Well, what do you know about weather observation systems --- things like density of observation stations, observations by satellite (will give cloud cover, though not detailed surface conditions), and how observations are used by weather-modeling programs? You have some reading to do, I think!
Then, you need to consider some general factors: constancy of weather at the two sites (most "weather" takes place over the temperate zone, and on out into the Atlantic), likelyhood for weather-changing events (fronts, storms) at the two sites, intended prediction-ahead time, and just what are the criteria for accuracy to be used (i.e., amount of precipitation, temperature daily min and max, cloud cover %, relative humidity -- what?).
If you seek a very long-term prediction, the historical average data dominate; but most people wouldn't refer to that as "weather", but climate.
-- Cheers, -- Mr. d.