[GRASS-dev] [GRASS GIS] #2014: r.sun using EPSG:3031 projection gives strange results

[GRASS GIS]

#2014: r.sun using EPSG:3031 projection gives strange results
---------------------------+------------------------------------------------
 Reporter:  pierreroudier  |       Owner:  grass-dev@…              
     Type:  defect         |      Status:  new                      
 Priority:  normal         |   Milestone:  7.0.0                    
Component:  Raster         |     Version:  svn-trunk                
 Keywords:  r.sun          |    Platform:  Linux                    
      Cpu:  x86-64         |  
---------------------------+------------------------------------------------

Comment(by hamish):

 Hi, I had a try using Mt. Erebus as the shadowing body & r.sun in mode 2
 (integrated over the entire day),

 {{{
 north:      -1280000
 south:      -1372500
 west:       240000
 east:       339500
 nsres:      100
 ewres:      100
 }}}

 and cubic-reprojected etopo1 for elevation. (can probably dig out some
 lidar for Ross Island from someone in our dept; 100m DEM from Colorado is
 currently offline due to flooding)

 test setup for grass 6.5 looked like this:

 {{{
 DAYS="001 030 045 060 070 080"
 for DAY in $DAYS ; do
    r.sun -s elevin=etopo1.ross_island step=0.05 \
       beam_rad=erebus.direct.$DAY diff=erebus.diffuse.$DAY day=$DAY &
 done
 wait

 DAYS="085 090 095 100 105 110 115"
 for DAY in $DAYS ; do
    r.sun -s elevin=etopo1.ross_island step=0.05 \
       beam_rad=erebus.direct.$DAY diff=erebus.diffuse.$DAY day=$DAY &
 done

 for map in `g.mlist pat=erebus.d*` ; do
    r.colors $map col=bcyr
 done
 }}}

 day 115 is near last-light of the autumn, not exactly the same sunrise/set
 as these values for McMurdo, but pretty close & I'm not sure which
 definition they're using.

 http://www.timeanddate.com/worldclock/astronomy.html?n=1032&month=4&year=2013&obj=sun&afl=-11&day=1


 There's an obvious low-sun-angle bug in the diffuse output map, which I'll
 file in a separate report, and obviously the default albedo and linke
 values in the module won't be much good, but the direct beam results seem
 ok if I look at the individual days. I'm still waiting for days 1 & 30 to
 finish, but looking at the result for day 45 makes it seem like for
 summertime there might be some counter-intuitive irradiation pattern, with
 more light on the true-south side of the mountain. If day 1 or 30 show the
 same, the next thing would be to run r.sun in mode 1 quarter-hourly over
 that one day and see if that can make more sense.

 todo: test with grass7. The main differences in the versions is that by
 default shadowing effect of the DEM is turned on in G7, while in grass 6.x
 you needed to use the -s flag; and the lat/lon reprojection code is backed
 out of the main loop for speed and to make parallelization/GPU assist
 easier.


 Hamish

-- 
Ticket URL: <https://trac.osgeo.org/grass/ticket/2014#comment:5>
GRASS GIS <http://grass.osgeo.org>