The unipolar pulsed-plasma electrolytic oxidation (PEO) of aluminum has been replaced by bipolar pulsed methods that use a so-called 'soft-sparking' mode. This method results in an effective reduction of intense microdischarges, which are detrimental to the oxide layer. In a previous publication, we developed an in-situ multivariable microdischarge control scheme using unipolar pulsing. Using this method, it is possible to restrict the mean microdischarge size to well-defined limits, while at the same time influencing the mean microdischarge energy, number density or spectral emission behaviour. This method operates well inside a frequency range of f = 1-20 kHz. Although this method shows highly desirable plasma control properties, the mechanisms defining this frequency-dependent controllability are unclear. The aim of this study is to visualize the spatio-temporal behavior of microdischarges in higher frequency ranges. First, a wavelet transform was performed to estimate the temporal evolution of microdischarge lifetimes. Ceramic coatings were then deposited on aluminum alloy substrates in an aqueous solution using unipolar pulsed galvanostatic PEO. The aluminum samples were coated for 30 min at frequencies of f_1 = 50 Hz, f_2 =5kHz and f_3 = 100 kHz. High-speed imaging was carried out utilizing four synchronized intensified charge-coupled device (ICCD) cameras, each with a 500 ns exposure time. At f_2 = 5 kHz, the microdischarges were still able to follow the electrical pulses. In this regime, the process can be divided into two stages, an initial charging of the substrate surface without plasma emission and a subsequent slower evolution of microdischarges. Equivalent circuit model descriptions are given for both processes. At f_3 = 100 kHz, microdischarges were not able to follow the pulse frequency, as the lifetimes and risetimes of the microdischarge characteristics were longer than the pulse length. Reignition at the same spatial location, clustering and permanent ignition through pulse periods were observed.