[GRASS-user] area stats on a lon/lat projection

[Glynn Clements]

Maning Sambale wrote:

> I know it should not be really practiced, but how accurate is the areal
> calculations (hectares) in v.report when the projection is lon/la wgs84?
> I have a vector data that covers several utm grids that I want to
> compute "hectarage".

AFAIK, the calculation is "exact", in the sense that it computes the
integral with respect to latitude.

The real issue is one of definition. Specifically, the polygon is
deemed to be bounded by straight lines (in lat-lon) between the
vertices, not geodesic lines (or straight lines in UTM). If your
polygon edges are large, this will make a difference.

If you want to know the precise calculation, see
G_ellipsoid_polygon_area() in lib/gis/area_poly1.c.

Also, bear in mind that projecting vector data (e.g. from UTM to
lat-lon) merely projects the vertices, so you are effectively changing
the shape of the area. In both cases (before and after projection),
the polygon is bounded by straight lines between the vertices
(straight lines in UTM before, straight lines in lat-lon after),
whereas applying a non-affine projection to a straight line will not
generally produce a straight line. I think that v.split may be of use
here (if I understand it correctly).

-- 
Glynn Clements <glynn at gclements.plus.com>