Abstract <p>Production interruption caused by an unfavorable combination of weather factors have a significant impact on the integral performance of objects and systems operating outdoors, which should be properly taken into consideration in their simulation models (“digital twins”). The paper proposes a method for stochastic modeling of a random process of “weather windows” dynamics in continuous time, i.e., with an arbitrary time resolution. Discrete records of regular meteorological observations are used to identify the parameters of this continuous stochastic generator. The proposed technique makes it possible to obtain an accurate model description of the random process from indirect and incomplete measurements, provided that certain assumptions about its nature are met. Examples and prospects of using the proposed approach in simulation models of marine transport systems are considered.</p>