I read that there is a direct relation between variation of pressure and seismic noise: high pressure correspond to an increase of noise, low pressure correspond to a decrease of noise.
It depends on what you define as seismic noise and which frequency range you refer to. If the range of interest is between 0.5 - 300 seconds, the most comprehensive work was published by Ardhuin et al., 2015 (see enclosed paper). Other frequency ranges have different mechanisms that drive them, such as suggested in the above answers.
Firstly, let me thank you for suggested pdf. Secondly, frequency can't be in unit of second but in hz. Period in second. Period of microseisms in the period range 3 to 300 s.