Microwave observations in the range 1 to 18 GHz with high spectral resolution (40 frequencies) have shown that many events display a complex microwave spectrum. From a set of 14 events with two or more spectral components, we find that two different classes of complex events can be distinguished. The first group (4 events) is characterized by a different temporal evolution of the spectral components, resulting in a change of the spectral shape. These events probably can be explained by gyrosynchrotron emission from two or more individual sources. The second class (10 events) has a constant spectral shape, so that the two spectral components vary together in intensity. For all ten events in this second class, the ratio of primary to secondary peak frequencies is remarkably similar, exhibiting an average value of 3.4, and both components show a common circular polarization. These properties suggest either a common source for the different spectral components or several sources which are closely coupled. An additional example of this class of burst was observed interferometrically to provide spatial resolution. This event suggests that the primary and secondary components have a similar location, but that the surface area of the secondary component is larger.

Microwave observations with exceptionally high spectral resolution are described for a set of 49 solar flares observed between May and October 1981. Total power data were obtained at 40 frequencies between 1 and 18 GHz by the Owens Valley frequency-agile interferometer with 10 s time resolution. Statistical analysis of this sample of microwave bursts established the following significant characteristics of their microwave spectra: (i) Most (≈ 80%) of the microwave events displayed complex spectra consisting of more than one component during some or all of their lifetime. Single spectral component bursts are rare. It is shown that the presence of more than one component can lead to significant errors when data with low spectral resolution are used to determine the low-side spectral index. (ii) The high-resolution data show that many bursts have a low-side spectral index that is larger than the maximum value of about 3 that might be expected from theory. Possible explanations include the effect of the underlying active region on the perceived burst spectrum and/or the necessity for more accurate calculations for bursts with low effective temperatures, (iii) the peak frequencies of the bursts are remarkably constant during their lifetimes. This is contrary to expectations based on simple models in which the source size and ambient field remain constant during the evolution of a burst.

First observational evidence of harmonic radiation at microwave frequencies during solar bursts is presented for the event of April 28, 1983. The recordings between 3.1 and 19.6 GHz show a typical continuum with a spectral maximum near 5.2 GHz. Superimposed fine structures with durations in the order of some seconds exhibit a very unusual spectral behaviour. Narrow-banded intensity peaks appeared at 5.2 and 11.8 GHz which were barely visible at 3.1, 8.4 and 19.6 GHz. These structures can be interpreted as harmonic emission. Harmonic radio emission can be generated either by plasma radiation, gyroradiation, electron-cyclotron maser or by nonlinear conversion processes. However, all of those mechanisms require extreme assumptions on the source and the ambient plasma in order to account for the observations.

Etudes des emissions microondes et decimetriques des eruptions solaires entre 3100 et 5205 MHz

The microwave radiation from solar flares sometimes shows short and intensive spikes which are superimposed on the burst continuum. In order to determine the upper frequency limit of their occurrence and the circular polarization, a statistical analysis has been performed on our digital microwave observations from 3.2 to 92.5 GHz. Additionally, fine structures have been investigated with a fast (5 ms) 32-channel spectrometer at 3.47 GHz. We found that ∼ 10% of the bursts show fine structures at 3.2 and 5.2 GHz, whereas none occurred above 8.4 GHz. Most of the observed spikes were very short (≤ 10 ms) and their bandwidth varied from below 0.5 MHz to more than 200 MHz. Simultaneous observations at two further frequencies showed no coincident spikes at the second and third harmonic. The observations can be explained by the theory of electron cyclotron masering if the observed bandwidths are determined by magnetic field inhomogeneities or if the rise times are independent of the source diameters. The latter would imply source sizes between 50 and 100 km.