the vast majority of those polls are before he was even in office over a span of weeks. since being in it's just been up and up for him. From 22 to 29 to 47.
Secondly take a look at the sample column. You see how most say "A" and some say "RV" and then the one at the top says "LV? those are 3 distinct samples, with "A" being polls of all americans, "RV" being registered voters, and "LV" being likely voters. Polls of all americans skew very liberal and polls of registered voters skew a bit liberal. (with regards to painting a picture of the electoral landscape) the reason for this is because there are a lot of poor liberals that can't get time off from work, live in urbanized areas with fewer polling places, etc. Polls of Americans in particular matter almost dick (electorally speaking) given half of America doesn't even vote. In any case, this is part of the reason why we have such a large spread among the sampled polls at realclearpolitics. THe methodology is different and they all tell different stories. Some polls are talking about American opinion (that wont amount to much), and other polls are talking about the electoral landscape. These are very different things.