks of Loss of Ecosystem Stability: The Epsilon Section of Correlation Sums

The assessment of risk related to breakdown in ecosystem stability, health and continuous delivery of services is one of the central issues in Global Change research. Although well established measures exist based on statistics, insurance mathematics or probability theory - they all depend on records of past events that happened (a) on the same or a similar site, (b) under the same or similar conditions, or (c) are facing the same or similar threats.

While such data may exist for specific well known systems (e.g. growth and yield in forestry and agriculture, earthquake frequency/intensity, volcano activity, ...), no such measures exist for sites, conditions or threats, that are devoid of past records.

One question remains: How to forecast risk for such systems in terms of changes in: (i) climate, (ii) atmospheric composition, (iii) human impact or (iv) expected ecosystem services.

In this presentation we will introduce a non-linaer measure of risk, i.e. the Epsilon section of the Correlation Sum from time series of observations, measurements or ecosystem model outputs. Results indicate, that even short time series of observations, measurements or modelled scenarios may be sufficient to assess risk at a fine scale, especially where standard methods of risk assessment fail due to lack of a sample data base.

We will show examples of method application for different ecosystems and discuss the potential value of our approach for policy advice in situations where prior expert knowledge is missing.