Average recon satellite takes roughly 7 days to overly the same spot, due to its orbit and rotation of earth. But due to off nadir (sideways) looking capability, revisit time (of sensor looking at a target) is usually advertised as less, as 2-3 days. (of course, the more sideways the sensor is - the greater the distance, so resolution suffers)
Assuming thus 2.5 revisit time and perfect, unison orbits for a large constellation of sats (neither may be applicable for Chinese satellite fleet right not) one might have complete coverage of earth with every point on earth being revisited every 10 minutes with a notional fleet of 360 satellites.
If 10 minutes is overkill, and one revisit each 60 minutes is enough, then 60 satellites could do the job.
But. Night time and cloud cover are still an issue. Great majority of today's high resolution satellites are optical, as precise heat finding from such distances, through atmosphere, is not really doable. Night time can take roughly 50% of the entire time during a day, depending on the time of the year and latitude.
So optical tracking during low light or no light is hard or impossible to do for optical satellites.
Furthermore, cloud cover can cause same issues. Assuming 30-50% of cloud cover at any given time and at any given area in the west pacific (that's based on a very quick meteo chart glance, it may be off) and compounded for night time, it may be that optical satellites are really trying to track targets during just 6 or so hours during a 24 hour period.
Of course this all assumes sufficient resolution for tracking. Which is actually not that demanding. While meter resolution might be needed to find and identify a target, once that is done, continued tracking of the same target would likely be achievable with smaller, cheaper satellites of lesser resolution. Within a reasonable window, of course. If large cloud cover or night time comes, tracking would cease and another round of initial finding and identification would have to be performed.
Radar based recon satellites are getting cheaper and smaller though. Just google what ICEEYE company is doing. Radar sats are getting cheap enough that constellations of dozens are planned by a commercial company, with resolution that might even be enough for identification is some situations, and one that is definitely enough to keep a track on a contact after something else has identified it.
So, one satellite identifies a contact, and then radar satellites track it during night time and adverse weather.
While I don't think we're quite there yet that satellites could monitor entire traffic in the pacific, i do believe progress is getting so quick that China might have that capability (with ever growing fleets of small form satellites, cheaper radar sats and geostationary optical sats) within the next 10 years.
