Data-driven dynamic emulation modelling for the optimal management of environmental systems

The optimal management of large environmental systems is often limited by the high computational burden associated to the process-based models commonly adopted to describe such systems. In this paper we propose a novel data-driven Dynamic Emulation Modelling approach for the construction of small, computationally efficient models that accurately emulate the main dynamics of the original process-based model, but with less computational requirements. The approach combines the many advantages of data-based modelling in representing complex, non-linear relationships, but preserves the state-space representation, which is both particularly effective in several applications (e.g. optimal management and data assimilation) and facilitates the ex-post physical interpretation of the emulator structure, thus enhancing the credibility of the model to stakeholders and decision-makers. The core mechanism is a novel variable selection procedure that is recursively applied to a data-set of input, state and output variables generated via simulation of the process-based model. The approach is demonstrated on a real-world case study concerning the optimal operation of a selective withdrawal reservoir (Tono Dam, Japan) suffering from downstream water quality problems. The emulator is identified on a data-set generated with a 1D coupled hydrodynamic-ecological model and subsequently used to design the optimal operating policy for the dam. Preliminary results show that the proposed approach significantly simplifies the learning of good operating policies and can highlight interesting properties of the system to be controlled. (C) 2011 Elsevier Ltd. All rights reserved.