Forecasting the State of a Geographic Region Using Neural Nets and Fisher Information: The San Luis Basin, Colorado Study

A critical need within sustainable environmental management is the ability to assess system changes over time. Indicators provide a clear means of tracking conditions, yet available data is often years in the past and might not reflect current system status. Further, tools are needed to aid managers in real-time decision making. This work presents a methodology that combines the power of an Artificial Neural Network and Information Theory to capture patterns in data, and then uses it to forecast time series representing system conditions. The novelty and strength of this approach is in employing Fisher Information to help bound the forecast. This methodology was applied to social, environmental and economic variables describing the San Luis Basin system in Colorado, USA from 1949 to 2010. Results indicate that the approach provides a viable management tool useful for projecting possible outcomes with the aim of avoiding unsustainable patterns and promoting sustainable trends.