1998A&A...336..371I

A stochastic model for type III bursts is introduced, discussed, and compared to observations. The active region is assumed to be inhomogeneous, with a large number of emerging magnetic fibers. At their bases, random energy release events take place, in the course of which electrons are accelerated, travel along the fibers and eventually undergo the bump-on-tail instability. In the non-linear regime, the formed Langmuir waves induce strong turbulence in the ambient plasma, with secondary electrostatic waves appearing. Wave-wave scattering finally leads to the emission of transverse electro-magnetic waves at the fundamental and the harmonic of the local plasma-frequency. The superposition of the emissions from all the fibers yields a model spectrogram for type III bursts (flux as a function of frequency and time). Peak-flux distributions of the model are compared to the ones of five observations of type III bursts. It turns out that, in a statistical sense, the model is largely compatible with the observations: the majority of the observations can be considered generated by a process which corresponds with the presented model. The details of the different sub-processes constituting the model play no decisive role concerning the statistical properties of the generated spectrograms, to describe them approximately by randomizing the unknown elements is sufficient. Therewith, the correspondence of the model with the data is not unique. Likewise, intrinsic shortness of observed type III events does not allow a strict enough discrimination between different possible sub-processes of the model through statistical tests. With that, the conclusion is that the observations are compatible with a model which assumes (i) a randomly structured active region, (ii) a flare-particle acceleration-process which is fragmented into a large number of sub-processes, (iii) a distribution of the accelerated particles which is a random fraction of the ambient density and of power-law form with random index, and (iv) the fragmentary acceleration events to occur randomly in time, i.e. the temporal structure of type III events to be random, without any correlations between the individual bursts.