See page 48 of Being Logical:
The simplest argument is one composed of two statements, a supporting statement or premise and a supported statement or conclusion. Usually, the context of the argument will allow you to tell which is which, but we attach what are called "logical indicators" to statements in order to mark them clearly as either premises or conclusions. [...] Common logical indicators for conclusions
Thus, the exposition of Conditional Argument (page 63) is a little bit sloppy: "if the weather is nice tomorrow, we will go on a picnic" is not an argument but a compound statement.
The complete argument will be:
"if the weather is nice tomorrow, we will go on a picnic; the weather is nice tomorrow. Therefore, we will go on a picnic."
It is an instance of the following "argument schema":
"if A, then B; A. Therefore, B".
Regarding validity, see page 60: validty regards "form" (the structure of the argument) and not "matter" (its content):
An argument is valid if its structure is sound, which means that its structure is such that true premises will ensure a true conclusion.
Thus, in order to assess validity of an argument, we have to consider the corresponding "argument schema", i.e. the formula obtained replacing statements with (sentential) variables:
An argument is valid when every "instance" (obtained substituting real sentences for the variables) of it has the nice property that:
if all the premises are true, so is the conclusion.
Please, note the clause "if all the premises are true"; in the above case, we have to assume that both premises are true.
But if we assume that the first premise: "if the weather is nice tomorrow, then we will go on a picnic" is true, then we cannot say "tomorrow there will not be good weather, but we can still decide to go on a picnic", because this is not the original premise.
The "argument schema":
"if P, then Q, and P; therefore Q",
is valid, as well as: "if P, then Q, and not-Q; therefore not-P".
The "argument schema": "if P, then Q, and Q; therefore P", instead, is not valid (see Affirming the consequent).
The validity of: "if P, then Q, and P; therefore Q" as well as that of: "if P, then Q, and not-Q; therefore not-P" are licensed by the truth table for the conditional propositional connective: "if..., then...".
For the first one, we assume that the first premise is true: thus rules out the 2nd line of the truth table (P true and Q false).
But we assume also P true, and this in turn rules out 3rd and 4th lines.
In conclusion, the assumption of both premises gives us only one possible state of affairs: P true and Q true: thus, we are licensed to safely conclude, from the two assumptions above, with the truth of Q.
Same approach to the second argument schema: the assumption of the first premise rules out the 2nd line of the truth table (P true and Q false).
And the assumption of not-Q as true (i.e. of Q as false) in turn rules out 1st and 3rd lines.
Again, we are left with only one line: the 4th one, where P is false, i.e. not-P is true.
Thus, we are licensed to safely conclude, from the two assumptions, with the truth of not-P.