In dynamic, complex and nonlinear systems, the analysis of transients, instabilities or unusual events is fundamental, as well as the knowledge of the system's behavior in the frequency space. What is usual in this type of study is to have discrete and digitally acquired data, due to a sampling process with a determined period or frequency. Extracting information from this data set is done by numerical methods, such as the Fourier Transform (FFT), the Hilbert-Huang Transform (HHT) or techniques from chaos mathematics (Attractors). In this way we have scarce data, with some loss of information, analyzed with numerical techniques that introduce, also, some dilution of the information. In particular, the techniques of analysis in the frequency space, by their very nature, blur, frequencies and bandwidths. This leads in some cases to the identification of unsuitable or very wide frequencies or intervals, which make the extracted information unusable. To this effect, we have studied the VMD technique that provides more real and representative frequencies of the system as well as narrower bandwidths that make possible better identification of phenomena and influences. This technique has been applied to the analysis of thermohydraulic instabilities in nuclear boiling water reactors (BWR). The cases had already been analyzed with STFT and HHT techniques. The results of the application of the VMD shows that the mode mixing is solved, so modes obtained are monofrequency and these frequency bands are narrower than in previous methodologies. This improved performance allows using backbone plots for identifying which modes are excited in instabilities events.