We study the origin of the background seismic noise averaged over long time by cross correlating of the vertical component of motion, which were first normalized by 1-bit coding. We use 1 year of recording at several stations of networks located in North America, western Europe, and Tanzania. We measure normalized amplitudes of Rayleigh waves reconstructed from correlation for all available station to station paths within the networks for positive and negative correlation times to determine the seasonally averaged azimuthal distribution of normalized background energy flow (NBEF) through the networks. We perform the analysis for the two spectral bands corresponding to the primary (10--20 s) and secondary (5--10 s) microseism and also for the 20--40 s band. The direction of the NBEF for the strongest spectral peak between 5 and 10 s is found to be very stable in time with signal mostly coming from the coastline, confirming that the secondary microseism is generated by the nonlinear interaction of the ocean swell with the coast. At the same time, the NBEF in the band of the primary microseism (10--20 s) has a very clear seasonal variability very similar to the behavior of the long-period (20--40 s) noise. This suggests that contrary to the secondary microseism, the primary microseism is not produced by a direct effect of the swell incident on coastlines but rather by the same process that generates the longer-period noise. By simultaneously analyzing networks in California, eastern United States, Europe, and Tanzania we are able to identify main source regions of the 10--20 s noise. They are located in the northern Atlantic and in the northern Pacific during the winter and in the Indian Ocean and in southern Pacific during the summer. These distributions of sources share a great similarity with the map of average ocean wave height map obtained by TOPEX-Poseidon. This suggests that the seismic noise for periods larger than 10 s is clearly related to ocean wave activity in deep water. The mechanism of its generation is likely to be similar to the one proposed for larger periods, namely, infragravity ocean waves.