Dynamical Model for Indoor Radon Concentration Monitoring
Abstract
In this paper,we will deal with interesting example of natural risk modeling - namely of indoor radon concentration. We will illustrate, how the traditional view based on fixed (i.e., time-invariant) coefficient models can be misleading. Then we formulate a nonparametric regression dynamic model as a doable alternative. Situation is not entirely standard here, as the smoothed and weight-giving variables are different. We illustrate on data coming from a rare intensive measurement campaign in an occupied house with simultaneous measurements taken in different rooms. We also illustrate how even such a simple model can produce behavior that throws some light on radon experts' discussions about natural radon concentration circadial movement phase. This is because the model acts as a linear but time-varying filter that can change not only amplitude but also phase between different rooms of the same house, suggesting that there might not be a universal phase of (e.g., daily) radon variation.