I'm comparing survey results of respondents' perceptions of creek environmental health to environmental health as measured by scientists with the city. What's the best way to analyze the difference?
The question is, what / how can your data be compared or related.
If you had measures on the same variables from respondents and scientists, you could compare them with tests like the t-test. Since yours are different types of variables (e.g., on different scales), it isn't meaningful to compare them. That is, you cannot "analyze the difference".
To relate variables, you need to be able to say that a given value in one variable(s) corresponds to a specific value in another variable(s). For the most part that isn't clear with your situation either. To the extent you can say that different respondents' perceptions correspond to particular streams, you could average those and get a single (mean) value of perception for that stream. Then you could use the scientifically measured variables for that stream and fit a multiple regression model.
Note that this is probably a large data loss, though. You may not have enough information left over for a multiple regression model. One possibility might be to standardize all the measured variables (and reverse some of them, if necessary, so that better is always in the same direction), run a principal components analysis, and use the first principal component as your measure of scientific stream health.
Your only other option, as far as I can see, is just to describe the data (for example, using means, SDs, etc.).