If dealing with multiple data files of similar layout especialy in a text format such as CSV a data step is likely to best in the long run. One reason would be that the code is more reuseable when you do the next set of runs. Also proc import, depending on the values may generate data sets with one or more variables as character in from one source file and numeric from another. The data step allows you to specify.
There are methods that allow reading multiple data files using wildcards and then having the name of the input file available as a variable in the data set.
We could use more information to provide a more specific response. Such as do the model projections have column headers? Naming convention?
If your data doesn't have column headers you could have code that starts as simply as
filename source "c:\path\*.csv"; /* reads all csv extension files in the folder identified by path*/
data raw;
infile source <various options> filename=Source;
input maxtemp mintemp precip wind radiation;
length InputFileName $ 200;
InputFileName=source;
run;
I recommend label and format assignments.
If it seems that you have a separate file for each variable then we have other bits to straighten out (such as who writes there models that way) to align your data.
You do need to provide some more examples of what you actually have.
... View more