Time Series Scenario Composition Framework in Supporting Environmental Simulation Tasks

Abstract

To answer the impacts under specific what-if scenarios together with simulation tools has been demanding in different environmental problems. In this contribution, a general software framework for time series scenario composition is proposed to deal with this issue. It is done through providing an interface to process available raw time series data and to compose scenarios of interest. These composed scenarios can be further converted to a set of time series data, e.g. boundary conditions, for simulation tasks in order to investigate the impacts. This software framework contains four modules: data pre-processing, event identification, process identification, and scenario composition. These modules mainly involve Time Series Knowledge Ming (TSKM), fuzzy logic and Multivariate Adaptive Regression Splines (MARS) to extract features from the raw time series data and then interconnect them. These extracted features together with other statistical information form the most basic elements, MetaEvents, for the semi-automatic scenario composition. Besides, a software prototype with two application examples containing measured hydrological and hydrodynamic data are used to demonstrate the benefit of the concept. The results present the capability of reproducing similar time series patterns from specific scenarios comparing to the original ones as well as the capability of generating new artificial time series data from composed scenarios based on the interest of users for simulation tasks. Overall, the framework provides an approach to fill the gap between raw data and simulation tools in engineering suitable manner.