Watershed area calculation

[Thomas Adams]

I'm getting discrepencies (which are expected) between various methods
of determining/calculating watershed areas. These are my 'methods':

(1) look them up based on USGS streamgage locations;
(2) use the GRASS 4.1 program v.area, applied to vector boundaries
determined elsewhere.
(3) approximate the area calc. using the count from r.stats, following
use of v.to.rast to get a raster mask area for each basin & multiply by
the *approximate* pixel area (3 arc-sec grid, ~90x90 m^2 between lats 38
to 40 deg. N)

My 'errors' are ranging 5-10%; I do need to be consistent, and
(obviously) would prefer what's best & what makes the most sense.

The vector boundaries in Method-2, above, appear reasonable when
compared to the underlying (3 arc-sec) terrain; so, my guess is that
this will provide the most consistent & reproducible areas.

All comments are appreciated...

Tom

-- 
Tom Adams                              adams at ohrfc.noaa.gov
                                       http://www.nws.noaa.gov/er/iln
Development & Operations Hydrologist   NWS-Ohio River Forecast Center
1901 S. State Route 134                Wilmington, OH 45177
(937) 383-0527 (VOICE)                 (937) 383-0033 (FAX)