Dynamic data driven simulation with soft data
Yuan Long, Xiaolin Hu · 2014
Dynamic driven simulation dynamically assimilates observation at runtime to improve the simulation results. Typically, the observations are data that are collected from sensors. In this paper we consider dynamic driven simulation with soft data, which are coming from human reports. Compared with the quantified hard data, soft are qualitative, fuzzy and subject to human judgment. This paper proposes a method to convert soft information to quantified based on fuzzy set theory, and then combines soft and hard to carry out assimilation. We apply this method to dynamic driven simulation of wildfire spread and show that the accuracy of simulation is significantly improved by assimilating both hard and soft data.