In a recent study presented in the article First global observations of the mesospheric potassium layer (Dawkins et al. 2014), space-borne resonance fluorescence observations confirmed previous LIDAR observations of a semi-annual variability of mesospheric meteoric $\ce{K}$, which is very different from the annual (seasonal) pattern observed for other metals such as $\ce{Na}$ and $\ce{Fe}$.
What causes the semi-annual mesospheric meteoric potassium variability?
Recent observations and modelling by Plane et al. (2015) and Plane et al. (2014) have described the most likely mechanisms for why potassium has its unique semi-annual variability, with a small summertime maximum and minima occurring at the equinoxes (Plane et al. 2015). The semi-annual variability is consistent across the globe (Plane et al. 2014).
In both papers, the cause of the unique semi-annual variability are narrowed down to 2 main atmospheric chemical dynamics, based in no small part on unique K chemical properties:
(Note: MLT = Mesosphere-Lower Thermosphere)
In summertime, the maxima observed is due to the cooler MLT temperatures being low enough to neutralise K+ ions to K.
During the equinoxes, O is at a minima, while CO2 and H2 exhibit a maxima - so, according to Plane et al. (2015)'s model, decreasing O, while increasing CO2 and H2 causes a reaction with K to form KHCO3, thus resulting in K being at a minima.
According to their models, KHCO3 returns to K by largely temperature independent photolysis.
Plane et al. (2014) describe there is a slight decrease in K in high latitudes in the presence of polar mesospheric clouds.
References
Plane et al. 2015, Resolving the strange behavior of extraterrestrial potassium in the upper atmosphere, Geophysical Research Letters
Plane et al. 2014, The Mesosphere and Metals: Chemistry and Changes, Chemical Reviews