The flows file contains monthly mean streamflows for 
over 1200 sites. Flow series have mostly been 
checked by eye for major discontinuities and 
outerliers; however, we have been "gentle" in
our decisions to edit or remove data. We were more
inclined to remove a site from the data set than
to edit its data. With the exception of European
Russian sites, we expect that most duplicate sites
have been merged or removed.

Each calendar year between the first year with streamflow 
record at a site and the last year with streamflow record 
comprises two (2) lines in the flow file. Missing data is 
indicated by -999. Years with no data at all (but within
the period spanned by a site's first and last years) will
appear in the record with 12 months of missing values.

Columns 1 - 20, first line, contain a site name that 
corresponds to the first columns of the header file.

Columns 23 - 25, first line, contain the number of years 
worth of records (some may be missing) in the file for this
site. Thus if you read  twice this number of lines in the
file, you will be at the first year of the next site's 
record.

Columns 27 - 30, first line, indicate the calendar year of 
the monthly flows (in m**3/s) that follow on this line and 
the next line.

Columns 32 - 37, first line, indicate the decimal latitude 
of the site; negative for southern hemisphere.

Columns 38 - 44 indicate the decimal longitude of the site;
negative for western hemisphere.

Monthly streamflows (m**3/s) follow in FORTRAN format:

 (t45,6(1x,e12.5)/,t45,6(1x,e12.5))

or as

 6 sets of 13-column flow rates for January - June flows
 starting in column 45, followed by 6 more 13-column
 flow rates for July - December on the following line,
 also starting in column 45.