[GRASSLIST:2246] Re: rectification and display questions

[Glynn Clements]

Kirk R. Wythers wrote:

> I am rectifying some raster images and have got a few questions for the 
> list.
> 
> 1. After setting points with i.points and running i.rectiy, I get the 
> expected message that mail will be sent when job is completed. However 
> I don't seem to get any mail (I assume it would be sent to the local 
> machine's user account that is logged in and running grass). Is there a 
> way to check or even change where mail is sent?

The system's logfiles may provide an indication of whether the mail
was sent. However, you can't change the destination; it's sent to the
current user, as determined by:

	getpwuid(getuid())->pw_name

[Given the number of recent problems/queries regarding this "feature",
and the fact that we are now trying to support Cygwin and MacOSX
(which may not support this without significant configuration effort),
I suggest that we consider getting rid of it.]

> 2. After quitting grass and restarting in the new location and mapset 
> (where i.rectify was projecting to), I can see the new files of the 
> geocoded tiffs:
> 
> g.list type=rast
> ----------------------------------------------
> raster files available in mapset ts2325:
> t59n.r10w.1.blue_r	t59n.r10w.1.green_r	t59n.r10w.1.red_r
> 
> I just used a _r for the extension that i.rectify asked for...
> 
> However, upon attempting to display in a monitor with r.rgb

You mean d.rgb, right?

> (after 
> setting region with g.region), all I get is a nice black image. Is 
> there something that I need to do to the color tables?

Quite possibly.

If the program which generated the maps didn't set the colour tables
correctly, you need to use "r.colors ... color=rules" to set a
grey-scale colour table which is appropriate for the data (typically,
black is 0 and white is the "100% intensity" value, e.g. 255 for 8-bit
integer data).

NB: you typically only need to use "color=rules" for the first map;
for the other two, you can use "rast=..." to copy the colour table
from the first map.

I can think of two likely reasons why the maps might be completely
black:

1. Something coerced floating-point data in the range 0.0 to 1.0 to
integers, resulting in all-zero output.

2. The data contains some extreme values (typically meant to represent
NULL but not interpreted as such), resulting in the colour table being
stretched over an extreme range, so that the "real" data only spans a
tiny proportion of the range.

The first step is to analyse the data with e.g. r.info (to check the
range), d.histogram (to check the distribution), d.legend (to
visualise the colour table) and/or d.rast (to view individual
channels).

-- 
Glynn Clements <glynn.clements at virgin.net>